144
Incorporating Forestry into Stormwater Management Programs: State of the Science and Business Model Evaluation for Nutrient Reduction and Volume Control PROJECT NO. 4837

Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs:

State of the Science and Business Model Evaluation for Nutrient Reduction and Volume Control

PROJECT NO.4837

Page 2: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs:

State of the Science and Business Model Evaluation for Nutrient Reduction and

Volume Control

Prepared by:

Trisha Moore Kansas State University, Department of Biological and Agricultural Engineering

Charles Barden Kansas State University, Department of Horticulture and Natural Resources

Pabodha Galgamuwa Kansas State University, Department of Horticulture and Natural Resources

Alireza Nooraei Kansas State University, Department of Biological and Agricultural Engineering

Co-sponsored by:

U.S. Endowment for Forestry and Communities

2020

Page 3: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

ii The Water Research Foundation

The Water Research Foundation (WRF) is a nonprofit (501c3) organization which provides a unified source for One Water research and a strong presence in relationships with partner organizations, government and regulatory agencies, and Congress. The foundation conducts research in all areas of drinking water, wastewater, stormwater, and water reuse. The Water Research Foundation’s research portfolio is valued at over $700 million.

WRF plays an important role in the translation and dissemination of applied research, technology demonstration, and education, through creation of research-based educational tools and technology exchange opportunities. WRF serves as a leader and model for collaboration across the water industry and its materials are used to inform policymakers and the public on the science, economic value, and environmental benefits of using and recovering resources found in water, as well as the feasibility of implementing new technologies.

For more information, contact: The Water Research Foundation

1199 North Fairfax Street, Suite 900 Alexandria, VA 22314-1445 P 571.384.2100

6666 West Quincy Avenue Denver, Colorado 80235-3098 P 303.347.6100

www.waterrf.org [email protected]

©Copyright 2020 by The Water Research Foundation. All rights reserved. Permission to copy must be obtained from The Water Research Foundation. WRF ISBN: 978-1-60573-457-6 WRF Project Number: SIWM12C15/4837

This report was prepared by the organization(s) named below as an account of work sponsored by The Water Research Foundation. Neither The Water Research Foundation, members of The Water Research Foundation, the organization(s) named below, nor any person acting on their behalf: (a) makes any warranty, express or implied, with respect to the use of any information, apparatus, method, or process disclosed in this report or that such use may not infringe on privately owned rights; or (b) assumes any liabilities with respect to the use of, or for damages resulting from the use of, any information, apparatus, method, or process disclosed in this report.

Prepared by Kansas State University

This document was reviewed by a panel of independent experts selected by The Water Research Foundation. Mention of trade names or commercial products or services does not constitute endorsement or recommendations for use. Similarly, omission of products or trade names indicates nothing concerning The Water Research Foundation's positions regarding product effectiveness or applicability.

Page 4: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs iii

Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD Kansas State University, Biological and Agricultural Engineering Department Charles Barden, PhD Kansas State University, Horticulture and Natural Resources Department Project Team: Pabodha Galgamuwa, PhD Kansas State University, Horticulture and Natural Resources Department Alireza Nooraei, BS Kansas State University, Biological and Agricultural Engineering Department Lisa Treese, BLA Kansas City Water Services Tom Jacobs, PhD Mid-America Regional Council WRF Project Subcommittee or Other Contributors Peter Stangel US Endowment for Forestry and Communities Alison Witheridge Denver Water Eric Kuehler US Forest Service Keith Cline Fairfax County Department of Public Works and Environmental Services AJ Lang North Carolina Department of Agriculture and Consumer Services Therese Walch City of Eugene Public Works Engineering WRF Staff John Albert Chief Research Officer Katy Lackey and Katie Henderson Research Managers

Page 5: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

iv The Water Research Foundation

Abstract and Benefits Abstract:

The overall goal of this project was to document the contribution of urban tree systems to stormwater nutrient and volume control in terms of their effectiveness at various scales, cost, desirability, and practicality. To achieve this goal, this project employed a combination of quantitative analyses of existing data, new modeling data, and qualitative program review. Key findings from this synthesis include: (1) rainfall capture by urban tree systems is non-trivial (e.g., 17% to 30% of annual precipitation), and can be reasonably estimated on an annual or event scale based on precipitation depth and tree type; (2) street sweeping is essential to managing phosphorus releases from decaying tree litter, though the rate of phosphorus release could be managed by selecting street trees with lower leaf phosphorus content; (3) tree costs are generally offset by their value when a suite of benefits (e.g., stormwater management, energy savings, increases in property value) are considered. An associated set of tools was developed to enable water utilities and other stormwater professionals to apply project findings, including: (1) simplified models to predict event and annual runoff reductions by urban tree canopy, (2) an urban tree cost-benefit value database, and (3) an urban forest hydrology curricula targeted to utilities to more effectively integrate urban tree systems within stormwater management frameworks.

Benefits:

• Contributions of urban tree systems to stormwater quantity and quality control are synthesized and statistically analyzed

• Simple models are presented to enable the prediction of rainfall capture, throughfall, and/or transpiration on an event or annual basis

• Economic cost and co-benefit values were compiled in a database tool to enable users to assess return on investment

• An educational module was developed to provide a watershed-based curricula aimed at integrating urban forestry with stormwater management

Keywords: Urban trees, stormwater management, interception, water quality, urban watershed, i-Tree Hydro, co-benefits, return on investment

Page 6: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs v

Contents Acknowledgments ........................................................................................................................................ iii Abstract and Benefits ................................................................................................................................... iv Tables ........................................................................................................................................................... vii Figures ......................................................................................................................................................... viii Acronyms and Abbreviations ........................................................................................................................ ix Executive Summary ....................................................................................................................................... xi Chapter 1: Introduction: The Case for Integrating Urban Tree Systems and Stormwater

Management .................................................................................................................................. 1 1.1 Objectives and Research Approach .................................................................................... 1 Chapter 2: State of the Science and Key Knowledge Gaps ......................................................................... 3 2.1 Literature Review Overview ............................................................................................... 3 2.2 Knowledge Gaps and Prioritized Research Needs .............................................................. 4 Chapter 3: Quantifying Urban Tree System Stormwater Benefits: A Meta-analysis ................................. 7 3.1 Meta-analysis Methods ...................................................................................................... 7 3.2 Meta-analysis Results ......................................................................................................... 9 3.2.1 Rainfall Partitioning Regression Equations .......................................................... 10 3.2.2 Transpiration Regression Equations .................................................................... 16 3.2.3 Water Quality Descriptive Statistics .................................................................... 18 3.3 Putting it Together: Application of Meta-analysis Results to Support Stormwater

Volume and Quality Impacts of Urban Tree Systems ....................................................... 26 3.3.1 Will Expanding Canopy Cover in Urban Open Spaces Reduce Pollutant Loads

Delivered to Urban Waters? ................................................................................ 27 3.3.2 Do Street Trees Act as Net Sources or Sinks of Nutrients to Urban

Stormwater? ........................................................................................................ 27 3.3.3 Do Trees Enhance Nutrient Removal Processes in Structure Green

Stormwater Infrastructure Practices? ................................................................. 28 Chapter 4: Scaling Hydrologic Benefits to the Watershed: i-Tree Hydro Modeling ................................ 29 4.1 Hydrologic Modeling Approach ........................................................................................ 29 4.1.1 Model Watershed Descriptions ........................................................................... 29 4.1.2 Model Creation, Calibration, and Performance Assessment .............................. 30 4.1.3 Tree Canopy Cover Scenarios .............................................................................. 32 4.2 Hydrologic Model Results and Implications ..................................................................... 33

4.2.1 Effect of Development Pattern: Comparisons among Study Watersheds and Forested Reference .............................................................................................. 33

4.2.2 Increasing Canopy Cover and the Effects of Landscape Context ......................... 33 4.2.3 Comparison of Meta-analysis Models: Scaling from Tree to Yard to

Watershed ........................................................................................................... 36

Chapter 5: Counting the Costs and Benefits: Urban Tree System Cost and Co-Benefit Database .......... 39 5.1 Urban Tree System Cost-Benefit Database Tool Overview .............................................. 39 5.1.1 Cost-Benefit Database Fact Sheet............................................................................ 41 5.1.2 Cost-Benefit Database User Guide .......................................................................... 41

Page 7: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

vi The Water Research Foundation

5.2 Cost-Benefit Database Takeaways .................................................................................... 41

Chapter 6: Curricula Development to Facilitate Integration of Urban Forestry and Stormwater Management ................................................................................................................................ 43

6.1 Survey of Existing Curricula and Needs ............................................................................ 43 Appendix A: Literature Review – Urban Tree Systems and Stormwater Quantity and Quality

Regulation: A State of the Science ................................................................................................ 45 Appendix B: Meta-analysis Supporting Information ................................................................................... 67 Appendix C: iTree-Hydro Model Calibration and Performance Assessment .............................................. 81 Appendix D: Stormwater and Forestry Survey ........................................................................................... 87 Glossary ..................................................................................................................................................... 115 References ................................................................................................................................................ 117

Page 8: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs vii

Tables 2-1 Factors That Influence How Trees Regulate Runoff Quantity and/or Quality ................................. 4 2-2 Knowledge Gaps Regarding Influence of Urban Tree System on Stormwater Quality and

Quantity ........................................................................................................................................... 5 3-1 Water Quality Processes and Their Representation in the Meta-analysis ...................................... 8 3-2 Summary of Model Variables for Predicting Urban Tree Hydrologic and Water Quality Effects .... 9 3-3 Summary of Studies Included in Meta-analysis of Rainfall Capture and Throughfall in Urban

Tree Canopies ................................................................................................................................ 10 3-4 Predictive Equations for Average Annual and Event Rainfall Capture and Throughfall ................ 13 3-5 Summary of Studies Included in the Urban Tree Transpiration Meta-analysis ............................. 17 3-6 Predictive Equations for Tree Canopy Transpiration ..................................................................... 17 3-7 Extrapolation of Urban Tree Decomposition Models to the Watershed Scale ............................. 23 3-8 Example Questions Related to the Role of Urban Trees in Regulating Stormwater Runoff in

Three Unique Landscape Contexts ................................................................................................ 27 4-1 Characteristics of Study Watersheds ............................................................................................. 30 4-2 I-Tree Hydro Model Inputs and Corresponding Sources of Data ................................................... 31 4-3 Description of Watershed Hydrologic Modeling Scenarios to Test Urban Forest System

Scalability ....................................................................................................................................... 32 5-1 General Co-benefit and Cost Categories Reported in the Literature for Urban Tree Systems ..... 41 6-1 Stormwater Utility Survey Results ................................................................................................. 44

Page 9: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

viii The Water Research Foundation

Figures 3-1 Boxplots of % Rainfall Capture and Throughfall under Evergreen and Deciduous Canopies ........ 11 3-2 Relationship between Annual and Event Scale Precipitation Depth and Rainfall Capture or

Throughfall ..................................................................................................................................... 14 3-3 Test of Throughfall and Rainfall Capture Regression Model Scalability ........................................ 15 3-4 Nutrient Concentrations Measured in Throughfall from Urban Tree Canopies and Direct

Precipitation ................................................................................................................................... 19 3-5 % Mass of Litter N and P Remaining as a Function of Time as Reported in Urban Litterbag

Studies ............................................................................................................................................ 20 3-6 Rate of P Mass Losses from Urban Leaf Litter over Short and Long Time Periods ........................ 22 3-7 Initial P Content Influences the Rate of Phosphorus Mass Loss in Urban Leaf Litter .................... 22 3-8 Box Plots Depicting Spread of Reported Total Dissolved Nitrogen (TDN) and Nitrate (NO3)

Concentrations and Loads Leached from Urban Tree and Turfgrass Systems .............................. 24 3-9 Box Plots Depicting Spread of Reported Total Dissolved Phosphorus (TDP) and Soluble

Reactive P (SRP) Concentrations and Loads Leached from Urban Tree and Turfgrass Systems ... 25 3-10 Box Plots Depicting Influent and Effluent Total Nitrogen (TN) and Phosphorus (TP) Loads by

Vegetation Type in Bioretention Systems ..................................................................................... 26 3-11 Example Nutrient Mass Loss Rates for Hypothetical Coniferous and Deciduous Street Trees ..... 28 4-1 Study Watershed Map ................................................................................................................... 30 4-2 Runoff Depth and Peak Flow for Reference and Urbanized Study Watersheds ........................... 33 4-3 Change in 25-mm Event Runoff Volume and Peak Flow with Increasing % Canopy Cover ........... 35 4-4 Change in 140-mm Event (10-yr 24-hr) Runoff Volume and Peak Flow with Increasing %

Canopy Cover ................................................................................................................................. 36 4-5 Comparison of Meta-analysis Equations for Event-Based Rainfall Capture with Changes in

Runoff Depth Predicted in i-Tree Hydro for Study Watersheds .................................................... 37

Page 10: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs ix

Acronyms and Abbreviations BMP best management practice DBH diameter at breast height ET evapotranspiration GSI green stormwater infrastructure LAI leaf area index I interception MLR multiple linear regression NSE Nash Sutcliffe Efficiency NO3 nitrate Pgr gross precipitation ROI return on investment Rs solar radiation SCM stormwater control measure SF stemflow SRP soluble reactive phosphorus TH throughfall TDN total dissolved nitrogen TDP total dissolved phosphorus TN total nitrogen TP total phosphorus VPD vapor pressure deficit

Page 11: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD
Page 12: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs xi

Executive Summary Green stormwater infrastructure is increasingly promoted and implemented to meet challenges associated with managing urban runoff quantity and quality in communities across the country. Urban trees are viewed as important components of green infrastructure networks; however, unlike their engineered green infrastructure counterparts (e.g., bioretention basins), less is known regarding the capacity of urban tree systems to regulate runoff quantity and quality. This knowledge gap hinders appropriate inclusion of urban tree systems within stormwater crediting and/or trading schemes. Water utilities trying to assess the costs and benefits associated with integrating urban trees into stormwater management programs would also benefit from guidance to quantify the impacts of urban trees on runoff quantity and quality. The overall goal of this project was to document the contribution of urban tree systems to stormwater nutrient and volume control in terms of their effectiveness at various scales, cost, desirability, and practicality. To achieve this goal, the project employed a combination of quantitative analyses, new modeling data, and qualitative program review to synthesize existing data and programs. The information generated from this project is intended to benefit water utilities and the stormwater regulatory, planning, and design communities as they aim to more effectively and formally integrate urban tree systems within stormwater regulatory and management frameworks.

This report is organized along the three primary objectives of this project:

• Objective 1: quantify contributions of trees and urban forests to stormwater quality and volume regulation at various scales. The overarching question addressed through this objective was: To what extent are urban trees/forests effective and scalable tools for stormwater management? This objective was addressed through a literature review (Chapter 2), quantitative statistical analyses of existing hydrologic and water quality data (Chapter 3), and hydrologic modeling (Chapter 4).

• Objective 2: understand the cost implications of tree/forest systems relative to other stormwater infrastructure practices and their value. The overarching question addressed through this objective was: Are urban trees/forests affordable and desirable components of stormwater programs and to the broader community? This objective was addressed through the creation of a database in which reported co-benefit values and economic costs associated with urban tree systems are compiled (Chapter 5). This tool provides a single source from which urban tree cost-benefit studies and return on investment can be estimated from customized queries.

• Objective 3: develop programmatic strategies and catalog existing curricula that may be adapted to effectively integrate trees and urban forests into stormwater management programs. The overarching question addressed through this objective was: How can urban trees/forests be practically integrated into existing stormwater utility programs? The primary outcome of this objective was a training module that can be utilized by stormwater utilities as an educational resource regarding the contributions of healthy urban tree systems to stormwater management (Chapter 6).

Although the answers to the questions posed above depend on the landscape and social contexts in which urban trees are situated, general answers stemming from this project follow:

1. To what extent are urban trees/forests effective and scalable tools for stormwater management? At the tree-scale, an urban tree captures, on average, 27% of the precipitation that falls on its canopy each year, with evergreen trees capturing more than deciduous (20% versus 32%). Through a meta-data analysis, the team produced a set of equations that can be used to predict precipitation capture on an annual or event basis. These equations were shown to scale up to a residential yard or urban forest plot. Soil water use in between storm events can also be an important consideration for

Page 13: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

xii The Water Research Foundation

urban runoff budgets. A complementary set of equations was developed to predict daily water use (also known as transpiration), which averaged 0.05 inches/day/ft2 canopy area, but ranged as high as 0.2 inches/day/ft2 canopy depending on tree species and climate conditions. A fact sheet was developed to demonstrate the use of these equations to provide stormwater managers with data-driven “rules of thumb” to determine, for example, the contribution of the urban tree canopy to runoff reduction targets. With respect to stormwater quality, urban tree litter generally released a pulse of dissolved phosphorus within days of falling, while dissolved nitrogen was released over a period of months. For leaves decaying on pervious surfaces, studies suggest that released nutrients are cycled in the soil underlying trees. However, when litter accumulates in street gutters or other impervious surfaces, nutrients are likely transported directly to storm drains; thus, timely litter management is essential to successfully integrate urban trees with stormwater management goals.

At the watershed scale, the answer is more complex. i-Tree Hydro modeling completed as part of this project indicated that increasing tree canopy alone would not substantially decrease runoff volume for a one-inch storm. However, when tree canopy expansion took place over impervious surfaces and was accompanied by a commensurate decrease in impervious area (to represent replacing existing impervious surfaces with an appropriate pervious planting area at the base of new trees) the effects were more substantial. For example, increasing tree canopy and associated permeable planting area by 50% resulted in a 10% to 30% decrease in runoff volume from the water quality storm across three different urban watersheds. As demonstrated by comparison to a reference forested watershed, the hydrologic impacts of urban development cannot be overcome with urban tree systems. Therefore, other stormwater infrastructure and hydrologically-sensitive development strategies are needed to restore more natural hydrologic regimes in urban watersheds.

2. Are urban trees affordable and desirable components of stormwater programs and to the broader community? For many communities, yes. An urban tree cost-benefit database was developed to compile existing economic costs and monetary valuations of co-benefits (including air quality regulation, carbon sequestration, stormwater control, water quality, energy savings, and increased real estate value) reported in urban tree studies. The median annualized life cycle cost ($51) and co-benefit value ($78) per tree per year (2017 dollars) results in an average return on investment for an urban tree of 50%. The database is structured such that users can query specific sets of co-benefits, tree system types, or other criteria to develop cost-benefit assessments or calculate a return on investment that is more tailored to local conditions. A fact sheet and user guide were developed to support the use of the database, which is available in both Microsoft Access and Excel formats.

3. How can urban trees/forests be practically integrated into existing stormwater utility programs? The short answer to this question is teamwork. The answers to the questions posed above indicate that (1) trees can complement other stormwater management practices to reduce runoff volume, particularly when opportunities to store and transpire additional runoff in the permeable soil systems in which urban trees grow are considered; and (2) life cycle costs associated with planting and maintaining urban trees are justifiable in light of the value of runoff reduction and other co-benefits they provide. These answers also indicate that mature, long-lived trees are most beneficial. Planting and removal costs comprised nearly 20% of median life cycle costs reported for urban trees; therefore, minimizing tree mortality and subsequent replacement is important from an economic standpoint. Furthermore, trees with well-developed canopies capture substantially more rainfall. For example, a 40-ft tall tulip poplar would be expected to capture approximately 7 times more rainfall than a 15-ft tree of the same species, due to the much larger canopy area. Therefore, it is important for water utilities to work with urban foresters, arborists, and/or other professionals with specialized knowledge to ensure trees are planted and maintained to optimize tree health. A survey of stormwater utility managers and municipal arborists from across the country indicated

Page 14: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs xiii

that some water utilities are already working with urban forestry partners to incorporate trees as part of their community’s green stormwater infrastructure. As described in Chapter 6, an educational curriculum was developed to assist other municipalities aiming to do likewise. This curriculum includes a general urban forest hydrology educational bulletin to supplement factsheets describing applications of the rainfall capture prediction equations and cost-benefit database described in Chapters 3 and 5, respectively. These materials are intended to support the practical integration of urban forestry and stormwater management programs.

Related WRF Research • Advancing and Optimizing Forested Watershed Protection (project 4595) • Source Catchments as Water Quality Treatment Assets: Industry Best Practices and Triple Bottom

Line Cost Evaluation of Catchment Management Practices (project 4570)

Page 15: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD
Page 16: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 1

CHAPTER 1

Introduction: The Case for Integrating Urban Tree Systems and Stormwater Management The role of trees in providing a variety of ecosystem services to urban communities – such as regulating urban microclimates and air quality, providing wildlife habitat, and increasing the sense of well-being among individuals – is well recognized (e.g., McPherson et al., 2005; Nowak et al., 2013). Trees also have the capacity to regulate urban hydrologic cycle and water quality, and thus the potential to contribute to stormwater management and, where combined sewer systems still exist, control of combined sewer overflows (Livesley et al., 2016; Moore et al., 2016). While recognition of these and other benefits has spurred adoption of tree planting and preservation campaigns in cities across the U.S. questions regarding the actual effectiveness of such programs remain (Livesley et al., 2016). Answering such questions remains critical to advancing concerted efforts to integrate urban tree/forest systems with urban water resource management. In addition to understanding their effectiveness as a stormwater control, utilities (stormwater, wastewater and/or water) also require information regarding the cost to implement and maintain trees/forests as part of the water management infrastructure.

Recent publication of three separate review articles (Berland et al., 2017; Cappiella et al., 2016; Kuehler et al., 2017) is indicative both of the swelling interest in the contribution of trees to stormwater management goals as well as the availability of data to begin quantifying this role. However, additional synthesis of these research efforts is needed to facilitate uptake by stormwater utilities or others aiming to practically incorporate trees in stormwater management programs. This project was aimed to help bridge this gap between the current state of the science and its application. By systematically analyzing empirical evidence regarding the potential effectiveness of urban tree systems to regulate stormwater quality and quantity at various scales, as well as the costs and benefits associated with these systems, this project is intended to serve as a foundation for future efforts to incorporate tree systems in stormwater regulatory and management frameworks. In the following sections, the project objectives and research approach are described, followed by a description of the report organization and associated project deliverables.

1.1 Objectives and Research Approach The overarching goal of this project was to document the contribution of urban tree systems to urban stormwater nutrient and volume control in terms of their effectiveness at various scales, cost, desirability and, practicality. To achieve this goal, the project team utilized a combination of quantitative synthesis, new modeling data, and qualitative program review to synthesize and analyze existing data and programs. The specific project objectives included:

• Objective 1: quantify contributions of trees and urban forests to stormwater quality and volume regulation at various scales. The overarching question addressed through this objective was: To what extent are urban trees/forests an effective and scalable tool for stormwater management? This objective is intended to address (1) the need for quantitative synthesis of the existing knowledge regarding the capacity for tree/forest systems to regulate stormwater nutrient and volume loads and (2) the need to better understand the role of street- and watershed-scale conditions on this capacity.

• Objective 2: Understand the cost implications of tree/forest systems relative to other stormwater

Page 17: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

2 The Water Research Foundation

infrastructure practices and their value. The overarching question addressed through this objective was: Are urban trees/forests an affordable and desirable component of stormwater programs and to the broader community? This objective will address the need to better understand the desirability of tree/forest systems as reflected by the value of the bundle of co-benefits they provide, as well as to readily enable stormwater utilities to compare economic costs to implement and/or maintain these systems relative to other types of stormwater infrastructure.

• Objective 3: Develop programmatic strategies and catalog existing curricula that may be adapted for use to effectively integrate trees and urban forests in stormwater management programs. The overarching question addressed through this objective was: How can urban trees/forests be practically integrated into existing stormwater utility programs? This objective was intended to provide educational curricula targeted to water utilities to improve their ability to integrate urban forest management into their activities.

This report is organized along the general set of tasks completed to address project objectives. A brief overview of these tasks and associated project deliverables follows. The project objective to which each tasks maps, as well as the report chapter in which it is detailed is included.

• Task 1: State of the science review regarding urban tree systems and their role in influencing stormwater runoff volume and quality (Objective 1; Chapter 2). A literature review was produced as a project deliverable from this task, and is included in Appendix A of this report.

• Task 2: Meta-analysis of tree hydrologic and water quality regulating functions (Objective 1; Chapter 3). The primary deliverable produced through this task was a fact sheet outlining the use of general equations resulting from the meta-analysis to predict quantities of precipitation capture, throughfall and transpiration by urban trees.

• Task 3: Model urban tree system landscape context scenarios to test scalability of stormwater control benefits (Objective 1; Chapter 4). In this task, the i-Tree Hydro model was utilized to examine how changes in urban tree canopy cover could influence watershed-scale runoff responses for three case study watersheds.

• Task 4: Compile a cost-benefit database to enable determination of return on investment for urban tree systems and comparison to costs of other green infrastructure (GI) types (Objective 2; Chapter 5). The primary project deliverables associated with this task include Microsoft Access and Excel database files and a fact sheet and user guide in which the database structure, contents and example applications are described.

• Task 5: Characterize existing efforts to integrate urban forestry and urban stormwater management and existing curricula and training materials to support such integration (Objective 3; Chapter 6). The primary project deliverable from this task was the production of watershed-based urban forestry training curricula targeted to stormwater utilities.

Page 18: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 3

CHAPTER 2

State of the Science and Key Knowledge Gaps A literature review was conducted to characterize the state of the science regarding the influence of urban tree systems on urban runoff hydrology and water quality and identify key knowledge gaps. This review was somewhat qualitative in nature, but laid the groundwork for the more systematic quantitative analysis conducted within the meta-analyses described in Chapter 3. This review built on the findings of three very recent reviews aimed to compile existing work regarding urban trees and their function in regulating runoff quality and/or hydrology (Berland et al., 2017; Cappiella et al., 2016; Kuehler et al., 2017). The review produced as part of this project is provided in its entirety in Appendix A. Key findings from this synthesis report are presented in the following sections. This review was conducted as part of project Objective 1 (to quantify contributions of trees and urban forests to stormwater quality and volume regulation at various scales), through which we aimed to address the question: To what extent are urban trees/forests an effective and scalable tool for stormwater management?

2.1 Literature Review Overview The recent publishing of three separate literature reviews on the topic of urban trees and their contribution to urban stormwater management highlights the timeliness and importance of this topic, but is also indicative of the growing repository of data collected in an effort to quantify this contribution. Since there is much overlap in the literature summarized by these recent reviews, the review prepared herein attempts to synthesize these efforts to highlight areas of consensus and research need. The review also incorporates studies that have been published since the reviews of Berland et al., (2017), Cappiella et al., (2016), and Kuehler et al., (2017). Broad areas of consensus emerging from the research to date include:

• Urban forest systems regulate stormwater hydrology through aboveground (canopy interception) and belowground (enhanced infiltration of incident precipitation and run-on; increasing soil water storage capacity via evapotranspiration). The magnitude to which these processes influence runoff volume depends on both meteorological/climatological (e.g., precipitation characteristics, wind, evaporative demand) and tree-specific (e.g., canopy architecture, phenology, bark characteristics) factors (Table 2-1).

• Climate and tree-specific factors may influence stormwater hydrology at magnitudes meaningful to stormwater managers; however, due to the interplay of these factors and their variability in time and space, it is difficult to generalize their effect, particularly on an event basis.

• Urban tree systems may serve as either nutrient pollutant sinks or sources to stormwater runoff. When implemented as part of structural GI systems, column and field studies indicate trees can enhance stormwater pollutant removal relative to unvegetated infiltration-based systems. The processes by which trees mediate water quality improvements likely include (1) reducing pollutant loads through stormwater volume regulation and (2) enhancing microbial processes that drive pollutant degradation and transformation within the canopy root zone. When implemented near to or over impervious surfaces, tree systems are likely to serve as a source of nitrogen (N) and phosphorus (P) to stormwater as these nutrients leach from litterfall and are exported via hydraulically expedient drainage systems. Therefore, street sweeping or other leaf litter management programs are needed to mitigate unintended nutrient exports from urban tree systems.

Page 19: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

4 The Water Research Foundation

These areas of consensus are expounded upon in the appended literature review (Appendix A).

Table 2-1. Factors That Influence How Trees Regulate Runoff Quantity and/or Quality.

Climate-based Definition Primary Processes Influenced1

Gross annual precipitation Total precipitation depth at a site over a year I, SF, T

Gross event precipitation Total precipitation depth at a site during the course of a single storm

I, SF, T

Precipitation intensity Depth of precipitation occurring per unit time I, SF

Antecedent rainfall Precipitation depth that has occurred prior to a storm event I, SF, T Wind speed Speed of wind at a site, typically measured 1.5 meters above ground level I, SF, T Evaporative demand

Upper limit to the rate of evaporation created by atmospheric driving forces (radiation, wind, temperature, humidity)

I, SF, T

Soil-based

Soil bulk density Mass of soil per unit soil volume; infiltration generally increases as soil bulk density decreases

Inf

Water holding capacity Quantity of water held in soil against gravitational forces. T, Inf

Soil moisture content Amount of water held in the soil at a given point in time T, Inf

Rooting volume Volume of soil or other media that can be occupied by tree roots; indirectly effects all hydrologic processes via tree health

Inf

Soil texture Description rooting volume soils on the basis of particle diameter, including sand (0.05-2 mm), silt (0.002-0.05 mm) and clay (< 0.002 mm) fractions

Inf

Hydraulic conductivity Rate at which water is transmitted through the soil profile Inf

Tree-based Evergreen or deciduous

Evergreen species retain foliage throughout year; deciduous drop foliage seasonally

I, SF, T

Height Height of tree as measured from the tree base to the top of the canopy I, SF, T Diameter at breast height Tree diameter as measured at 1.4 m (4.5 ft) above ground level I, SF, T

Leaf area index Characteristic of tree canopy, generally defined as leaf area per unit ground area I, SF Leaf morphology Describes leaf shape and orientation relative to stem and branches I, SF

Branch angle Describes orientation of tree branches relative to main trunk/bole. Small angles indicate near vertical canopy; larger angles indicate rounder canopy

I, SF

Bark texture Describes relative roughness or smoothness of bark along tree branches and trunk. May change through time in species that slough bark.

I, SF

Root density Volume of roots per unit volume soil I, SF, T 1Indicates predominant hydrologic and water quality response influenced by given variable. Categorized as interception (I), stemflow (SF), transpiration (T), and infiltration capacity (Inf)

2.2 Knowledge Gaps and Prioritized Research Needs In addition to fully scoping the state of the science, research synthesis efforts such as this are needed to more clearly understand remaining knowledge gaps that hinder adoption of urban tree systems for stormwater management. The knowledge gaps identified in this and previous reviews of urban tree systems can be broadly categorized by the type and scale of research needed to fill the gap (Table 2-2).

Page 20: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 5

Table 2-2. Knowledge Gaps Regarding Influence of Urban Tree System on Stormwater Quality and Quantity. Knowledge gap Key questions

Hydrologic regulation

To what extent and at what scale do canopy characteristics influence runoff peak timing and magnitude? To what extent do urban trees enhance soil infiltration capacity? What is the role of urban tree transpiration on stormwater volume reduction?

Water quality regulation What are the driving mechanisms and conditions in which urban tree systems are a pollutant sink versus source?

Scaling to watershed

Do existing watershed models adequately represent urban tree ecohydrologic processes? How do interactions with built infrastructure constrain or enhance hydrologic benefits provided by urban tree systems?

Scaling to socio-ecological systems

How do sociocultural factors influence implementation and function of urban tree systems for stormwater regulation?

In the opinion of the research team, research priorities should shift to studies aimed to scale tree hydrologic functions from the individual tree to the watershed through research aimed to understand how physical processes scale, how to represent these processes adequately in watershed models, and how to account for social and cultural influences on urban tree system structure and function for stormwater management. This opinion is based in part on conversations and informal feedback provided by stormwater professionals over the course of this project. Understanding how urban tree systems function at the watershed scale as a component of socioecological systems is essential to provide a scientific basis for policies incorporating watershed-scale networks of connected tree systems.

Page 21: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD
Page 22: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 7

CHAPTER 3

Quantifying Urban Tree System Stormwater Benefits: A Meta-analysis

As demonstrated in the previous chapter, existing studies clearly indicate urban trees and forests influence stormwater runoff quantity and quality. However, predicting the magnitude of this influence for a given situation is challenging given the complex interactions among numerous climatic, meteorological and tree-specific variables that control the processes through which trees moderate runoff quantity and quality. As part of the project goal to address the question “To what extent are urban trees/forests an effective and scalable tool for stormwater management?” (Objective 1), we conducted a quantitative statistical analysis, known as a meta-analysis. The primary goals of this analysis were to:

• Produce sets of regression models that, although empirical in nature, would provide stormwater managers with a reasonable estimate of the contributions urban trees make to stormwater quantity goals that are easier to use than existing mechanistic models.

• Identify key environmental and/or structural characteristics of urban trees that contribute to their hydrologic and/or water quality functions and their relative importance

• Develop recommendations to prioritize and fill existing gaps in available data.

The following sections present the methods by which the meta-analysis was conducted, the resulting prediction equations, and an example application of the models. Supporting information is provided in Appendix B.

3.1 Meta-analysis Methods Meta-analysis is a statistical tool by which to synthesize independent research studies such that general conclusions can be made while taking into account (potentially) interactive effects of other variables. In this study, separate meta-analyses were conducted related to: (1) rainfall partitioning in the tree canopy, (2) transpiration, (3) and water quality. For all components of the meta-analysis, data were extracted from peer-reviewed literature through database (Scopus, ISI Web of Science, ProQuest) and Google Scholar searches. For cases in which data were presented graphically, the data extraction application WebPlotDigitizer 4.1 (Rohatgi, 2018) was used to obtain x,y coordinate pairs for use in the meta-analysis. To avoid duplicating data points in the dataset, only original data were included. Consistent with recommended meta-analysis procedures, measurements collected from the same species within the same study were averaged to reduce bias in analysis results that may occur when a few studies comprise a disproportionate number of data points (e.g., Brok et al., 2008). Differences in rainfall partitioning and transpiration rates between trees in urban areas and natural forests have been observed (Park and Cameron, 2008); therefore, preference was given to studies conducted in urban settings, though data from open-grown trees (i.e., no overlap in tree canopy and no understory) in rural areas were also considered to compile a more robust dataset. Preference was also given for water quality studies conducted in urban or peri-urban areas due to differences in pollutant loading via atmospheric deposition and other sources, leaf litter composition, and decomposition environments characteristic of urban tree systems relative to rural trees (e.g., Tulloss and Cadenesso, 2015; McDonnel et al., 1997; Dorendorf et al., 2015). An additional requirement for inclusion in the rainfall partitioning or transpiration analysis was that the original data were collected at the individual tree scale. Two studies

Page 23: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

8 The Water Research Foundation

(Bu et al., 2016; Inkilainen et al., 2013) were identified in which throughfall was measured at the plot or yard scales. These studies were not utilized in the analysis, but were used to assess the performance of the resulting models when extrapolated beyond the tree scale (Section 3.2.1).

For the rainfall partitioning and transpiration component of the meta-analysis, multiple linear regressions (MLR) were used to identify statistically significant relationships between predictor variables and reported throughfall, rainfall capture or transpiration rates. Prior to conducting MLR, the underlying assumptions of this statistical approach were checked using Q-Q plots, variance inflation factors (to demonstrate predictor variables were independent of one another), and scatterplots of residuals versus predicted variables (to demonstrate the variance in error terms across the range of predictor variables was similar rather than heteroscedastic). All statistical analyses were conducted using the JMP, Version 14.1 statistical software package (SAS Institute Inc. Cary, NC, 1989-2019). Separate models were developed for the annual and event time scales since utilities may have interest in, for example, estimating annual avoided runoff or the proportion of the water quality storm event that could be captured by urban tree systems.

The water quality component of the analyses was limited to nitrogen and phosphorus water quality data, and was organized conceptually according to the primary processes through which trees may influence water quality: (1) above canopy-level processes including washoff of atmospheric deposition and/or canopy leaching by throughfall; (2) ground surface processes such as litter nutrient release and mineralization; and (3) belowground processes such as plant uptake and microbial nutrient transformation within the root zone (Table 3-1). Due to the small number of water quality studies identified relative to the number of different observed predictor variables and study designs employed across these studies, the meta-analysis approach was limited to determination of descriptive statistics rather than a formal regression analysis. While this approach does not produce predictive equations like the hydrologic component of the meta-analysis, it does provide an indication of the capacity of trees to act as a nutrient sink and/or source in urban environments.

Table 3-1. Water Quality Processes and Their Representation in the Meta-analysis.

Source: Tree sketch from http://www.freestockphotos.biz/stockphoto/15117.

To characterize nutrient releases via decomposition of urban tree litter, exponential decay curves were fit to litter decomposition datasets to provide an indication of the rate at which nutrient mass was lost from decaying litter (Equation 3-1) where M(t) is the mass remaining at time t, M0 is the initial nutrient mass, and k is the decay rate coefficient.

M(t) = M0e-kt (Equation 3-1)

-Ln(M(t)/ M0)= kt (Equation 3-2)

Water quality

process Reference

system Data collected to characterize process

Throughfall washoff & leaching

Bulk precipitation Measured throughfall nutrient concentrations and loads

Litter decomposition

NA Nutrient mass loss rates from urban tree litter

Biological uptake, belowground processes

Managed turfgrass

Leachate nutrient concentrations and loads measured below urban tree root zones (used as proxy for net nutrient retention in root zone); International Stormwater BMP database nutrient load retention in bioretention with trees, herbaceous or no vegetation

4. Li erfall 2. Belowground

Nutrient cycling & immobiliza on

3. Throughfall Washoff & leaching

1. Uptake

Page 24: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 9

The rate of mass loss k was determined from the slope of the line relating the natural logarithm of M(t)/M0 with time t (Equation 3-2). This mathematical manipulation is convenient as it allows the rate of mass loss, k, to be determined as the slope of a line. Higher values of k indicate a higher rate of nutrient mass loss. For studies in which nutrient losses were reported in terms of nutrient content (e.g., mass nitrogen per unit mass remaining leaf material), nutrient mass remaining was obtained by multiplying the nutrient content by the remaining litter mass. Belowground processes were characterized using nutrient loads measured in drainage from tree system root zones as an indication of the net effect of nutrient uptake/sequestration and release. This process was characterized for trees in two different landscape/management contexts: trees planted in urban parks or forests and trees incorporated as part of engineered stormwater management systems (here, bioretention). Data for the latter context was obtained exclusively from the International Stormwater BMP Database (BMP Database, n.d.).

In many cases, it is helpful to compare the relative magnitude of nutrient sequestration or delivery associated with urban trees with a reference system to get a sense of how adding (or taking away) trees would affect nutrient loads. When possible, data for appropriate reference systems (Table 3-1) were also analyzed to provide this context. Statistical texts between nutrient loads associated with urban trees and their reference system were conducted using nonparametric Wilcoxon rank test for two-way comparisons or Steel-Dwass test for multiple comparisons. All statistical tests were conducted using JMP, Version 14.1 software (SAS Institute Inc. Cary, NC, 1989-2019).

3.2 Meta-analysis Results A total of 50 studies were identified that met the requirements for inclusion in the meta-analysis of urban tree hydrologic and water quality benefits (Table 3-2). Although there are many other climatic, environmental and tree-specific variables controlling hydrologic and water quality processes mediated by trees (e.g., Table 2-1), meta-analyses were limited to groups of explanatory variables reported across the collection of studies identified for the analysis. The majority of these studies were conducted in the United States, though studies from Australia and Europe were also identified. The presiding climate at each of these data collection sites was coarsely classified by the Köppen-Geiger climate system (Beck et al., 2018). Additional details regarding the datasets included in each component of the meta-analysis are presented in the following sections as well as in Appendix B.

Table 3-2. Summary of Model Variables for Predicting Urban Tree Hydrologic and Water Quality Effects. Potential explanatory variables for each process reflect variables that were reported across a majority of studies

included in the meta-analysis. Citation information for studies provided in Tables B-1, B-2 and B-5. Process Predicted variables Potential explanatory variables included in

analysis Analysis time scale (n =

number of studies)

Rainfall partitioning

Throughfall Rainfall capture1

Climatic: rainfall depth, intensity, Köppen climate Tree: evergreen-ness, bark roughness, DBH, LAI, Height

Rainfall event (n=10) Annual (n = 16)

Transpiration Sapflow rate Whole tree transpiration

Climatic: Solar radiation, vapor pressure deficit, rainfall depth, Köppen climate Tree: evergreen-ness, xylem anatomy DBH, LAI

Daily transpiration (n= 4) Average seasonal (n = 8)

Water quality

Throughfall nutrient washoff

Tree system: precipitation nutrient content; annual precipitation Annual (n = 14)

Litter nutrient mass release

Leaf litter characteristics: initial N and P content; N:P ratio; C:N ratio, lignin content Annual (n = 8)

Below root zone nutrient leaching

Climatic: annual precipitation Tree: leachate volume Annual (n = 4)

1Rainfall capture reflects the combination of interception and stemflow processes (calculated as the difference between gross precipitation and throughfall).

Page 25: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

10 The Water Research Foundation

3.2.1 Rainfall Partitioning Regression Equations The major fates of precipitation falling on tree canopies are interception, throughfall and stemflow. Of these, throughfall and stemflow are typically measured in the field, and interception is then determined as the difference total precipitation, throughfall and stemflow. Throughfall was measured in all of the studies included in the dataset. However, stemflow was only measured in 11 of the studies and, thus, did not provide a true measure of interception. To maintain consistency in reporting across all datasets, stemflow and interception – which represent the difference between total rainfall and throughfall – were grouped together and termed “rainfall capture.” Implicit in this approach is the assumption that rainfall that is retained on leaf and stem surfaces or that is funneled to the base of the tree via stemflow is effectively removed from the stormwater runoff budget through evaporation or infiltration. Provided that soils at the base of the tree are maintained to promote infiltration, this assumption should be valid. Following this approach, the regression equations were developed to predict either throughfall or rainfall capture. Throughfall and rainfall capture data included in the meta-analysis encompassed a range of tree species, sizes, growth forms, and climate (Table 3-3), which likely improves the transferability of resulting models from one site to another as opposed to if the data were constrained to a narrow set of tree structural characteristics and/or climatic variables. It should be noted that all studies with deciduous trees spanned the leaf on and leaf off periods (with the exception of one study that was conducted only during the leaf off period). Thus, this dataset should represent rainfall partitioning under deciduous canopies across a range in phenological stages.

Table 3-3. Summary of Studies Included in Meta-analysis of Rainfall Capture and Throughfall in Urban Tree Canopies.

All studies presented annual results, but not all studies presented event-based results. See Table B-1 for study citation and details.

Dataset characteristic Annual-scale models Event-based models # unique studies 16 10

# data points 35 1,231 # different species 28 19

Evergreen-ness 14 Deciduous; 21 Evergreen 431 Deciduous; 717 evergreen DBH (cm) (# data points) 10.1 to 95 (25) 10.1 to 95 (1,049)

LAI (# data points) 1 to 7.75 (24) 1.7 to 7 (1,156) Site Koppen climate

classification (# data points) Tropical (7); Temperate-humid (9); Mediterranean (11); Semi-arid (8)

Tropical (155); Temperate-humid (134); Mediterranean (340); Semi-arid (602)

Site annual rainfall (mm/year) 270 to 2130 270 to 2130

Interception and total rainfall capture were significantly greater within evergreen tree canopies, while throughfall was significantly less under evergreen tree canopies relative to deciduous canopies (Figure 3-1). Given this difference, separate predictive equations were developed for deciduous and evergreen trees.

Page 26: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 11

Figure 3-1. Boxplots of % Rainfall Capture and Throughfall under Evergreen and Deciduous Canopies. Rainfall capture (interception + stemflow) was significantly greater within evergreen canopies (p = 0.0015), while

throughfall was significantly higher under deciduous canopies (p = 0.006). Horizontal lines of the box plots represent (from bottom to top) the 10%, 25%, 50% (i.e., median), 75% and 90% quartile values. The solid gray line

in both plots represents the overall mean. Studies from which data were compiled are cited in Table B-1.

Regression models in which rainfall fate (i.e., rainfall capture and throughfall) was expressed as an absolute depth performed much better than when expressed as a relative percentage of total precipitation. Therefore, all regression models produced through the meta-analysis predict rainfall capture and throughfall as a depth. Both the annual and event scale datasets exhibited heteroskedacity; that is, the variability in observed rainfall capture and throughfall rates was not random across the range in precipitation depth but, rather, became increasingly variable with increasing precipitation depth. As noted above, this condition violates the assumptions of linear regression so, to correct for it, throughfall, rainfall capture and precipitation depth were all log-transformed using the natural logarithm. The resulting regression equations are presented in Table 3-4 and described further in the following sections. Users of these equations may choose to translate predicted depths to a relative percentage of rainfall by dividing by the gross annual precipitation (for annual-scale models) or by the storm event depth of interest (for event-scale models). To demonstrate the use of these equations and increase the ease with which they can be applied, a pre-programmed excel workbook and accompanying factsheet were also developed.

At the annual time scale, precipitation was the best predictor of annual throughfall and rainfall capture for both deciduous and evergreen trees, and the relationship was best modeled as a simple linear equation (Figure 3-2; Table 3-4). Adding leaf area index (LAI) improved the model predictions slightly, and accounted for up to 12% of the observed variability in rainfall capture data. Although adding LAI as a predictive variable did not improve model performance substantially, a separate set of LAI-based equations are presented for users interested in quantifying the potential to increase rainfall capture by selecting trees with greater leaf areas (Table 3-4). DBH was also positively correlated with annual rainfall capture and negatively correlated with throughfall but it was not statistically significant and was not included in the final regression models. Significant relationships between throughfall or capture rates and climate classification were not observed, but this may have been associated with the relatively small size of the dataset as opposed to a true lack of climate zone influence at annual scales.

At the event time scale, precipitation remained the strongest predictor of rainfall capture and throughfall. At this time scale, more accurate models were developed when leaf on and leaf off periods for deciduous trees were separated; therefore, equations for three “tree types” were developed: evergreen trees, deciduous trees with leaves on, and deciduous trees during leaf off periods. As with annual scale models, addition of LAI as a predictive variable slightly improved model performance, particularly for capture models. Because some users of these equations may be interested in quantifying

Page 27: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

12 The Water Research Foundation

the extent to which rainfall capture is enhanced by selecting trees with greater leaf area, LAI-based equations are presented. Other tree characteristics and event-specific meteorological variables were found to correlate with rainfall capture and throughfall rates but these correlations were not statistically significant. For example, tree DBH was positively correlated with rainfall capture rates but it did not appreciably improve model fit or predictive power and was not included. Bark structure, which was assigned a relative classification of “smooth” or “rough” based on tree species, was also found to influence precipitation fate, with smooth-barked trees exhibiting higher rates of throughfall and lower rates of rainfall capture than rough-barked trees. However, the effect was small, and uncertainty in associated models was high enough that predictions incorporating bark roughness were not significantly different. Therefore, as with annual scale models, the regression equations presented in Table 3-4 require precipitation depth and tree type as inputs to predict capture and throughfall for individual precipitation events. For users who are also able to obtain LAI estimates for the trees being modeled, equations incorporating LAI as a predictive variable area also presented.

As evidenced by the scatter of points in Figure 3-2, there remains additional, unexplained variability in the observed throughfall and, especially, rainfall capture datasets. We suspect much of this variability can be attributed to site- and event-specific meteorological variables that were not incorporated in this dataset because they were not measured and/or reported in published studies. This supposition is partially supported by the influence of climate zone on observed throughfall and rainfall interception rates. For example, event-based precipitation partitioning data from humid/tropical climates tended to have higher throughfall rates than data collected in Mediterranean climates across the same range in event size (Appendix B), which may indicate vapor pressure deficit and other factors regulating evaporative demands play an important role in determining precipitation fate on an event basis. Average rainfall event intensity (i.e., rainfall depth per unit time) also appears to constrain rainfall capture, with a sharp drop in % rainfall capture (and corresponding increase in throughfall) occurring at precipitation intensities near 4 mm hr-1 (Appendix B). Rainfall intensity data were only available for a subset (2) of event-based studies and the explanatory power of regression equations (R2 = 0.26) was less than that for precipitation depth-based models; therefore, intensity was not included as a variable in final predictive equations despite its likely importance.

Page 28: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 13

Table 3-4. Predictive Equations for Average Annual and Event Rainfall Capture and Throughfall. Given differences between partitioning processes in deciduous and evergreen trees (Figure 3-1), separate

equations are presented. Goodness-of-fit and number of data points used to develop relationship indicated by R2 and n values, respectively. Studies from which meta-regression data extracted are cited in Table B-1.

Tree type Predictive equation: average annual rainfall capture and throughfall

Evergreen Capture (mm): = exp(0.67*ln(Pannual) + 0.98) Throughfall (mm) = exp(1.09*ln(Pannual) – 1.015)

R2 = 0.58 (n=20) R2 = 0.92 (n=21)

Deciduous Capture (mm): = exp(0.67*ln(Pannual) + 0.52) Throughfall (mm) = exp(1.09*ln(Pannual) – 0.88)

R2 = 0.53 (n=14) R2 = 0.98 (n=14)

Equation terms: Rainfall capture and throughfall are in mm/unit evergreen or deciduous canopy area on an annual basis; Pannual is gross annual precipitation depth (mm). These equations not recommended for precipitation depths less than 150 mm.

Tree type LAI-based Predictive equation: average annual rainfall capture and throughfall

Evergreen Capture (mm): = exp(0.82*ln(Pannual) + 0.08*LAI + 0.82) Throughfall (mm) = exp(0.82*ln(Pannual) – 0.02*LAI – 0.99)

R2 = 0.62 (n=21) R2 = 0.92 (n=21)

Deciduous Capture (mm): = exp(0.82*ln(Pannual) + 0.08*LAI + 0.30) Throughfall (mm) = exp(1.09*ln(Pannual) – 0.02*LAI – 0.83)

R2 = 0.65 (n=14) R2 = 0.98 (n=14)

Equation terms: Rainfall capture and throughfall are in mm/unit evergreen or deciduous canopy area on an annual basis; Pannual is gross annual precipitation depth (mm); LAI is leaf area index. These equations not recommended for precipitation depths less than 150 mm.

Tree type Predictive equation: rainfall event capture and throughfall

Evergreen Capture (mm): = exp(0.74*ln(Pevent) – 0.58) Throughfall (mm) = exp(1.21*ln(Pevent) – 1.06)

R2 = 0.69 (n=692) R2 = 0.87 (n=692)

Deciduous – leaf on

Capture (mm): = exp(0.74*ln(Pevent) – 0.82) Throughfall (mm) = exp(1.21*ln(Pevent) – 0.83)

R2 = 0.76 (n=353) R2 = 0.94 (n=353)

Deciduous – leaf off

Capture (mm): = exp(0.74*ln(Pevent) – 0.88) Throughfall (mm) = exp(1.21*ln(Pevent) – 0.63)

R2 = 0.78 (n=76) R2 = 0.92 (n=76)

Equation terms: Rainfall capture and throughfall are in mm/unit evergreen or deciduous canopy area on an annual basis; Pevent is gross event precipitation depth (mm). Minimum recommended event precipitation depth is 1 mm.

Tree type LAI-based Predictive equation: rainfall event capture and throughfall

Evergreen Capture (mm): = exp(0.74*ln(Pevent) + 0.15*LAI – 1.00) Throughfall (mm) = exp(1.21*ln(Pevent) – 0.08*LAI – 0.83)

R2 = 0.70 (n=692) R2 = 0.87 (n=692)

Deciduous – leaf on

Capture (mm): = exp(0.74*ln(Pevent) + 0.15*LAI – 1.16) Throughfall (mm) = exp(1.21*ln(Pevent) – 0.08*LAI – 0.64)

R2 = 0.79 (n=353) R2 = 0.94 (n=353)

Deciduous – leaf off

Capture (mm): = exp(0.74*ln(Pevent) + 0.15*LAI – 1.75) Throughfall (mm) = exp(1.21*ln(Pevent) – 0.08*LAI – 0.32)

R2 = 0.90 (n=76) R2 = 0.92 (n=76)

Equation terms: Rainfall capture and throughfall are in mm/unit evergreen or deciduous canopy area on a storm event basis; Pevent is gross precipitation for a given storm event; LAI is leaf area index. Minimum recommended event precipitation depth is 1 mm

Page 29: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

14 The Water Research Foundation

Figure 3-2. Relationship between Annual and Event Scale Precipitation Depth and Rainfall Capture or Throughfall.

Precipitation and evergreen-ness (annual scale: evergreen or deciduous; event scale: evergreen, deciduous leaf off or deciduous leaf on) explained the majority of variation in observed rainfall capture and throughfall at the annual

scale (left panel) and event scale (right panel). Data were log transformed to correct for heteroscedasticity of observed data. Studies from which data points comprising graphs were extracted are cited in Table B-1.

3.2.1.1 Scalability of Annual Throughfall and Rainfall Capture Models It should be emphasized that the regression equations presented in Table 3-4 reflect rainfall capture and throughfall rates for individual urban trees. However, it is likely that users of these models would want to extrapolate these models to application at larger scales (e.g., development site, street, watershed). We are not aware of a mechanism that would cause the relationship between throughfall and rainfall capture rates as presented here (on a per unit canopy basis) to scale in a non-linear fashion and so believe that these models can be linearly extrapolated. To test this supposition, the scalability of annual scale models was tested against two independent datasets in which throughfall was measured at larger scales. In the first dataset, yard-scale throughfall measurements were reported for 16 residential lawns in the Raleigh, North Carolina area (Inkiläinen et al., 2013). The second study reported plot-scale throughfall for urban riparian buffers (Bu et al., 2016). The range in canopy cover represented across both of these studies spanned 30% to 90%. In general, our model tended to under predict throughfall (Figure 3-3), thus over predicting rainfall capture, with percent error1 between measured and modeled values ranging from 1% to 13% (mean 6%). This range in performance was in line with the regression models developed by the authors of these studies specific to their respective datasets. Therefore, we believe that the regression models presented in Table 3-4 can provide meaningful information to stormwater managers at a variety of spatial scales in in spite of their simplicity and uncertainty in 1 Percent error calculated as the absolute value of (measured value – modeled value)/measured value x 100%

-3

-2

-1

0

1

2

3

4

ln(C

aptu

re, m

m)

-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5ln(Precip, mm)

-5

-4

-3

-2

-1

0

1

2

3

4

ln(T

hrou

ghfa

ll, m

m)

-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5ln(Precip, mm)

ELeaf offLeaf on

4

4.5

5

5.5

6

6.5

7

ln(C

aptu

re_a

nnua

l, m

m)

5.5 6 6.5 7 7.5ln(P_annual, mm)

4.5

5

5.5

6

6.5

7

7.5

8

ln(T

hrou

ghfa

ll, m

m)

5.5 6 6.5 7 7.5ln(Precip, mm)

Evergreen-ness

DeciduousEvergreen

Page 30: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 15

predicted values. This is, of course, just one instance to demonstrate model performance, and additional evaluation is needed at both annual and event time scales.

Figure 3-3. Test of Throughfall and Rainfall Capture Regression Model Scalability. Regression equations were applied to predict throughfall as a function of gross precipitation and canopy cover at

the yard or plot scales and compared to measurements from two independent datasets (Bu et al., 2016; Inkiläinen et al., 2013). 1:1 line (solid black line) indicates regression equations obtained from the meta-analysis tended to

underpredict observed throughfall rates, with % errors ranging from 1% to 13%.

3.2.1.2 Strengths and Limitations of Precipitation Partitioning Meta-analysis Models The intent of these equations is to provide relatively simple models of rainfall capture (interception and stemflow) and throughfall rates at annual and storm event scales. The sum of interception and stemflow components (or, conversely, the difference between gross precipitation and throughfall) represents the portion of direct rainfall that is potentially captured in the tree canopy or delivered to the tree base (and infiltrated assuming adequate infiltration capacity). As indicated by the R2 values in Table 3-4, precipitation depth explained 40% to 50% of the variability in observed rainfall capture rates reported among the 16 studies and over 80% of variability in throughfall measurements. Inclusion of DBH in the event-scale rainfall capture model for deciduous trees accounted for an additional 5% of observed variability. Strengths and limitations of these regression models are listed here:

Strengths of Model Simplicity Evergreen-ness, Pgr and DBH are relatively easy to determine. Thus, stormwater managers or other users of this model could obtain event- to annual-scale estimates of urban tree canopy interception with less parameter uncertainty than mechanistic models involving many parameters such as the modified Rutter equation utilized in i-Tree Hydro (Wang et al., 2008). These equations are also relatively easy to apply, and may be more desirable for obtaining planning-scale estimate of rainfall capture by urban tree canopy.

Limitations of Model Simplicity There is still a lot of unexplained variability (50% to 60%) in observed rainfall capture data that are not explained by Pgr and DBH. This indicates that there are other important variables controlling interception rates. For example, tree characteristics such as canopy shape and leaf/stem orientation have been shown to influence interception rates substantially (Livesley et al., 2014). In addition, this simple model does not capture the dynamic nature of variables such as LAI, which we suspect plays a substantial role in interception processes by which we are not able to adequately represent with available data.

Strengths of Model Approach The regression equations presented in Table 3-4 are based on data reported in 16 studies specific to urban trees or open canopy tree systems representative of urban tree growing conditions collected

Page 31: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

16 The Water Research Foundation

across a variety of climate zones (see Appendix B). Thus, these simple models should be generally representative of urban tree systems.

Limitations of Model Approach As with all empirical models, these models are limited to the experimental data upon which they are based and, as result, may not perform as well as models based on physical processes (such as the Rutter equation utilized in i-Tree Hydro). This limitation may be partially addressed by additional testing against data that were not used to develop regression equations, but, moreover, this is a limitation of regression models in general. An additional limitation of this approach is that predictive variables are limited to the set reported in a majority of studies. In this instance, variables such as LAI, gap fraction or other variables that may influence interception were not reported by enough studies to incorporate in the meta-analysis.

3.2.2 Transpiration Regression Equations Transpiration represents the process by which trees and other vegetation extract water stored in the soil profile to meet water demands for growth, evaporative cooling, and other processes. Reviews of transpiration rates from non-urban forests have been performed (e.g., Wullschleger et al., 2001; Ford et al., 2011); however, urban trees typically grow in isolated or otherwise open conditions in which leaf surfaces in both the horizontal and vertical directions may be more exposed to climatic drivers of transpiration processes (Kjelgren et al., 2016; Peters et al., 2010). Therefore, the meta-analysis conducted herein focused exclusively on trees grown in urban areas and/or in open canopy conditions (that is, no canopy overlap with adjacent trees). Other criteria for inclusion in the transpiration data set in addition to those stated in Section 3.1 included: (1) study presented original field measurements of water use in an urban/open canopy system (2) the measurement method followed established/reliable methods by which to scale water use measurements to the whole tree; and (3) transpiration rates represented periods in which trees were not water limited. This latter criterion was added to address inconsistencies in the way soil moisture was reported (or not reported at all) in studies and unreported quantities of irrigation or other water subsidies. As a result, transpiration rates considered in this analysis reflect peak water use following rainfall/runoff, which may be of greater interest to stormwater managers who are interested in the potential to dry out soil in between rainfall events. By only using periods in which water availability was not a limiting factor to transpiration, soil moisture and precipitation can essentially be removed from predictive equations such that the influence of other climate and tree factors can be examined. This criterion was determined on the basis of soil moisture or soil water potential measurements when reported or statements from study authors that trees had access to adequate water if soil moisture was not reported.

Eight studies encompassing transpiration measurements on over 140 trees representing 29 different species were identified that met these criteria (Table 3-5). As with the precipitation partitioning meta-analysis, prediction equations were developed to estimate tree water use at two scales: the seasonal scale and the daily scale. Either scale could be applicable to stormwater management for the purposes of, for example, estimating the quantity of water trees can remove from the underlying soil profile as averaged over the growing season (seasonal scale) or the maximum rate of water extraction during the peak of the growing season (daily scale). For comparison among studies, all transpiration rates were expressed on a per unit canopy area basis (mm/day/m2 canopy area). As indicated in Table 3-5, each tree species in the dataset was classified by its xylem or wood anatomy, which has been shown to influence tree transpiration, particularly at high vapor pressures (Ford et al., 2011). Additional detail regarding this classification is provided in Appendix B. Briefly, the size and distribution of vessels through which trees transport water within their sapwood can be classified as diffuse porous (in which vessels are relatively evenly distributed throughout sapwood), ring porous (in which vessel size decreases within

Page 32: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 17

each growth ring), or tracheid (typical of conifers, in which water transport occurs in small conduits known as tracheids rather than larger vessels in diffuse and ring porous species). Of these, diffuse porous trees have been associated with higher water use because, whereas other trees tend to close their stomates at higher vapor pressure deficits, diffuse porous woods are able to continue transpiring (Ford et al., 2011). Within the data analyzed here, examples of diffuse porous trees included maple, linden, and sycamore, while ring porous species included oak, elm and ash trees.

Table 3-5. Summary of Studies Included in the Urban Tree Transpiration Meta-analysis. Citation information and additional details for the studies summarized in this table are provided in Table B-2.

Tree Characteristics Study (n = 8) site variables # trees 144a Annual rainfall (mm) 270 to 2130 # different species 29 Median study period climate variable valuesb (semi-arid sites; temperate sites)

Evergreen-ness 25 Deciduous 10 Evergreen

Wood anatomy

15 Diffuse porous 9 Ring porous 1 Semi-ring porous 8 Conifer 2 Brachychiton

DBH (cm) 2.6 to 67 (n = 30) Vapor pressure deficit Solar radiation Temperature

1.7 kPa; 1.1 kPa 238 W/m2; 208 W/m2 24 C; 21.8 C LAI 1.7 to 17.8 (n = 11)

areduced to 35 datapoints after averaging across species within studies b data points grouped by temperate (n = 18) and semi-arid (n = 17) study sites. Climate variables were averaged over the same period (e.g., growing season) as transpiration rates were reported.

Predictive equations for urban tree water use on a seasonal and daily basis are presented in Table 3-6. For clarity, this presentation is limited to general relationships that should be applicable across climate zones described in Table 3-5. Other relationships between transpiration and tree trait/climate variables were explored; however, due to limitations in how these data were (or were not) reported across studies, these relationships are reserved for Appendix C as they are not believed to represent the full dataset well. Additional details regarding relationships between transpiration rates and other tree trait and climate variables are provided in Appendix B. As indicated by the R2 values in Table 3-6, these equations do not explain all of the variability in transpiration rates compiled from urban tree studies herein. If more precise estimates of tree water use are required, tree-specific approaches such as the modified reference ET method proposed by Litvak et al., (2017) could be utilized.

Table 3-6. Predictive Equations for Tree Canopy Transpiration. Citation information for studies used in developing regression equations is provided in Table B-2.

Equation type Predictive equation: daily tree water use Tr Equation fita Average daily water use (averaged over growing season; mm d-1)

Diffuse porous: Tr = 0.048*Rs – 0.43*T + 0.55 Ring porous: Tr = 0.031*Rs + 0.21*T – 10.81

Conifer: Tr = 0.023*Rs – 3.20*VPD – 0.39

R2 = 0.58 (n = 12) R2 = 0.79 (n = 8) R2 = 0.58 (n = 8)

Daily water use (mm d-1) Diffuse porous: Tr = exp(-2.97+0.56*ln(Rs) + 0.2*ln(VPD)) Ring porous: Tr = exp(-3.02+0.56* ln(Rs) + 0.2* ln(VPD))

Conifer: Tr = exp(-3.20+0.56* ln(Rs) + 0.2* ln(VPD))

R2 = 0.28 (n = 235) R2 = 0.29 (n = 349) R2 = 0.40 (n = 536)

Equation terms: Transpiration (Tr) is in mm/day/m2 canopy area; Rs is average daily solar radiation in W/m2; VPD is average daily vapor pressure deficit in kPa; T is temperature in degrees Celsius; Equations most appropriate for solar radiation values ranging from 40 W/m2 to 350 W/m2 and VPD up to 3 kPa. aEquation fit indicated by R2 value. Value in parentheses denotes number of data points comprising each regression equation

A fact sheet demonstrating the use of transpiration models in conjunction with precipitation partitioning models was developed as part of this project. Applying these regression models requires the user to

Page 33: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

18 The Water Research Foundation

determine values of input climatic variables (solar radiation, temperature and/or vapor pressure deficit) in the appropriate units. For example, if one wanted to determine an average daily transpiration rate for the purpose of estimating how much water trees planted in a bioretention or other green infrastructure system could use in between runoff events to recover the soil water storage capacity of the system at a site where the average daily VPD is 2.5 kPa and solar radiation is 250 W m-2, estimated water use would range from 1.1 mm d-1 for conifers (e.g., pine or spruce) to 1.3 mm d-1 and 1.4 mm d-1 for ring porous trees (e.g., oaks) and diffuse porous trees (e.g., maple), respectively.

3.2.2.1 Strengths and Limitations of Transpiration Meta-analysis Models The intent of this analysis was to develop tools to support estimates of tree water use in the landscape or within stormwater controls that could be applied relatively easily by stormwater managers or designers. To this end, the simplicity of the equations presented in Table 3-4 relative to other, more mechanistic transpiration models is the primary strength of this approach. Herein also lays the primary weakness of these models. Transpiration is a complex process dictated by a multitude of physical and biological controls. The regression equations presented in Table 3-6 greatly oversimplify this process, using tree wood type as a surrogate for biological variables while reducing the myriad of physical climate factors to solar radiation, vapor pressure deficit and temperature. Therefore, these equations should be regarded as planning-level estimates of tree water use.

3.2.3 Water Quality Descriptive Statistics Trees can influence water quality through a variety of different physical, chemical and biological processes. In this literature survey, these processes were lumped into three primary categories: (1) canopy wash-off and leaching; (2) litter decomposition and (3) belowground leaching. Thirty studies were identified in which data for one or more of these processes in urban or peri-urban tree systems were reported. As noted in the methods, this review was limited to nitrogen and phosphorus forms and did not consider metals or other ions. These data are presented in the following sections to better characterize the magnitude of nutrient delivery or sequestration through each of these three process categories. A tree-scale nutrient budget is then developed to illustrate relative nutrient mass loading rates in three different urban tree landscape contexts.

3.2.3.1 Canopy Wash-off Pollutants deposited on tree canopy surfaces can be washed off during rainfall events and delivered to urban ground surfaces via throughfall. Indeed, studies traversing regional urbanization gradients have identified canopy throughfall as a contributor to “hotspots” of nitrogen deposition in rural and urban landscapes (Tulloss and Cadenasso, 2015). In this review, 14 studies were identified in which throughfall deposition of nitrogen and phosphorus was measured under urban or suburban tree canopies. Of the 14 studies, only 4 reported throughfall phosphorus content. Most of these studies also presented nutrient content of precipitation falling at the site, thus providing a useful reference against which to compare throughfall concentrations. Across all studies, throughfall nutrient concentrations were significantly greater than precipitation (Figure 3-4). It is generally believed that throughfall nutrient fluxes are the result of physical washoff of atmospheric deposition as opposed to nutrients leaching from the leaves themselves (Cappellato and Peters, 1995). Therefore, observed nutrient enrichment in throughfall is indicative of the role of urban trees in filtering and concentrating dispersed air pollutants within their canopies. It can also be conjectured that nutrients delivered to urban surfaces via throughfall do not constitute a new pollutant source but, rather, a concentrated delivery of atmospherically derived substances. Still, throughfall that falls directly on paved surfaces represents a nutrient load that stormwater managers should consider. Conversely, if the tree canopy overhangs permeable surfaces, it is likely that throughfall nutrient loads are sequestered in underlying soils and vegetation (as evidenced in Section 3.2.3.3). While an extensive review of nutrient washoff concentrations from other urban

Page 34: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 19

surfaces was not undertaken herein, it should be noted that total nitrogen concentrations observed in throughfall (median values of 3 mg l-1) were similar in magnitude to mean concentrations reported in national stormwater quality surveys (Scheuler, 2003). However, the proportion of NO3- tended to be about double that observed in runoff from rooftops and streets (Yang and Toor, 2017), which is also indicative of the atmospheric origin of nitrate loads in throughfall.

Stemflow can also serve as a concentrated pathway for nutrients and other soluble pollutants that have accumulated on canopy surfaces and, provided the opportunity to infiltrate, stemflow-derived nutrients will also be cycled via belowground processes. While reported stemflow nutrient concentrations are similar to those reported in throughfall (Chiwa et al., 2003; Johnson and Lehmann, 2006; Xiao and McPherson, 2011b), stemflow nutrient loads are relatively small volume of stemflow that is generated relative to throughfall (Section 3.2.1). Therefore, nutrient load delivery via stemflow was not considered explicitly in this analysis.

Figure 3-4. Nutrient Concentrations Measured in Throughfall from Urban Tree Canopies and Direct Precipitation. The number of data points n is given for each dataset, which included ammonium (NH4), nitrate (NO3) and total

nitrogen (TN), Soluble reactive phosphorus (SRP) and total phosphorus (TP) concentrations. Significant differences between precipitation and throughfall concentrations are indicated with a ** (p < 0.05) or * (p < 0.1). Citation

information for data used to create box plots is provided in Table B-5.

3.2.3.2 Leaf Litter Decomposition While nutrients release from decomposing tree litter constitutes an essential nutrient source to aquatic ecosystems in natural forested watersheds (Benfield, 1997), this process may be viewed as a pollutant source to stormwater in urban watersheds (Selbig, 2016). This may be particularly true for litter that has accumulated on impervious surfaces, the high hydraulic connectivity of which limits opportunity for nutrient retention and/or transformation via biological processes. To characterize the potential magnitude of nutrient releases from decaying tree litter to urban stormwater systems, rates of nutrient loss from decomposing urban tree litter were compiled from the literature. Nutrient losses via decomposition are generally measured in one of two ways: laboratory leaching experiments, in which litter fragments are mixed with deionized water in a test tube and monitored for dissolved nutrient concentrations over time, or via litterbag experiments, in which small mesh bags are filled with representative tree litter samples, deployed in the field for a set period of time and then collected to measure changes in litter mass and nutrient content. Although laboratory leaching experiments offer a more controlled and repeatable approach for measuring nutrient leaching rates, only litterbag experiments were considered herein as they are considered more representative of the conditions and timeframes under which urban leaves undergo decomposition. However, we did consider laboratory leaching studies to support interpretation of these results. A total of eight studies were identified in

Page 35: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

20 The Water Research Foundation

which litterbag experiments were employed to measure phosphorus and/or nitrogen release from urban tree litter. The majority (6 of 8) measured decomposition rates on permeable surfaces only (e.g., soil surface below the tree canopy) while two considered decomposition rates on impervious surfaces as well. While decomposition processes may differ between permeable and impervious surfaces (Bratt et al., 2017), there were not enough data to indicate differences were statistically significant. Therefore, all data were lumped together regardless of decomposition environment.

Before delving into the analysis of nutrient dynamics observed this collection of urban leaf litter decomposition studies, it is instructive to consider conceptual models of leaf litter decomposition. In general, nutrient dynamics follow three phases: an initial leaching phase, followed by microbial immobilization and mineralization (Blair, 1988). During microbial immobilization, nutrient uptake by microbes that colonize leaf litter may increase N and P content beyond initial levels. Mineralization represents the process by which microbes break down organic matter and release previously immobilized nutrients. The time scale over which these processes take place can vary from hours to years; for example, decaying leaves have been observed to leach a substantial proportion of phosphorus over the scale of hours to a few weeks (Bratt et al., 2017; Dorney, 1986; Hobbie et al., 2014) while net nitrogen mineralization may continue for several years (Pavao-Zuckerman and Coleman, 2005; Pouyat and Carreiro, 2003). Leaf litter mass, on the other hand, tends to decrease through time and is often fit to single or double exponential decay functions (Adair et al., 2010; Hobbie et al., 2014). From a stormwater management standpoint, the primary interest is not so much in the mass of remaining leaves or their nutrient content but in the product of these two quantities: the mass load of nutrients remaining (or lost). Therefore, in presenting the data for urban leaf studies, all nutrient quantities are reported in units of mass. Changes in total phosphorus and nitrogen mass reported across all urban leaf litter decomposition studies are presented in Figure 3-5. Since mass losses from decaying leaves tend to follow an exponential decay function (Equation 3-1), we attempted to fit exponential decay functions to nutrient mass data as well. Doing so provides a useful metric for comparing the relative rate of nutrient mass loss in the decay constant k term that integrates across all decomposition and microbial nutrient cycling processes.

Figure 3-5. % Mass of Litter N and P Remaining as a Function of Time as Reported in Urban Litterbag Studies. The exponential decay constant k (yr-1) was obtained for all datasets. While overall changes in litter mass followed

an exponential decay pattern (left panel), nutrient dynamics are poorly described by the exponential decay function as indicated by low R2 values. Citations for data used to create plot provided in Table B-5.

Bulk litter decomposition generally followed an exponential decay function across all studies. In contrast, nutrient dynamics were more complex and defied the explanatory power of both single and double (not shown) exponential decay functions, particularly for phosphorus. The scatter of points both above and below initial litter TN and TP masses (100% on the y-axis of Figure 3-7) indicates microbial immobilization and mineralization processes were at play within the litter, though the relative timing

Page 36: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 21

and duration of these processes varied across species and studies. It is not surprising that these data do not conform to a simple mathematical expression as leaf litter in these studies was subject to different environmental (e.g., temperature, moisture regimes, nutrient availability in decomposition environment, decomposer communities) and litter traits (e.g., leaf litter lignin content and other characteristics regulating the ease with which decomposer communities can access litter nutrients) that regulate decomposition. In an attempt to better understand nutrient release rates observed across studies and species, separate exponential decay curves were fit to each species and, when possible, to species as grouped by nutrient content or other tree trait factors (Appendix B). General insights are outlined in the points below.

Phosphorus Losses Occur over Short Time Scales…Unless Litter Is P-Deficient Substantial phosphorus losses from fallen leaves have been observed to occur over the span of a few weeks in a variety of field conditions (e.g., Bratt et al., 2017; Enloe et al., 2015). Likewise, litter has been observed to lose up to 60% of its phosphorus mass within 1 to 2 hours in laboratory leaching experiments (Bratt et al., 2017; Dorney, 1986). Phosphorus releases from decomposing urban tree litter were also observed to occur over a relatively short time frame in the dataset presented herein; this trend is particularly evident when data are plotted as the log of the mass fraction of phosphorus remaining (Figure 3-6). In this plot, the slope of the line relating the fraction of remaining phosphorus and time represents the decay constant k (Equation 3-1). As indicated by the steeper slope of points in the figure, TP mass loss rates were higher early in decomposition study period. This is important from a stormwater management standpoint, as leaf collection activities are typically managed to collect leaves within a few weeks to months of leaf fall, a time that corresponds with the greatest net losses of TP. To better represent rates of P leaching immediately following litterfall, separate exponential decay functions were fit to a “quick loss” and “slow loss” period. The best fit across the datasets was obtained when the quick loss period was set to 0.2 years (10 weeks). Beyond this period, P losses no longer exhibited a significant relationship with time, indicating that other environmental conditions associated with microbial immobilization and mineralization processes were likely more important. Individual tree species generally followed the “quick” and “slow” loss trend as well; however, resulting k values were only significant for F. pennsylvanica (k = 5.7) and Quercus species (k = 5), likely due to the small number of data points relative to the highly variable processes of phosphorus immobilization and mineralization taking place in all datasets (Appendix B).

In observing the scatter of P mass loss rates across different studies and tree species, a logical question arises: Is the litter of some tree species more susceptible to P loss than other species? While our dataset is not large enough to statistically test for differences in species-specific P loss rates, we can begin to detect trends between predicted loss rates and litter composition. Of the litter composition characteristics that were reported across most of the studies in this dataset (lignin content, initial N and P content, C:N ratio, P:N ratio), the initial P content was the best predictor of quick and slow P losses (Figure 3-9), although these loss rates also exhibited a significant negative correlation with the litter N:P ratio. Thus, we expect that species whose litter has a high initial P content (which is typically coincident with a low N:P ratio) are likely to have litter loss rates on the high end (i.e., exponential decay rates on the order of 5 yr-1 or more) while trees with low initial P content will have lower P loss rates. In this urban tree dataset, the litter of Pinus species had the lowest initial litter P content (0.4 g kg-1) and actually gained P mass in the initial 10 weeks of the decay period (as indicated by the negative k value in Figure 3-7) before releasing P later in the decomposition process. Litter quality has been shown to exert a strong control on decomposition in rural forests (Cornwell et al., 2008) and, therefore, similar controls should be expected in urban environments. However, additional data are needed to understand the relative magnitude of differences among species with different litter characteristics and within decomposition environments unique to urban areas (e.g., street gutters).

Page 37: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

22 The Water Research Foundation

Figure 3-6. Rate of P Mass Losses from Urban Leaf Litter over Short and Long Time Periods. P exhibits a quick loss period in the initial stages of decomposition, then follows a slower loss period in which

changes in P mass are highly variable and probably driven more by environmental conditions than time. Studies from which data extracted cited in Table B-5.

Figure 3-7. Initial P Content Influences the Rate Of Phosphorus Mass Loss in Urban Leaf Litter. Decay rate constants (k) were determined for short-term P mass losses for litter with high (> 1.3 g P kg-1; k = 4.8 yr-

1), medium (0.5 – 1.3 g P kg-1; k = 2.9 yr-1), and low (< 0.5 g P kg-1; k = 0.2 yr-1) initial P content. Decay rates for the high and medium categories were significant at the p < 0.01 level; the fit for the low P content group was not

statistically significant. Studies from which data were extracted are cited in Table B-5.

Nitrogen Losses Occur over Longer Time Periods Relative to Phosphorus In contrast with the phosphorus dataset analyzed herein, the nitrogen dataset exhibited a positive but weak correlation with time over the first 8 weeks of the collective set of decomposition studies, followed by a period of nitrogen release (k = 0.32 yr-1; p = 0.0001). Individual tree species generally followed this trend as well; however, resulting regression equations were not statistically significant, likely due to the small number of data points relative to the highly variable processes of nitrogen immobilization and mineralization (Appendix B). Initial N content was not a good predictor of loss rates; rather, N losses appeared to be influenced by the relative amounts of carbon and phosphorus remaining in the litter. Laboratory litter decomposition studies have found N:P ratios controlled nitrogen and

Page 38: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 23

phosphorus releases (Gusewell and Gessner, 2009), and, N releases herein were correlated with N:P ratios greater than 20, although this trend was not statistically significant. Additional analysis of other environmental and litter trait factors included in this dataset and/or collection of additional nutrient loss data in decomposition environments unique to urban areas is warranted to improve understanding of nitrogen loss dynamics from urban leaf litter.

Despite Periods of Nutrient Mass Gain, Most Datasets Exhibited Net Nutrient Losses from Decaying Tree Litter Leaf litter has been observed to “scavenge” nutrients, particularly nitrogen, as microbes take up nitrogen and phosphorus from the decomposition environment (e.g., surrounding soil, potentially stormwater when litter is exposed to runoff) and incorporate them into the microbial biomass through a process known as immobilization. This is likely the process that resulted in nutrient masses that exceeded the initial nutrient mass in some datasets. This condition was more common for nitrogen (43% of observations) compared to phosphorus (31% of observations). However, most datasets ultimately reflected nutrient mass losses over the study period, particularly in the case of phosphorus (see point on short term leaching above). Extending these results to urban stormwater management, it is important to acknowledge the dual role of tree litter as both a source (through leaching and mineralization) and sink (through immobilization) of nutrients. However, as will be discussed in the following section, retaining leaf litter on impervious surfaces with hopes of net nutrient uptake through microbial immobilization probably is not a good strategy based on evidence from litter removal studies (Selbig, 2016).

The Nutrient Mass Loss Models Presented Herein Scale to the Watershed The exponential decay models produced from the compiled dataset herein were applied to watershed-scale monitoring studies designed to estimate nutrient loads attributable to urban tree leaf decomposition (Bratt et al., 2017; Selbig, 2016). These watershed studies were all conducted in the upper Midwest (Minnesota and Wisconsin) which may introduce bias, but still provides a means to assess the accuracy and scalability of decomposition models. By constraining nutrient mass decay constants values k and time to coincide with the dominant species and timelines over which watershed nutrient loads were monitored in these studies, estimated nutrient loads from these models compared favorably with watershed-scale study estimates (Table 3-7). While additional testing of litter nutrient mass models is warranted, the initial assessment presented in Table 3-7 indicates these models could be useful to stormwater managers for estimating nutrient load contributions of street trees given tree types and leaf collection frequencies.

Table 3-7. Extrapolation of Urban Tree Litter Decomposition Models to the Watershed Scale. Citation Urban canopy and study description Estimated leaf litter inputs Model parameter values

Study authors This study TDN (kg)

TDP (kg)

TDN (kg)

TDP (kg)

k (yr-1) t (yr)

Bratt et al., 2017

0.17 km2 watershed; 51% impervious; 38% mean impervious canopy cover; oak species and sugar

maple dominant; watershed outlet monitored Dec – March 2012/13 and 2013/14.

NA

3.4 10.4 3.1 kN = 1 kP = 5.9

0.33

Selbig, 2016

0.01 km2 watershed; 45% impervious; 64% canopy cover (17% street canopy); seasonal nutrient loads

from 2 years watershed monitoring

0.18 0.13 0.18 0.17 kN = 0.32; kp = 5.9

0.25

3.2.3.3 Belowground Processes Trees play direct and indirect roles in augmenting nutrient cycling and sequestration through a variety of processes occurring within the root zone. These processes include direct uptake through tree roots, immobilization in microbial biomass living in association with tree roots, nitrification and denitrification

Page 39: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

24 The Water Research Foundation

by soil microbes residing in the root zone that likely utilize root exudates as a carbon source (Bettez and Groffman, 2012; Groffman et al., 2002), and by influencing soil C:N ratios and other soil properties related to increased nutrient retention (Livesley et al., 2016). These processes control nutrient delivery to groundwater, which could serve as an important pathway by which nutrients are delivered to urban waterways via baseflow (Nidzgorski and Hobbie, 2016; Janke et al., 2014). As an indicator of the role of urban trees in controlling nutrient leaching to groundwater, this literature survey focused on nutrient concentrations reported in soil pore water observed below the primary rooting depth of urban trees. Through the literature review process, four studies were identified in which nitrate and/or dissolved phosphorus below urban trees were reported. As with other water quality processes, it is useful to have a reference against which to compare nutrient leaching rates below trees. As a likely alternative to tree canopy cover in urban areas from which leaching could occur, leaching rates below turfgrass systems were selected as this reference. Two of the four tree leaching studies also measured leaching rates below turfgrass systems; three additional turfgrass leaching studies were identified to supplement this dataset. It should be noted that, with the exception of one turfgrass system that was not fertilized at all, all other turfgrass systems included in this dataset were fertilized at recommended rates.

Average nutrient concentrations and loads reported for urban tree and/or turf systems in this collection of studies are presented as boxplots in Figures 3-8 and 3-9. As with throughfall leaching studies, the majority of work to date has focused on nitrate leaching, with only one study reporting total dissolved nitrogen (which includes organic forms) or phosphorus forms leaching rates below urban trees (Nidzgorski and Hobbie, 2016). Due to the limited number of studies, observed leaching rates cannot be extrapolated to other urban tree systems with high confidence. However, these data do provide an indication of the potential magnitude of nutrients leaving urban tree systems below ground, and, for the majority of soluble nitrogen and phosphorus forms, suggest urban trees have greater capacity to control belowground nutrient leaching relative to turfgrass. This may be particularly true for soluble phosphorus forms (Figure 3-11). Total dissolved and soluble reactive phosphorus concentrations and loads tended to be less in urban trees than turf, but the small size of the P leaching dataset limited statistical analysis. Additional examination of P leaching dynamics under urban trees and their turfgrass counterparts is warranted to better understand the potential significance of urban trees in limiting delivery of highly bioavailable forms of P to urban waters.

Figure 3-8. Box Plots Depicting Spread of Reported Total Dissolved Nitrogen (TDN) and Nitrate (NO3) Concentrations and Loads Leached from Urban Tree and Turfgrass Systems.

Median values and number of data points n are displayed for all boxes. As denoted by the *, NO3 loads under trees were significantly less than under turf at the p < 0.1 level; all other differences were insignificant. Studies from

which data extracted cited in Table B-5.

Page 40: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 25

Figure 3-9. Box Plots Depicting Spread of Reported Total Dissolved Phosphorus (TDP) and Soluble Reactive P (SRP) Concentrations and Loads Leached from Urban Tree and Turfgrass Systems.

Median values and number of data points n are displayed for all boxes. As denoted by the *, TDP loads and concentrations under trees were significantly less than under turf at the p < 0.1 level; however, such comparisons

are limited by the small number of data points. Studies from which data extracted cited in Table B-5.

In addition to providing insight regarding the potential range of nutrient concentrations and loads potentially leached from urban tree systems relative to the likely alternative to tree cover (i.e., urban turf grass), the studies also raise questions regarding the relative water balance of these systems. Our expectation, as that of the authors of the studies comprising the belowground process dataset, was that leaching volume below urban trees would be less than that from urban turf due to the ability to draw water from a greater portion of the soil profile through a tree’s more extensive root system. However, both of the studies examining both trees and turf observed the opposite; higher leachate volumes were consistently collected below urban trees than reference turf systems (Groffman et al., 2009; Nidzgorski and Hobbie, 2016). Across the entire belowground process dataset, reported leachate volumes were not correlated with precipitation depth, which was also counter to expectation. Reasons could include some combination of measurement error, greater infiltration under tree canopy and thus greater water availability for leaching (e.g., Rahman et al., 2019) and/or differences in soil structure and other physical subsurface controls. Thus, in addition to collecting more leaching data in urban tree systems, a better understanding subsurface hydraulic controls on leaching volume is needed to better predict nutrient mass loads.

As an additional indication of belowground processes, influent and effluent concentrations and loads for bioretention systems cataloged in the International Stormwater BMP database (BMP Database, n.d.) were examined. Systems were grouped into one of three general categories according to dominant vegetation type reported in the database: trees and shrubs (9 datasets), turfgrass (8 datasets), or native grass and forbes (6 datasets). While total nitrogen and phosphorus concentrations leaving these systems tended to be less than influent concentrations, the difference was not statistically significant across any of the vegetation types (Figure 3-10). However, when system performance was considered on an influent/effluent load basis, total nitrogen and phosphorus loads exiting systems in the tree and shrub vegetation category were significantly different than influent loads. Significant nitrogen load reductions were also observed for the turfgrass dataset, but not for total P. Influent and effluent nutrient loads were not significantly different across the native grass and forbes vegetation category.

Page 41: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

26 The Water Research Foundation

Figure 3-10. Box Plots Depicting Influent and Effluent Total Nitrogen (TN) and Phosphorus (TP) Loads by Vegetation Type in Bioretention Systems.

Pairs of influent/effluent data in which significant load reductions were detected are denoted with ** (p < 0.05). Source: Data from BMP database, n.d.

As with all datasets, care must be taken in the interpretation. This analysis only considers differences in system vegetation type but not other design characteristics known to influence water quality performance in bioretention systems such as media composition, hydraulic residence time, system age, or plant/root density. This analysis does indicate the need to consider nutrient loads alongside concentrations when comparing system performance across vegetation type and/or other design parameters. For instance, in a meta-analysis of phosphorus removal by green stormwater infrastructure practices on a concentration basis, Schechter et al., (2013) also found that vegetation type and density did not have a significant effect on effluent P concentrations. Due to the limited number of sites represented in this dataset, care should be taken in generalizing results to other bioretention systems on the basis of vegetation type alone until additional analysis of other design features is undertaken. Still, it is interesting to note that, similar to observed trends in TP leaching rates below urban turfgrass and trees depicted in Figure 3-9, bioretention systems with woody vegetation tended to be more effective in TP removal than turfgrass bioretention systems.

3.3 Putting it Together – Application of Meta-analysis Results to Support Stormwater Volume and Quality Impacts of Urban Tree Systems The landscape context of a given urban tree system will influence its role as a source versus sink of nutrients to urban stormwater. For example, litter decomposition and nutrient leaching/mineralization is more likely to contribute nutrients to stormwater pollutant load budget for trees overhanging impervious surfaces while these same nutrients may be retained within the tree root zone-soil system in when trees are located in urban open spaces. In the following section, trees in three different landscape contexts are considered to illustrate how the results of the meta-analyses presented in this chapter can be used to examine the role of these systems in runoff quantity and quality regulation (Table 3-8). While the water quality analyses presented herein lend more to a qualitative assessment, insights to assist

Page 42: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 27

stormwater managers in assessing water quality impacts of urban tree systems can still be gleaned.

Table 3-8. Example Questions Related to the Role of Urban Trees in Regulating Stormwater Runoff in Three Unique Landscape Contexts.

Tree system landscape context Example question Key processes Open space trees (e.g., upland forest,

riparian forest or other tree systems over permeable surfaces)

Will expanding tree canopy cover in urban open spaces influence urban water quality? Precipitation

partitioning, ET, throughfall nutrient

washoff, litter decomposition,

belowground cycling

Street trees Do street trees act as a net source or sink of nutrients to urban stormwater?

Trees in structural stormwater practices Do trees enhance pollutant removal processes in structural stormwater practices?

3.3.1 Will Expanding Canopy Cover in Urban Open Spaces Reduce Pollutant Loads Delivered to Urban Waters? The data presented herein suggest that yes, urban tree canopy in open spaces will reduce nutrient loads delivered downstream relative to open turfgrass, which is the most likely alternative to tree cover in urban parks, residential areas, and other open space areas. It is likely that water quality benefits associated with open space trees comes through regulation of pollutant loads leached to groundwater and stream baseflow, particularly with respect to soluble phosphorus forms. Furthermore, reported nutrient fluxes measured below tree root zones indicate that, if one considers nutrient inputs from atmospheric deposition/throughfall and litter decomposition, open space trees likely act as a net nutrient sink. It should be noted that these observations are based on a relatively small data sets, particularly for P leaching rates.

3.3.2 Do Street Trees Act as Net Sources or Sinks of Nutrients to Urban Stormwater? In the context of street trees, opportunities to infiltrate and process nutrients delivered via throughfall and decaying leaf litter within the root zone are limited relative to trees growing over pervious surfaces. Nutrient mass loss rates compiled herein, as well as the small number of watershed-scale studies in which tree litter contributions to runoff nutrient loads have been estimated, indicate that trees overhanging impervious surfaces are likely to act as net sources of nutrients to runoff. These data also indicate the magnitude of nutrient losses from decaying litter to stormwater runoff is likely to vary from one tree species to another depending on the quantity and quality of litterfall. Figure 3-11 illustrates the potential range in nutrient loss rates associated with evergreen and deciduous tree canopies and takes into account differences in litter quantity and quality produced by these tree systems. (Here, litter quality describes how easily nutrients can be leached or degraded and release by the microbial communities that colonize tree litter.) Coniferous evergreen species tend to produce relatively small quantities of low quality litter relative to deciduous trees, and, as result, are likely to release a relatively small nutrient mass compared to deciduous trees. The mass loss rates presented in Figure 3-11 are scaled to represent litter from a 100 m2 canopy area, which is approximately representative of the canopy area for a single, mature deciduous tree (Nidzgorski and Hobbie, 2016). These mass loss rates reflect the potential stock of dissolved nutrients released from the litter over time; actual delivery of released nutrients to runoff would depend on the timing and intensity of rainfall during the litterfall period. Still, this data summary is intended to help stormwater managers put into context the potential nutrient load associated with tree litter and time street sweeping or other litter collection efforts accordingly.

Finally, while street trees may have a water quality benefit by reducing potential runoff volumes from underlying impervious surfaces via rainfall capture processes (Section 3.2.1), the effect is probably

Page 43: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

28 The Water Research Foundation

minimal in light of data indicating nutrient concentrations delivered by trees via throughfall are similar to those from impervious surfaces (Section 3.2.3.1). Thus, the data strongly suggest that, without an effective litter collection program, street trees or other tree canopy overhanging (directly connected) impervious surfaces is very likely to act as a net source of nutrients to downstream waters.

Figure 3-11. Example Nutrient Mass Loss Rates for Hypothetical Coniferous and Deciduous Street Trees. Decay rates based on initial P content as described in Section 3.2.3.2 to be representative of low P content conifer litter (initial P = 0.4 g/kg) and deciduous litter with moderate (0.75 g P/kg) and high (1.5 g P/kg) initial P contents.

Other assumptions: 100 m2 deciduous canopy generates 10 kg litter mass while evergreen generates 5 kg (Nadelhoffer et al., 1983; Sun and Zhao, 2016); Initial N content of freshly fallen litter taken as median values

reported across studies analyzed in Section 3.2.3.2 for urban deciduous (11.2 g N per kg litter) and coniferous (4.2 g N per kg litter) litter.

3.3.3 Do Trees Enhance Nutrient Removal Processes in Structural Green Stormwater Infrastructure Practices? The analysis of bioretention water quality datasets in the International Stormwater BMP database (BMP Database, n.d.) suggests that the answer to this question is also “Yes.” While bioretention system planted with trees tended to have lower nutrient concentrations in drainage waters leaving the cell compared to influence concentrations, the real difference was observed from a nutrient mass load standpoint. This indicates the importance of volume reduction processes (e.g., seepage to underlying soils, evapotranspiration) in removing nutrients from stormwater runoff. It should be emphasized that this analysis was not designed to fully attribute observed changes in water quality to the type of vegetation system present as the influence of other important controls (e.g., media composition and nutrient content, hydraulic residence time, underlying soil types) were not considered. Still, as demonstrated in column studies (e.g., Denman et al., 2016), the role of vegetation in enhancing nutrient retention processes cannot be understated. Although this analysis does not unequivocally indicate that trees are more effective than other vegetation types in green stormwater infrastructure systems, it is interesting that trends in phosphorus mass reductions from systems with trees versus turfgrass were similar to those reported across the collection of landscape leaching studies compiled in Section 3.2.3.1 (trees were more effective in retaining P in both). Thus, additional data collection and/or analyses are warranted to better characterize how trees augment pollutant removal processes in GSI systems.

Page 44: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 29

CHAPTER 4

Scaling Hydrologic Benefits to the Watershed: i-Tree Hydro Modeling As reflected by the literature review and meta-analysis, nearly all of the data characterizing the hydrologic effects of urban tree systems have been collected at spatial scales ranging from individual trees to residential yards. In contrast, stormwater managers are generally interested in hydrologic outcomes at scales ranging from development sites up to the watershed. It is very challenging to experimentally test the effect of tree canopy cover on urban hydrology at these larger spatial scales due to the difficulty of designing (and funding) studies with enough replication or control over the range of biophysical and social factors that influence watershed-scale hydrologic responses and, to our knowledge, no such study has been conducted. Hydrologic models, on the other hand, can provide a useful tool for examining how a change in the watershed, such as increasing canopy cover, may influence runoff responses provided they are properly calibrated. In this chapter, the results of watershed modeling activities aimed to demonstrate the capacity for urban tree systems to mitigate runoff volume in urban watersheds are presented. The primary objective of these activities was to bridge the gap between the scale at which urban tree system hydrologic effects have been measured (tree to residential yard-scale) and the scale at which hydrologic effects are desired (watershed scale) to provide insight to the question: are the hydrologic benefits of urban tree systems scalable? Model selection, creation and results are presented in the following sections.

4.1 Hydrologic Modeling Approach There are a variety of hydrologic modeling tools that have been developed and applied in urban watersheds (e.g., source). Each of these tools comes with its own set of strengths and limitations. Since the objective of this study was to specifically assess the hydrologic effects of tree canopy cover, the i-Tree Hydro model was selected since it is one of few hydrologic models to specifically incorporate tree canopy characteristics and model their effects on ecohydrological processes such as interception and evapotranspiration (Wang et al., 2008). I-Tree Hydro also represents a balance between model complexity (as it incorporates mechanistic models for interception) and usability (as the model is parsimonized to only require the most influential variables, thereby reducing some of the burden associated with model creation) such that it could be a practical tool for stormwater managers and stormwater engineers to adopt. A distinct disadvantage of the tool is that it does not explicitly represent stormwater systems or drainage practices (e.g., stormwater sewers, constructed stormwater ponds) common in urban landscapes, though there is some ability to calibrate for these features using other model inputs.

4.1.1 Model Watershed Descriptions The i-Tree Hydro model was applied to three urbanized watersheds in the Kansas City, KS metropolitan area. A small forested watershed was also included as a reference (Table 4-1; Figure 4-1). This region is characterized by a continental climate and receives, on average, 995 mm of precipitation annually, which is distributed approximately normally over the calendar year. Study watersheds were selected to be relatively near in proximity to one another and within the same hydrophysiographic province with the intent of minimizing the effects of other hydrologically-important variables (e.g., watershed slope, soils and underlying geology) so that the effects of urban forest cover could be more clearly assessed. These watersheds were also selected to represent a range in development patterns (reflective of

Page 45: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

30 The Water Research Foundation

commercial, residential and mixed development) and tree cover (14% to 45%). Streamflow and precipitation data collected have been monitored at the outlet of each of these study watersheds through a joint venture between the county, cities and state department of transportation in this region, thus providing a valuable source of information needed to calibrate each study watershed. The study area was part of a natural resource inventory (MARC, 2014) through which a high resolution land use land cover dataset was developed (MARC, 2013), as well as an i-Tree Eco project in which the general structure and composition of urban forest in the region was documented (Nowak et al., 2013).

Table 4-1. Characteristics of Study Watersheds. Watersheds are labeled in order of increasing % forest cover; labels correspond to labeling in Figure 4-1. A small,

predominantly forested watershed was also modeled as a reference. Watershed characteristic Urb1 (residential) Urb2 (commercial) Urb3 (mix) Reference

Area (km2) 4.2 2.5 2.7 0.92 Mean % slope (min, max) across

watershed 2.9%

(0.2 to 9%) 4.6%

(0.1 to 13.2%) 4.1%

(0.1 to 18.5%) 5.8%

(1.5 to 13.6%) % forest cover 14% 21% 45% 74%

% herbaceous/turf 36% 12% 14% 24.5% % cultivated cropland 5% 0% 0% 0%

% total impervious area 45% 67% 41% 0.17% Soil texture Silt loam & silty clay

loam Silt loam & silty clay

loam Silt loam & silty

clay loam Silt loam & silty clay

loam

Figure 4-1. Study Watershed Map. All sites were located in the Kansas City, Kansas metropolitan area. Study site names correspond to sites described

in Table 4-1.

4.1.2 Model Creation, Calibration, and Performance Assessment Numerous input parameters are required to create an i-Tree Hydro model. These input parameters, and the sources from which they were obtained, are presented in Table 4-2. As noted in the table, the values

Page 46: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 31

of a subset of the input parameters were determined through the calibration process. These so-called calibration parameters generally characterize some aspect of the watershed that is difficult to measure directly (e.g., unsaturated zone time delay) and so are determined by adjusting values to match modeled and measured hydrologic responses. The land cover dataset upon which model inputs for % impervious and vegetation cover were based was developed from 2011 imagery; therefore, streamflow and precipitation data from 2011 to 2013 were used to calibrate and assess model performance. 2012 was a drought year and, combined with several periods in which errors were identified in streamflow and precipitation sensor data, was not considered in model creation. The majority (87.5%) of rainfall events during the calibration and assessment periods were 25 mm (1 inch) or less, but ranged up to 110 mm (4.3 inches). This range in precipitation enabled model calibration and assessment across a range of streamflow conditions. Visual inspection of observed and modeled hydrographs were supplemented with the Nash-Sutcliffe Efficiency (NSE) ratio to characterize how well modeled runoff hydrographs matched observed streamflow data on both an event and continuous basis. Model calibration was conducted using i-Tree Hydro’s built in model calibration tool. Additional manual adjustment of calibration parameters was conducted as needed to ensure values fell within appropriate ranges and to obtain NSE ratios of at least 0.5, which is generally deemed acceptable for hydrologic studies of this nature (Engel et al., 2007). Despite instances in which the fit between modeled and observed streamflow was poor for individual events, overall average NSEs for model watersheds ranged from 0.52 to 0.66 for the 2013 calibration period and from 0.48 to 0.71 for the 2011 model assessment period. Additional information regarding the calibration and model assessment process, final parameter values used to characterize each of the study watersheds, and hydrographs comparing observed and modeled streamflow data from which NSE metrics were calculated, are presented in Appendix C.

Table 4-2. i-Tree Hydro Model Inputs and Corresponding Sources of Data. Model inputs Data sources

Model area characteristics Digital elevation model National elevation dataset Watershed area Arc-hydro watershed delineation

Simulation period Selected to correspond to desired calibration/validation dataset or scenario run

Observed streamflow data (used in calibration) Johnson County Stormwatch network Meteorological data (used in calibration) Landcover parameters

% tree, evergreen tree, shrub, herbaceous, water, bare soil and impervious cover MARC1 LULC1 natural resource inventory

Leaf area index i-Tree Eco tree survey data (Nowak et al., 2013) Shrub and herbaceous leaf area index Literature values with field verification % impervious area (total and directly connected) Literature values with field verification

Pervious/impervious cover beneath tree canopy MARC1 LULC1 natural resource inventory with field verification

Hydrologic parameters Soil characteristics: texture, infiltration parameters (wetting front suction, wetted moisture content, surface hydraulic conductivity), initial moisture content

STATSGO soil database, field infiltration tests; additional adjustment of infiltration parameters through calibration

Root zone depth Literature values Leaf phenology and storage: leaf on, off and transition periods Literature values Calibration parameters: Tree and shrub bark area index; leaf storage; pervious and impervious depression storage; transmissivity at saturation; unsaturated zone time delay; soil macropore %; subsurface and surface flow constants; % watershed controlled by infiltration excess versus saturation overflow runoff processes

Model default values used as initial starting point; final values determined through auto-calibration with comparison to available literature values

1Abbreviations: MARC is the Mid America Regional Council; LULC is Land use land cover

Page 47: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

32 The Water Research Foundation

4.1.3 Tree Canopy Cover Scenarios To test the influence of tree canopy cover in these study watersheds, a series of canopy cover change scenarios was developed (Table 4-3). These scenarios were intended to represent different planning or policy mechanisms through which urban tree canopy could be expanded in urban areas. The first scenario represented programs in which the outcome is an increase in tree canopy cover over pervious surfaces. For example, efforts to expand tree planting in parks, residential yards, or riparian areas could be considered under this scenario. The second set of scenarios was intended to represent programs or policies in which tree canopy expansion over impervious surfaces is targeted. For example, “green street” programs or policies requiring higher levels of canopy cover in parking lots or other impervious areas would fall under these scenarios. Two different sets of impervious canopy cover scenarios were developed. In the first, canopy cover over impervious surfaces was increased but the total impervious surface area remained unchanged. In the second, every 10% increase in tree canopy over impervious surfaces was accompanied by a 1% decrease in total impervious surface area. This scenario was intended to represent situations in which pervious area at the base of the tree (e.g., a tree pit) replaced part of the existing impervious area, while the former (i.e., no change in total impervious area) could represent planting programs in which trees are planted in existing pervious medians or rights-of-way. Model inputs for all scenarios are provided in Appendix C. Across all three scenarios, existing landcover conditions in the calibrated study watershed models were adjusted to represent a 25%, 50%, 75% or 100% increase in pervious or impervious canopy cover. Stormwater managers are generally concerned with treating runoff quality and/or volume associated with small, frequent rainfall events and flood mitigation/safe passage of larger, less frequent events. Thus, changes in watershed runoff response were modeled for the so-called water quality event (which is generally defined as the rainfall depth greater than 85% to 90% of annual rainfall events) and 10-year, 24-hour event. In the greater Kansas City metropolitan area, the water quality event is equivalent to 25 mm (1 inch) of precipitation while the 10-year, 24-hour event corresponds to 140 mm (5.5 inches) of rainfall. Each of these storm depths were distributed in time using the SCS Type II rainfall distribution over a 24-hour period (NRCS, 1986).

Table 4-3. Description of Watershed Hydrologic Modeling Scenarios to Test Urban Forest System Scalability. Baseline conditions Description1 Levels Water quality (25 mm), and 10 year-24 hr (140 mm) design storm events

Establish hydrologic response to water quality event and larger, infrequent storm (e.g., 10-year 24-hour) across the gradient in impervious surface cover and tree canopy cover represented in reference and developed study watersheds

Canopy cover change scenarios1 1. Increase % canopy cover over pervious surfaces

Represents replacing open turf grass by increasing tree canopy cover in residential yards, parks, open spaces, and/or expanding extant upland or riparian forest areas. Represented in i-Tree Hydro by increasing % tree cover and % tree cover over pervious surfaces. Total impervious surface area remains unchanged.

25%, 50%, 100% over existing conditions

2. Increase % canopy cover over impervious surfaces

Represents increasing canopy cover over streets, parking lots or other impervious surfaces by planting trees in existing pervious areas (e.g., right of way) such that canopy overhangs impervious surface. Represented in i-Tree Hydro by increasing % tree cover and % of tree cover over impervious surfaces. Total impervious surface area remains unchanged.

25%, 50%, 100% over existing conditions

3. Increase % canopy cover and tree planting area over impervious surfaces

Represents increasing canopy cover over streets, parking lots or other impervious surfaces by planting trees such that canopy overhangs impervious surface while creating new or expanding pervious planting area at the base of trees. Represented in i-Tree Hydro by increasing % tree cover and % of tree cover over impervious surfaces. Total impervious surface area decreases.

25%, 50%, 100% over existing conditions

1Model inputs for baseline and all canopy cover change scenarios are detailed in Appendix C.

Page 48: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 33

4.2 Hydrologic Model Results and Implications The following sections describe differences in hydrologic response of the study watersheds as characterized by runoff volume and peak flow, and the potential contribution of urban trees systems in these watersheds to regulate this response. This section concludes with a comparison of relative runoff volume reductions for the water quality and 10-year storm predicted by the meta-analysis regression equations presented in Section 3.2 with predictions from calibrated i-Tree models for each of the study watersheds.

4.2.1 Effect of Development Pattern: Comparisons among Study Watersheds and Forested Reference To provide a sense of how hydrologic responses in the three study watersheds differ from the native vegetated conditions (i.e., native grassland and gallery forests), runoff volumes and peak flows are presented for each of the study watersheds and the reference watershed (Figure 4-2). Since model watersheds were all different sizes, hydrologic metrics were scaled to a depth basis to enable direct comparison. Runoff volumes and predicted peak flows in the developed watersheds exceeded that predicted for the reference watershed for both the water quality storm (25.4 mm) by about an order of magnitude and by about three times for the 10-year storm (140 mm) runoff volume. In the case of the smaller water quality event, runoff volume and peak flow predicted for the urban watersheds generally increased with increasing impervious area.

Figure 4-2. Runoff Depth and Peak Flow for Reference and Urbanized Study Watersheds.

The following sections present tree canopy cover scenarios for each of the study watersheds. While return to a “natural” hydrologic condition is not expected, these scenarios provide insight to the ability to shift urban hydrologic responses to a more natural condition just through increasing the number of trees and associated canopy cover.

4.2.2 Increasing Canopy Cover and the Effects of Landscape Context Predicted changes in runoff volume and peak flow for the water quality event suggest that increasing canopy cover over impervious surfaces provides greater hydrologic benefits that increasing canopy cover within open space or other surfaces that are already pervious. As illustrated in Figure 4-3, converting up to 100% of the herbaceous land cover in each of the study watersheds to tree canopy only resulted in a 0% to 6% reduction in runoff volume and peak flow relative to existing conditions for the water quality event (25 mm rainfall). This result may reflect a limitation of the i-Tree model – in this model, soil infiltration parameters are held constant over the entire watershed area, so there is no difference in infiltration below tree canopy or open turf grass in the model. Thus, it is likely that the

Page 49: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

34 The Water Research Foundation

small difference that arises is associated with the additional surface storage provided by tree canopy. In contrast, runoff volume and peak flow reductions ranged from 2% to 10% across all watersheds for a 25% increase in canopy cover over impervious surfaces, while model predictions for a 50% increase in impervious canopy cover ranged from approximately 10% to 30%. As expected, the hydrologic effect was more pronounced when the increase in tree canopy cover was accompanied by a decrease in total impervious area as would occur if impervious canopy cover was being achieved by creating new and/or larger tree pits in existing impervious surfaces. While increasing tree cover over pervious areas is not the most strategic canopy expansion approach from a stormwater hydrology perspective, there may be water quality-related benefits such as delivering and retaining more atmospherically-derived pollutants to the root zone and reducing phosphorus leaching to groundwater and/or stream baseflow (Section 3.2.3). Regardless of the landscape context, it should be noted that the runoff responses still exceeded that of the reference watershed by an order of magnitude even for under the maximum canopy expansion scenarios. Thus, additional green stormwater infrastructure would be required to fully restore watershed hydrologic conditions to that of a more natural, vegetated state.

In the case of large, infrequent storms often associated with flooding in urban watersheds (here, the 10-year, 24-hour event), the effect of expanding urban canopy cover was negligible for all levels of canopy expansion over pervious or impervious surfaces (Figure 4-4). However, reductions in impervious surface cover associated with the tree pit scenario resulted in modest reductions in runoff volume, particularly in the most impervious of the three study watersheds (Urb2). In all canopy expansion scenarios, runoff volumes remained 3 to 4 times higher than observed in the reference forested watershed. In addition, peak flow did not change appreciably across any of the study watersheds for any canopy expansion scenario. This indicates the limitations of tree canopy alone for mitigating large events, but also underscores the potential to regulate stormwater volumes by increasing not just tree canopy but the pervious area around the tree base in which infiltration may occur.

Page 50: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 35

Figure 4-3. Change in 25-mm Event Runoff Volume and Peak Flow with Increasing % Canopy Cover.

Scenarios include increasing canopy cover over pervious surfaces only (green triangles), over impervious surfaces only (closed gray circles), and over impervious surfaces with a commensurate reduction in impervious surface to represent creating a pervious tree pit at the base of each tree (estimated as 10% of total canopy area; open black

circles).

Page 51: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

36 The Water Research Foundation

Figure 4-4. Change in 140-mm Event (10-yr 24-hr) Runoff Volume and Peak Flow with Increasing % Canopy Cover.

Scenarios include increasing canopy cover over pervious surfaces only (green triangles), over impervious surfaces only (closed gray circles), and over impervious surfaces with a commensurate reduction in impervious surface to represent creating a pervious tree pit at the base of each tree (estimated as 10% of total canopy area; open black

circles).

4.2.3 Comparison of Meta-analysis Models: Scaling from Tree to Yard to Watershed In the previous chapter, predictive equations developed through meta-analysis of existing precipitation partitioning studies were proposed (Table 3-4). Here, these simple regression equations are compared with watershed modeling results to demonstrate the potential to extrapolate these equations to larger spatial scales than the tree-scale at which the data underpinning these models was collected. Because meta-regression equations only represent canopy processes and not rainfall-runoff processes, the comparison was limited to the watershed scenario in which canopy cover was expanded over impervious surfaces (Scenario 2 in Table 4-3). While reductions in throughfall over impervious surface do not scale directly to the depth of runoff generated from the impervious surface, the relative change between the two metrics should be similar given the relatively linear change in runoff depth from impervious surfaces after the initial surface storage depth has been exceeded (e.g., Schuler, 1987). For this comparison, the volume of rainfall capture associated with the water quality event (25 mm) was calculated using the event-scale capture equation for deciduous trees (which dominate canopy cover in the study watersheds) for existing conditions and each of the canopy cover expansion scenarios (25%, 50%, 75% and 100% impervious canopy coverage). The % change in incident precipitation delivered to

Page 52: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 37

underlying impervious surfaces as predicted by the meta-analysis rainfall capture equation was then compared to % change in runoff depth as predicted in i-Tree Hydro for the canopy cover over impervious surface scenarios. Again, this comparison is not intended to imply that the meta-analysis equations are appropriate for predicting runoff depth or other hydrologic responses following delivery bulk rainfall or throughfall to impervious surfaces and is best limited to considerations of canopy expansion over impervious surfaces from which runoff responses are less complex. Rather, the intent here is to simply compare the results of the simplified regression equations to that of calibrated hydrologic models.

Figure 4-5 presents a comparison of changes in rainfall delivery to impervious surfaces underlying tree canopy as predicted by the event-based deciduous tree rainfall capture equation from the meta-analysis (Chapter 3; Table 3-4) and % change in runoff depth for the same impervious canopy expansion scenarios modeled in i-Tree Hydro. Despite differences in equations applied to predict rainfall capture (i-Tree Hydro uses more mechanistic Rutter and Gash models of interception processes; Wang et al., 2008) and differences in canopy level versus ground level hydrologic processes, the meta-analysis rainfall capture equations provide a reasonable indication of watershed-scale runoff response for two of the three study watersheds, but underestimated runoff reductions in Urb3. As mentioned in the preceding section, the relatively large change in predicted runoff depth from the Urb3 watershed is not fully understood, and based on the comparison here, likely originates from calibration parameters controlling surface and subsurface flow responses as opposed to canopy processes.

Figure 4-5. Comparison of Meta-analysis Equations for Event-Based Rainfall Capture (light blue bar) with Changes in Runoff Depth Predicted in i-Tree Hydro (dark blue bar) for Study Watersheds.

This comparison indicates meta-analysis equations may not scale linearly to the watershed scale, potentially due to effects of variables such as spatial orientation and hydraulic connectivity of impervious surfaces.

Page 53: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD
Page 54: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 39

CHAPTER 5

Counting the Costs and Benefits: Urban Tree System Cost and Co-benefit Database Urban trees and forests are known to provide a variety of benefits, such as air quality regulation, reducing building heating and cooling costs and contributing to the overall health and well-being of city inhabitants. Benefits like these have driven efforts to preserve and/or expand urban tree canopy in many cities and have been described in recent reviews such as by Song et al., (2018). However, establishing and maintaining trees entails additional economic costs for municipalities. Therefore, it is important that utilities are able to evaluate the costs and benefits of urban trees and forests to justify continued investment in urban tree systems. In response to this need, an urban tree cost-benefit database has been assembled to provide a single, comprehensive resource for utilities, or others interested in assessing costs, benefits, and/or return on investment associated with urban tree systems. This database was developed to fulfill Objective 2 of this project: to provide a basis for comparing the costs and overall value of urban tree systems with other stormwater infrastructure.

The primary goals of this database were to:

• Provide a comprehensive compilation of datasets in which costs and/or benefits of urban trees have been economically valued

• Produce a searchable tool with which stormwater managers or others interested in assessing costs, benefits, and/or return on investment (ROI) associated with urban tree systems can obtain this information

The database tool produced through this project intended to enable evaluation of the project question: Are urban trees/forests an affordable and desirable component of stormwater programs and to the broader community? The answer to this question will vary from one community and/or municipal program to another. Without doubt, planting and maintaining healthy tree systems as part of a municipality’s green infrastructure come with tangible economic costs. The value of co-benefits provided by those trees – which include a suite of other economic, social and environmental services that are very real but whose value is not always accounted for in traditional economic cost-benefit analyses – is likely to depend on the values of a community and/or the types of environmental issues the community faces. The database tool described herein was intended to present existing cost and co-benefit values ascribed to urban tree systems in such a way that gives the user flexibility in considering the types of costs and benefits most relevant to their organization. The following sections provide an overview of the database tool, as well as the fact sheet and user guide produced to support its use.

5.1 Urban Tree System Cost-Benefit Database Tool Overview A Microsoft Access database containing records of urban tree cost and co-benefit studies was assembled. A notable advantage of building this database in Microsoft Access was the ability for users of the database to run queries to summarize data for specific sets of costs or co-benefits as they desire. However, as not all potential users of the database tool may have access to or experience using Microsoft Access, a companion database was also developed in Microsoft Excel, albeit with less functionality than the Access version.

The structure, contents and example applications of the database tool are more fully described in a fact

Page 55: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

40 The Water Research Foundation

sheet and user guide. Briefly, a literature review was conducted to identify published studies in which economic cost or co-benefits associated with urban tree systems were reported. For a study to be included in the database, the following selection criteria had to be met:

• Studies were peer-reviewed or, if not peer-reviewed, utilized methods that have been peer-reviewed to quantify tree costs and/or benefits

• Biophysical or social metrics used to quantify co-benefits provided by trees were economically valuated (i.e., ascribed monetary value)

• Costs and benefits presented in the paper were primary data; that is, the study authors did not use values determined from other studies. This allowed double counting costs or values from the same study.

These selection criteria resulted in a final pool of 39 studies. Of these, 12 reported economic costs of planting and maintaining urban trees (25 records total) and 34 reported co-benefits and associated economic values (182 records total). (Some studies report both costs and benefits; hence the total does not sum to 39). The majority of studies identified for inclusion in the database were conducted in the United States, through studies conducted in foreign countries (e.g., Australia, Canada, China, India and Portugal) were also included. Information regarding the location at which cost and/or co-benefit data were generated are included in the database so that users of the database tool can consider costs and values from similar cities, climatic zones, etc. Given differences in costs expected to maintain, for example, street trees relative to forested outlots or urban green spaces, cost and benefit values were grouped into subcategories describing the tree system type when specified within the study. Most of the studies included in the database focused on “street trees” or urban forests in general. Some studies specifically looking at “parks” and “residential” tree/forest systems were included as well. The categorization of tree system types is intended to give users of the database tool the option to consider costs, benefits, and/or ROI for a specific tree system if desired.

To enable comparison between cost and benefit studies, the database only includes those studies in which a monetary value was ascribed. While this criteria enables a more straightforward evaluation of the return on investment presented by urban trees, which was one of the goals of this objective, it excludes those benefits for which monetary values have not been ascribed. As indicated in Table 5-1, many of the social benefits that urban tree systems are known to provide were not ultimately included in the database due to a lack of studies in which these benefits were monetized. This is a weakness of utilizing economic value as a common unit to aggregate benefits and costs and reflects a broader challenge to ecosystem service-based decision making. All co-benefit values were adjusted to 2017 dollars using the U.S. Bureau of Labor Statistics Consumer Price Index (CPI) inflation calculator on an annual basis to enable comparison among values reported at different points in time.

Page 56: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 41

Table 5-1. General Co-benefit and Cost Categories Reported in the Literature for Urban Tree Systems. The database is limited to those categories for which costs and/or benefit values were monetized (indicated by *).

Environmental benefits Economic benefits Economic cost categories Control urban heat island effect Increase real estate value* Planting*

Carbon sequestration* Energy savings* Establishment* Air quality control* Increase business income Pruning*

Enhance biodiversity Social benefits Litter Management* Stormwater runoff control* Reduce stress Infrastructure repair*

Nutrient retention/water pollution control* Social cohesion Disease control* Encourage physical activity Liabilities* Reduce crime/increase public safety Administration* Increase public health/well-being Removal*

5.1.1 Cost-Benefit Database Fact Sheet The database fact sheet provides a brief (2 page) overview outlining the approach by which the database was assembled and summary statistics of compiled tree costs and benefits. The fact sheet is intended to orient potential users to the database and its capabilities but without providing an overwhelming amount of detail. As highlighted in the fact sheet, the co-benefit with the highest ascribed value from the studies included in the database was increased real estate value (median value of $55 per tree per year), followed by water quality ($7.91/tree/year), energy savings ($7.12/tree/year), stormwater control ($4.90/tree/year), air quality improvements ($2.43/tree/year) and carbon sequestration ($0.40/tree/year). The median total life cycle cost from the compiled studies was $51 per tree per year, with tree pruning (median cost $25/tree/year) and removal ($7.26/tree/year) the most costly expenditures.

5.1.2 Cost-Benefit Database User Guide A separate user guide was produced as an extension of the fact sheet to provide a more detailed overview of the database format and, specifically, to provide users with a series of example queries to demonstrate the types of questions that the compiled cost and benefit data can be used to examine. Examples outlined in the user guide include:

• What is the value of stormwater management benefits reported for street trees? • What is the average value of co-benefits reported per tree? • What is the return on investment for urban trees? • What are the life cycle costs of urban trees per unit volume of stormwater capture?

In addition to guiding users through example queries to examine these questions, the database contains the same queries that users could simply use directly and/or modify as desired to obtain cost and benefit values of interest. A separate user guide was not developed for the companion Microsoft Excel version of the database tool; however, the Excel database contains a “Read Me” tab in which a general overview of its contents and use, including a simple automated ROI calculator, intended to assist users in successfully using the tool.

5.2 Cost-Benefit Database Takeaways As noted above, different communities are likely to place different values on various co-benefits provided by trees. However, the following general conclusions can be drawn from the data assembled in the cost-benefit database:

Page 57: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

42 The Water Research Foundation

• To minimize the economic costs of urban trees, providing conditions to grow healthy, long-lived trees is important as planting, establishment and removal costs comprised over 25% of total life cycle tree costs.

• The cost per average volume of stormwater runoff avoided obtained from the studies included in the database was approximately $19/m3 runoff per year. This unit cost could be compared to other green stormwater infrastructure. Tools such as the stormwater best management practices Whole Life Cost Tool (NAS, 2014) or the US EPA Opti-Tool (EPA, 2018) can be used to obtain analogous estimates of the annual cost to capture runoff volume with a variety of traditional and green stormwater infrastructure practices.

• If stormwater management is the only benefit considered, then the return on investment is low (negative) as life cycle costs far exceed the benefits. However, if a suite of benefits is considered, then a positive return on investment is obtained, indicating trees are a good investment from the standpoint of the broader community.

Page 58: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 43

CHAPTER 6

Curricula Development to Facilitate Integration of Urban Forestry and Stormwater Management The preceding chapters have described datasets and associated tools developed to enable utilities and other stormwater professionals to better quantify and account for the contribution of urban tree systems as explicit components of an area’s green stormwater infrastructure and/or determine the economic costs and co-benefit values associated with expanding tree canopy cover. The goal of the work presented in this chapter was to develop curricula aimed to help integrate urban forestry with stormwater management programs. Curricula development was supported by two preliminary tasks: (1) a literature survey of existing forestry training materials and (2) a survey of municipal stormwater and forestry programs. As described in the following sections, this curricula was developed to equip utilities with the knowledge to answer the question: How can urban trees/forests be practically integrated into existing stormwater utility programs?

6.1 Survey of Existing Curricula and Needs To better understand the needs of the target stormwater management communities for training materials, a survey was developed by the project team and its advisory council. The survey was administered via Qualtrics and targeted primarily to stormwater Utility managers and municipal arborists through various email mailing lists and newsletter invitations. A total of 52 stormwater utility representatives and 9 municipal arborists completed the survey. While respondents represented 15 difference states and D.CS, nearly half came from one of three states (California, Florida and Texas). The majority of respondents (64%) represented municipalities of more than 100,000 people, while 17% and 19% worked in municipalities with populations ranging from 75,000 to 100,000 and less than 75,000, respectively. Survey questions and responses are summarized in Table 6-1.

Survey responses indicated that many municipalities do consider trees at some level within their stormwater management programs, and that a majority currently collaborate with a municipal arborist or other forestry expert for managing trees. A high proportion (85%) also indicated that they would be willing to participate in educational activities such as reading a short bulletin or watching a webinar to learn more about the influence of trees on stormwater quantity and quality.

In addition to this survey of practitioners, a review of forest management curriculum was completed to identify existing training materials communicating the hydrologic roles of urban trees that could be adapted for use by stormwater utilities and municipal arborists. Surprisingly, there were only two peer-peer educational materials that addressed the influence trees have on hydrology. One set of materials was from a discontinued “Master Watershed Steward” program based in Oregon (Oregon State University), with a heavy focus on hydrology for salmon. The other set of materials was from Pennsylvania (Penn State) for training “Master Forest Stewards.” There were several examples in the Pennsylvania lessons comparing stream hydrology and storm flow for forested versus developed watersheds. Neither curriculum focused on urban forest or stormwater issues.

Due to this lack of appropriate education materials to adapt, a new “Forestry Effects on Stormwater” module was developed in this project, and can be viewed in Appendix D. The module draws heavily on the Stormwater Management Benefits of Trees report prepared for the Vermont Department of Forest,

Page 59: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

44 The Water Research Foundation

Parks and Recreation (Moore et al., 2014), along with information from the Chesapeake Bay watershed manual series (Cappiella et al., 2005, 2006a, 2006b). Results of the data analyses conducted herein were also incorporated. In accord with survey responses reported in Table 6-1, the resulting educational bulletin was developed to take about 15 to 20 minutes to read. This bulletin is included in Appendix D.

Table 6-1. Stormwater Utility Survey Results.

Survey Question Response summary Q1 - Does your municipality account for the effect of existing wooded areas and other landscape trees when planning for stormwater quantity and quality issues?

61% “yes”

Q2 - Does your municipality actively promote the protection of existing trees or the establishment and management of new wooded areas or other landscape trees for their effect on stormwater quantity and quality issues?

63% “yes”

Q3 - Does your department incorporate trees into engineered stormwater structures? 60% “yes” Q4 - Does your municipality have regulations regarding trees in riparian (stream side) set-backs?

57% “yes”

Q5 - Does your municipality have regulations regarding trees in flood plains? 66% “yes” Q6 - Do staff in your department collaborate with the city forester or municipal arborist on tree management issues?

67% “yes”

Q7 - When planning a new stormwater project, do you consider the projected amount of tree canopy or wooded areas in the development watershed? Q7a - If yes, what methods or models do you utilize to estimate the influence of various land uses, including tree canopy or wooded areas?

44% “yes”

Q8 - If offered free of charge, how likely would you be to participate in an educational module (read a bulletin for 20 minutes, view a webinar for 1 hour) on the effect trees and wooded areas have on stormwater quantity and quality?

85% extremely or somewhat likely to participate.

Q9 - What locations of trees do you consider when assessing stormwater effects? Check all that apply.

All six types of trees were considered to have an effect on stormwater, with Riparian trees being chosen most often by 24%, and Natural wooded areas chosen by 21%.

Q10 - What values and benefits do you attribute to the location of trees above? Check all that apply.

Refer to Appendix D for these results.

Q11 - What regulations, policies or programs are actively used in your municipality or state to promote tree planting and retention of tree cover?

21 responses, with 19 indicating “Yes” there are regulations and policies.

Page 60: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 45

APPENDIX A

Literature Review – Urban Tree Systems and Stormwater Quantity and Quality Regulation: A State of the Science A.1 Introduction Growing interest in the contribution of urban tree and forest systems to stormwater quality and quantity management is evidenced by several recent reviews on the topic (Berland et al., 2017; Cappiella et al., 2016; Kuehler et al., 2017), convening of expert panels to determine stormwater regulatory crediting structures (e.g., Cappiella et al., 2016), and an increasing number of organizations promoting trees as part of the green stormwater infrastructure. Given recent activity around this topic, the purpose of this synthesis is to present a state-of-the-art understanding regarding the effectiveness of urban tree systems as a component of stormwater management programs. Here, the term “urban tree systems” is used broadly to encompass any of the myriad forms in which trees may occur in urban landscapes, including but not limited to naturally established riparian and upland forest tracts to individual trees planted in residential lawns or streetscapes. Soils supporting tree systems may range from native mineral soils, disturbed urban soils, or engineered structural soils imported to the site. Urban tree systems, whether defined at the individual or stand scales, may influence runoff hydrology and quality through a variety of mechanisms. As defined in Table A-1, these mechanisms control the fate of precipitation falling directly on the tree canopy (as interception, throughfall or stemflow) as well as run-on delivered to the subcanopy area from adjacent urban surfaces, which may infiltrate into tree system soils or continue down gradient as surface runoff.

An understanding the effectiveness of urban trees for stormwater quantity and quality management requires an understanding of the magnitude to which these various hydrologic and water quality processes occur within the urban tree canopy. To this end, recent literature review efforts (Berland et al., 2017; Cappiella et al., 2016; Kuehler et al., 2017) have compiled much of the peer-reviewed literature in which the various fates of precipitation over tree canopy are quantified. Therefore, this review aims to synthesize the recent review efforts and relevant studies published in their wake for the purpose of (1) highlighting common areas of understanding, (2) identifying core areas of research need and (3) laying the foundation for a quantitative meta-analysis that will add to the narratives set forth in these existing reviews. Following an overview of the set of recent reviews of stormwater regulation by urban tree systems (Section A.2), this review is organized along the primary mechanisms controlling the influence of urban trees on stormwater runoff hydrology and water quality. The implications of this knowledge base to stormwater management are then discussed along with a summary of existing knowledge gaps (Section A.3). Throughout this review, the results of field data are emphasized, though a limited discussion of modeling is presented with efforts to quantify the effects of urban tree systems on stormwater hydrology at the watershed scale.

Page 61: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

46 The Water Research Foundation

Table A-1. Definitions of Hydrologic Processes and Mechanisms Associated with Urban Tree Systems. Term Definition

Gross precipitation Precipitation measured above the tree canopy or in an open area. Fate: Interception, throughfall, stemflow, runoff or infiltration

Interception (Section A.2.1)

Rainfall abstracted through storage on and evaporation from canopy surfaces (i.e., foliage and/or stems) that does not reach the ground; equal to the difference between total precipitation and the sum of throughfall and stemflow. Fate: evaporation. Relevant time scale: during and immediately after precipitation event.

Throughfall (Section A.2.1)

Rainfall that falls through leaf canopy to ground either by falling freely through or dripping from the canopy. Fate: Infiltration or surface runoff. Relevant time scale: during and immediately after precipitation event.

Stemflow (Section A.2.1)

Portion of rainfall that reaches ground by flowing down tree stems and/or tree trunk (or bole), initiated after canopy surface storage is filled. Fate: Infiltration or surface runoff. Relevant time scale: during and immediately after precipitation event.

Evapotranspiration (Section A.2.2)

Process by which soil water is removed from the soil profile through a combination of evaporation from the soil surface and transpiration through leaf stomata. Relevant time scales: days to weeks in between precipitation events.

Surface run-on (Section A.2.2)

Stormwater runoff generated outside the tree canopy that is transported to soils underlying the canopy as result of topographic position. Fate: infiltration or surface runoff. Relevant time scale: during and immediately after precipitation event.

Infiltration (Section A.2.2)

Process by which water moves into the soil profile. Fate: interflow, groundwater recharge, evapotranspiration. Relevant time scale: during and immediately after precipitation event.

Surface runoff (Section A.2.2)

May occur as result of infiltration excess overland flow (Hortonian flow, in which precipitation rate exceeds soil infiltration rate) or saturation excess overland flow (in which runoff is generated due to soils that have saturated to the surface). Relevant time scale: during and hours after precipitation event.

A.2 Overview of Recent Syntheses Given growing interest in quantifying the role of urban tree systems in regulating stormwater hydrology and quality, three literature reviews have recently been completed to compile available data pertaining to influences of trees on the urban hydrologic cycle and runoff quality. Two of these reviews were published in the peer-reviewed literature (Berland et al., 2016; Kuehler et al., 2017) while a third review by Cappiella et al., (2016) was completed as part of an expert panel to the Chesapeake Bay Program. Among the studies reviewed in these publications, 95 presented measurements associated with tree or forest hydrology and its implications at scales ranging from individual trees (79 studies) to watersheds (16 studies). Seventeen of the studies included in these reviews addressed some aspect of water quality associated with tree systems. In addition to hydrologic and water quality studies, Cappiella et al., (2016) reviewed 48 studies addressing tree growth and survival rates in urban settings as is relevant to projecting changes in stormwater regulation provided by trees through time. Table A-2 summarizes selected studies in which precipitation partitioning data were collected from urban or open-grown trees.

Table A-2. Summary of Studies in Which Hydrologic and/or Water Quality Influences of Urban Trees Were Experimentally Quantified.

Source Location Tree system type/scale Mechanisms addresseda

Measurement method

Li et al., (2008) Fort Pierce, FL urban I, THF, SF (n = 6 events, 3 cultivars x 3 reps; n = 54 total)

Field: I calculated from measured P, THFb, and SF

Xiao and McPherson (2011b)

San Francisco, CA

Urban/individual tree

I, THF, SF (n = 75)

Field: I calculated from measured P, THFb, and SF

Xiao and McPherson (2016)

San Francisco, CA (laboratory experiment)

Urban/ individual tree branch

I (n = 160)

Lab: Simulated rainfall, change in weight of harvested branch

Page 62: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 47

Livesley et al., (2014)

Melbourne, Australia

Urban/individual tree

I, THF, SF (n = 55)

Field: I calculated from measured P, THFb, and SF

Staelens et al., (2008)

Ghent, Belgium

Forest/ Isolated tree (12 m x 15 m)

I, THF, SF (n = 205 events; 1 tree; n = 205 total)

Field: I calculated from measured P, THFb, and SF

Guevara-Escobar et al., (2007)

Queretaro City, Mex.

Urban/individual tree

I, THF, SF (n = 19 events; 1 tree; n = 19 total)

Field: I calculated from measured P, THFc and SF

Xiao et al., 2000 Davis, CA

Urban/individual tree

I, THF, SF (n = 134)

Field: I calculated from measured P, THFc and SF

Asadian (2010)

Vancouver, BC

Urban/individual tree

I, THF (n = 318 events, 54 trees; n = 7042 total)

Field: I calculated from measured P and THFb, assumed no SF

Asadian and Weiler (2009)

Vancouver, BC

Urban/individual tree

I, THF (n = 7 events, 6 trees; n = 42 total)

Field: I calculated from measured P and THFb, assumed no SF

Inkilainen et al., (2013)

Raleigh, NC Urban, yard-scale

I, THF (n = 320)

Field: I calculated from measured P and THFb, and assumed SF

Park and Cameron, 2008

Soberania National Park (SNP), Republic of Panama

Plantation, tree scale (6 x 6 m spacing)

I, THF, ST (n = 14 events, 5 species x 3 replicates n = 461 total)

Field: I calculated from measured P, THFb, and SF

Bu et al., 2016 Taihu Lake watershed, southeastern Yixing City, Jiangsu Province, China

Plot: Tree-grass buffer (16 m x 13 m plots)

I, THF, ST (n = 32 events, 3 planting densities x 3 replicates; n = 210 total)

Field: I calculated from measured P, THFb, and SF

Van Stan et al., 2015

NE Maryland Individual tree scale

I, THF, SF (n = 52 storm events, 2 tree species x 3 replicates; n = 312)

Field: I calculated from measured P, THFb, and SF

Schooling and Carlyle-Moses, 2015

City of Kamloops, British Columbia, Canada

Isolated deciduous park trees

SF (N = 37 trees; 86 events)

Field: measured SF

Kermavnar and Vilhar, 2017

Ljubljana, Slovenia Urban forest stands (2,500 m2)

I, THF, SF Field: I calculated from measured P and THFb; SF estimated from literature

Sadeghi et al., 2016 Tehran, Iran Afforestation plots (open understory)/ stand scale (300 m2)

I, THF, SF (n = 165 events, 4 species, n = 660 total)

Field: I calculated from measured P, THFb, and SF

ahydrologic mechanisms abbreviated as Int (interception), ET (evapotranspiration), T (throughfall), S (Stemflow), Inf (Infiltration), RO (runon). Water quality mechanisms abbreviated as F (filtration), U (uptake), L (litterfall). bThroughfall measured via point measurements in which grid of gauges positioned below canopy cThroughfall measured via area method in which collection surface positioned below canopy to obtain integrated throughfall measurement

Relative to urban tree systems, there is a relatively rich body of literature pertaining to non-urban forest hydrology, the likes of which have also been reviewed (e.g., Crockford and Richardson, 2000; Barbier et al., 2009). However, as reviewed by Kuehler et al., (2017), structural differences in the canopies of

Page 63: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

48 The Water Research Foundation

natural forests and open grown trees typical of urban areas are expected to give rise to differences in hydrologic and water quality processes. Therefore, this and other reviews of urban tree hydrology have focused on those studies conducted in urban contexts. However, where appropriate, we have included non-urban studies to, for example, provide additional support for explanatory mechanisms that may promote or constrain stormwater regulation by urban trees.

The following sections review the current state of knowledge regarding the mechanisms by which urban trees may regulate stormwater hydrology and water quality. Hydrologic mechanisms are presented as those that occur within the tree canopy (Section A.2.1) and at the ground surface (Section A.2.2). The state of knowledge regarding the influence of urban tree systems on stormwater runoff quality is then presented (Section A.2.3), and considers the counter effects of trees as nutrient sinks (Section A.2.3.1) and sources (Section A.2.3.2). Efforts to extrapolate beyond the scale of individual trees, at which the majority of field data have been collect, to the watershed scale are next considered (Section A.2.4). This review concludes with a summary of the implications to stormwater management efforts, or, where appropriate, highlights existing gaps in understanding what those implications are.

A.2.1 Aboveground Processes: Interception, Throughfall, and Stemflow Precipitation falling on the tree canopy is partitioned among interception, throughfall and stemflow. Of these, interception serves to decrease the amount of precipitation reaching the ground surface through evaporation back to the atmosphere, while throughfall and stemflow alter the spatial and temporal distribution of precipitation reaching the ground. In stemflow, a portion of the precipitation falling on the canopy is directed along stems and branches and then funneled to the base of the tree along the trunk (or bole). Throughfall reaches the ground below the canopy by falling either directly through or by concentrating into large droplets on leaves and falling in localized streams to the ground surface (Park and Cameron, 2008). As indicated in Table A-2, these interrelated mechanisms are often studied concomitantly. Field studies to date suggest the depth of canopy interception increases with precipitation depth during an initial wetting period, then reaches a threshold saturation depth that remains somewhat constant even as precipitation depth increases (Xiao et al., 2000; Xiao and McPherson, 2016; Van Stan et al., 2012). In contrast, the depth of throughfall and stemflow tend to increase approximately linearly with gross precipitation depth after canopy saturation (Xiao et al., 2000; Asadian, 2010; Xiao and McPherson, 2011b; Livesley et al., 2014; Park and Cameron, 2008), though interactions with precipitation characteristics can cause nonlinearities in this relationship (Crockford and Richardson, 2000; Schooling and Carlyle-Moses, 2015). This linear behavior has been observed in studies of natural forests, interception in which has been estimated as a function of gross rainfall with a simple linear model by Gash (1979) that is still in use today (e.g., Gonzalez-Sosa et al., 2017). While measured as a depth, studies typically report interception, throughfall and stemflow as a percentage of annual gross precipitation (Cappiella et al., 2016). However, in comparing interception rates across studies conducted under differing climatic conditions, it is also instructive to consider canopy interception as a depth given the nonlinear relationship between interception and gross precipitation depth. Figure A-1 summarizes the results of studies included in and published since the recent reviews by Cappiella et al., (2016), Berland et al., (2017), and Kuehler et al., (2017) in which precipitation partitioning was measured in urban and open grown tree systems (see Table A-2 for more detailed listing of reviewed studies). Across these studies, canopy interception was reported to range from 9 to 60% of gross annual precipitation, while throughfall and stemflow ranged from 37 to 89% and 0.3 to 17%, respectively (Figure A-1). With the exception of a few outliers, observed rates of interception by urban tree systems generally fall within the range for natural forests (10 to 24% and 15 to 46% of annual precipitation in deciduous and evergreen stands, respectively, as reviewed by Cappiella et al., 2016).

Page 64: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 49

Figure A-1. Box Plot Depicting % of Gross Annual Precipitation as Partitioned among Interception (left),

Stemflow (middle) and Throughfall (right) Observed in Urban and/or Open Canopied Tree Systems. Citation information for studies from which data used to create boxplots were extracted is provided in Table A-2.

Figure A-2. Box Plots Depicting % of Gross Annual Precipitation as Partitioned among Interception (top),

Stemflow (middle) and Throughfall (bottom) below Urban Deciduous and Evergreen Tree Canopies. Citation information for studies from which data used to create boxplots were extracted is provided in Table A-2.

As illustrated by Figure A-1, reported values of rainfall partitioning by urban trees are highly variable, even when expressed as an annual average. Separating by evergreen and deciduous trees can help explain some of this variability (Figure A-2). Other sources of variability include climate, precipitation characteristics, canopy architecture, measurement error, and interactions among these factors (Crockford and Richardson, 2000). Reviews by Kuehler et al., (2017) and Berland et al., (2017) discuss some of the physical and biological factors thought to influence rainfall partitioning in tree canopies, and thus, the potential for trees to regulate stormwater hydrology. These factors are summarized in Table A-3, along with the direction of their effect as evidenced by the literature to date. While additional work is needed to identify which of these factors are of the greatest importance in determining whether precipitation falling on tree canopies reaches the ground (Berland et al., 2017), the existing literature provides a starting point from which to surmise the importance of these factors to stormwater management efforts. The following subsections review the quantitative evidence regarding the importance of tree- and atmospheric-based variables on stormwater regulating potential of urban tree systems. From this evidence base, we then provide some general estimates of the potential impact of

Page 65: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

50 The Water Research Foundation

these variables upon stormwater hydrology. It should be noted that these estimates are coarse by nature and are intended to serve as an example. Every tree system is different and, therefore, general conclusions about rainfall partitioning and subsequent impacts to stormwater runoff may not apply to every system, even within similar species mixes and climate regions (Kermavnar and Vilhar, 2017).

Table A-3. Factors That Influence Precipitation Partitioning into Canopy Interception (I), Throughfall (TH) and Stemflow (SF), as percentages of gross precipitation.

The “+” indicates positive correlations generally observed while “-“ indicate negative correlations and “+/-“ indicated correlation is mixed or unknown. Variables compiled from Berland et al., (2017) and Kuehler et al.,

(2017). Tree Variables I TH SF

Evergreenness + - +/-

Leaf area index + - +/-

Crown size/tree age + - +

Bark roughness + - -

Branch angle - + +

Leaf angle - + +/-

Leaf hydrophobicity - + +

Evapotranspiration rate + - -

Crown openness - + +

Atmospheric characteristics I TH SF

Precip intensity (drop size) +/- + +/-

Duration + - +/-

Antecedent + - -

Temperature - + -

evaporative demand + - -

Wind +/- +/- +

gross precipitation - + +

rain angle from horizontal +/- +/- -

vapor pressure deficit + - -

A.2.1.1 Tree-Based Controls on Canopy Partitioning in Urban Tree Systems Numerous studies have indicated evergreenness bears a particular influence on precipitation partitioning, with evergreens tending to intercept more water than deciduous trees (Xiao et al., 2000; Asadian, 2010; Xiao and McPherson, 2016; Barbier et al., 2009; Livesley et al., 2014; Sadeghi et al., 2016). Indeed, as indicated by Figure A-2, canopy interception rates reviewed herein for representative urban tree systems are on average 27% higher for evergreens than those reported for deciduous tree species (30.3 ± 5% versus 21.9 ± 3.4% of annual precipitation. Where the data presented in this set of studies allowed, the average depth of interception per event was computed by dividing the total annual interception depth by the number of precipitation events recorded. On an average annual event basis, interception rates among the set of studies reviewed equated to 3.5 ± 2.5 mm and 4.9 ± 2.6 mm per event for evergreen (n = 9) and deciduous trees (n = 9), respectively. Greater canopy interception depths computed across deciduous tree studies indicates these studies were subjected to higher average rainfall depths than the set of studies including evergreens. Still, with the exception of Kermavnar and Vilhar (2017), studies in which interception rates in evergreen and deciduous urban tree systems were directly compared under similar climatic conditions reported higher interception rates by evergreens on an average annual basis (Xiao et al., 2000; Asadian, 2010; Xiao and McPherson, 2011b; Sadeghi et al.,

Page 66: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 51

2016), mirroring observations in natural forest systems (e.g., as reviewed by Barbier et al., 2009 and Cappiella et al., 2016). It is generally agreed that higher interception rates by evergreens are driven by greater storage capacity of leaf surfaces, which evergreens retain year-round, relative to that of stems. The importance of leaf surfaces in interception processes is evidenced by comparisons of rainfall partitioning during leaf on and leaf off periods in deciduous tree systems. Interception rates during leaf on periods have been reported to be 30% to 67% higher during leaf off (Staelens et al., 2008; Asadian, 2010; Xiao and McPherson, 2011b). As reviewed by Kuehler et al., (2017), efforts to explicitly measure the storage capacity of leaves and stems of different tree species also indicate greater storage capacity of foliage (mean 0.44 ± 0.14 mm) over branch surfaces (mean 0.25 ±0.05 mm). Although these studies did not isolate causal factors driving variability in storage among species, leaf morphology is likely an important factor. For example, Xiao and McPherson (2016) surmise that the rigidity and roughness of Chinese pistache (Pistacia chinensis) contributed to its high leaf storage capacity (mean 1.51 mm) relative to the smooth and flexible leaves of the Bradford pear (Pyrus calleryana; mean leaf storage 0.57 mm). However, as reviewed by Kuehler et al., (2017), a portion of foliar storage is temporary as some losses from leaf drip are expected following cessation of precipitation.

Differences in canopy architecture, leaf and bark orientation, roughness and hydrophobicity, tree size and health, and other morphological characteristics give rise to additional variability in canopy storage among and within evergreen and deciduous tree groupings. While the relative influence of these factors is seldom, if ever, isolated in field studies, several studies have indicated individual and/or sets of tree structural traits that promote various partitioning pathways. For example, Van Stan et al., (2015) measured canopy interception rates to be 5% higher for Yellow poplar (Liriodendron tulipifera), which is characterized by rough bark and a moderately angled (~ 45 degree) branching pattern, than the American beech (Fagus grandifolia), which is characterized by smooth bark and vertically oriented leaves. They attributed this difference to greater water storage in the thicker, rougher bark of L. tulipifera and slower conveyance of rain droplets along the less steeply angled branches. In contrast, the erect branches and smooth bark of F. grandifolia promoted annual average stemflow rates that were about 5 times higher in this species (5.3%) relative to L. tulipifera (0.9%). Despite differences in interception and stemflow between these two species, annual average throughfall rates were not substantially different. Schooling and Carlyle-Moses (2015) observed “optimal” characteristics for stemflow generation included smooth bark and multiple inclined branches leading to the tree trunk, and reported stemflow rates up to 12% of total event precipitation for trees with these characteristics. Bark roughness has also been identified as an influential factor in precipitation partitioning in natural boreal forests, with rougher barked-species tending to have higher rates of interception and lower rates of stemflow (Barbier et al., 2009). Park and Cameron (2008) examined the effects of several canopy traits (leaf area index (LAI), crown length and crown openness) on throughfall for five tree species in a plantation in Panama. Although each of these traits was correlated with throughfall rates, live crown length was the only trait for which the correlation was significant at rainfall depth classes greater than 20 mm. Still, it is likely that traits such as these interact to form a bundle of characteristics that promote throughfall or, conversely, interception. For example, Acacia mangium, which had the highest LAI, longest crown length and lowest crown openness of the five species examined by Park and Cameron (2008), had the lowest throughfall rates (73% of average total rainfall), while throughfall rates among the other species were significantly higher (84% to 89% of average total rainfall). Sadeghi et al., (2016) also found interception rates were positively correlated with crown length. This trait seemed to override other traits associated with high interception rates; for instance, despite having high percent canopy cover and leaf area index, a stand of black locust (Robinia pseudoacacia) had the highest rates of throughfall, which the authors attributed to the trees low crown length. The predictive models developed in these studies do not have high predictive power; thus, there are likely other factors related to tree morphology or storm event characteristics were at play (Park and Cameron, 2008).

Page 67: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

52 The Water Research Foundation

Since it potentially affects canopy architecture, tree management practices also matter. For instance, the manner in which trees are pruned can influence the storage capacity of the canopy and its potential for stemflow generation. In one study, coppicing tended to result in higher (though not significantly different) rates of throughfall from trees of the same species in which pollarding or lopping had been practiced, while lopping tended to significantly increase stemflow rates across events ranging from 2.7 mm to greater than 40 mm (Kaushal et al., 2017).

A.2.1.2 Atmospheric-Based Controls on Canopy Partitioning in Urban Tree Systems Atmospheric-based controls include broad climate variables, such as the climate zone and annual precipitation distribution, as well as event-scale precipitation characteristics, such as intensity and duration, frequency, antecedent dry period, wind and evaporative demand (Table A-3). As indicated by recent reviews on the subject, precipitation depth and intensity are believed to be the primary event-based drivers of precipitation partitioning in urban tree canopies. Multiple studies have shown strong positive correlations between throughfall rates and total precipitation depth and/or intensity (Xiao et al., 2000; Guevera-Escobar et al., 2007; Staelens et al., 2008; Asadian, 2010; Xiao and McPherson, 2011b; Park and Cameron, 2008; Sadeghi et al., 2016). As reviewed by Kuehler et al., (2017), annual average throughfall reported from regions characterized by lower intensity storms (30%) tends to be lower than that reported in regions with higher intensity storms (93%). Positive correlations have also been reported between precipitation depth and interception or stemflow depth (Xiao et al., 2000; Guevera-Escobar et al., 2007; Staelens et al., 2008; Xiao and McPherson, 2011b; Schooling and Carlyle-Moses, 2015; Park and Cameron, 2008; Sadeghi et al., 2016); however, these relationships tend to be more variable. Event rainfall depths in these studies have ranged from less than 5 mm to greater than 90 mm, demonstrating the range over which correlations between precipitation variables and precipitation partitioning have been observed.

Perhaps reflecting their importance in canopy partitioning processes, relationships between precipitation depth or intensity and rates of interception, throughfall and stemflow have received the most attention in the literature to date. However, as reviewed by Kuehler et al., (2017) and Berland et al., (2017), precipitation partitioning, and thus the potential of the urban tree canopy to regulate runoff, is also influenced by a host of other meteorological factors, including wind speed and rainfall inclination (Van Stan et al., 2011; Schooling and Carlyle-Moses, 2015), and vapor pressure deficit (Staelens et al., 2008; Schooling and Carlyle-Moses; Van Stan et al., 2015). Of these, Schooling and Carlyle-Moses (2015) observed that, following precipitation depth, vapor pressure deficit (VPD) was the second most common explanatory variable for stemflow across a diverse sample of deciduous tree types, particularly during leaf-on season. VPD has also been observed to have a positive influence on interception losses (Van Stan et al., 2012). As it represents evaporative demand, these studies indicate that stem and leaf evaporation are an influential mechanism for canopy interception losses and, therefore, stormwater volume management (Van Stan et al., 2015). Wind, which also promotes canopy evaporation and can drive rainfall deeper into the canopy, is also positively correlated with interception losses up to a point. Van Stan et al., (2012) found that interception rates increased along an increasing gradient of VPD and wind speed up to a threshold wind speed of about 4 to 6 m s-1. After this point, interception losses decreased. A similar wind speed threshold (4 m s-1) was observed by Xiao et al., (2000). Reduced interception losses at high wind speeds is likely due to increased stemflow and/or throughfall at higher wind speeds. The importance of wind was also highlighted by Crockford and Richardson (2000), who provide an example of two events in a eucalypt forest with similar intensity and duration but differing wind speeds. In this example, interception rates were about 2 mm more during a storm with low wind compared to an event with high wind speeds (4 to 5.5 m s-1).

Interactions between atmospheric-based controls and tree traits are evident. For example, climate

Page 68: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 53

controls tree phenology and growth rates in a given location, which, as discussed in Section A.2.1.1, effect interception and other partitioning processes. Depending on seasonal precipitation patterns, trees, and deciduous species in particular, may be more or less effective in regulating stormwater hydrology. For example, Kermavnar and Vilhar (2017) suspected that the occurrence of higher intensity storms during the dormant season amplified high throughfall rates observed in their study of urban forest stands in Slovenia. The interplay of storm event and tree traits has also been observed to influence the amount of precipitation delivered to the base of the tree as stemflow, with the same set of tree characteristics (tall, broad and dense canopies) promoting stemflow under events large enough to saturate the canopy but preventing penetration to the trunk during events that didn’t fully saturate the canopy (Schooling and Carlyle-Moses, 2015).

A.2.1.3 Implications of Canopy Partitioning to Stormwater Hydrology From a stormwater management perspective, canopy interception processes constitute a reduction in stormwater volume; thus, maximizing canopy interception by selection of trees with traits that promote interception and understanding the relative canopy losses possible during a storm event with particular characteristics storm events seems a beneficial pursuit. Existing research also indicates the potential for stemflow and throughfall to contribute to stormwater management goals. Stemflow is concentrated at the base of the tree, and therefore presents an opportunity for focused infiltration (Kuehler et al., 2017). Throughfall changes the temporal distribution of incident rainfall, causing a time lag that could be beneficial for peak flow management goals (Kuehler et al., 2017; Livesley et al., 2014). To summarize, there are several pathways by which precipitation partitioning within the canopy of urban tree systems may influence stormwater hydrology:

• Volume reduction via interception • Reducing and delaying peak flow via temporary storage in canopy • Increasing infiltration opportunity via focused delivery of water to tree base

Do the changes in hydrology manifested by these processes occur at a scale of interest to stormwater managers? While they are extrapolated from measurements at the scale of single trees, existing data provide some evidence that precipitation partitioning by trees – and, more specifically, the interspecies traits that influence partitioning – can exert appreciable controls on stormwater hydrology. Consider canopy interception by mature Fagus grandifolia (500 L per event) and Liriodendron tulipifera (650 L per event) observed by Van Stan et al., (2015). Differences in interception were attributed to differences in bark roughness and branching structure of the two species. If these results were applied to both sides of a 1-block (100 m; 330 ft) “green street” in which F. grandifoia and/or L. tulipifera were planted at a 6.1 m (20 ft) spacing, the volume of precipitation retained in the canopy, and thus removed from the runoff budget, would range up to 22 m3 (5800 gal) per storm. In this example, the potential difference in interception capacity between the two species equates to about 5 m3 per storm. But what is the contribution of precipitation partitioned to either throughfall or stemflow? To understand the potential for runoff generation by non-intercepted water below the canopies of urban tree systems, one must consider the mechanisms controlling the fate of rainfall reaching the ground via throughfall or stemflow – this is the topic of the following section.

A.2.2 Processes at the Soil Interface: Infiltration, ET and Runoff Precipitation delivered to the ground surface via throughfall or stemflow may enter the soil profile via infiltration or may leave the canopy area as surface runoff. In the case of infiltration, water is stored in the soil profile and effectively removed from the surface runoff water budget. Evapotranspiration (ET), which includes transpiration (i.e., water removed from the soil profile by vegetation to meet growth needs and atmospheric evaporative demands), serves as an important mechanism for restoring soil

Page 69: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

54 The Water Research Foundation

storage capacity for subsequent storms. Recent reviews by Cappiella et al., (2016), Berland et al., (2017) and Kuehler et al., (2017) agree that there is relatively little research available to quantify these functions for urban trees and their ultimate effect on stormwater hydrology. The following sections review what is known regarding the role of trees in mediating these processes in urban contexts and with a summary of work needed to better understand the implications of these processes to stormwater runoff management presented in Section A.3.1.

A.2.2.1 Infiltration Recent reviews agree there is ample evidence to conclude that trees serve to enhance infiltration rates relative to adjacent pervious surfaces (Table A-3). Potential mechanisms underlying higher infiltration under tree canopies include concentrated infiltration at the base of trees associated with stemflow and enhanced hydraulic conductivity and preferential flow along root channels (see reviews by Johnson and Lehmann, 2006; Van Stan and Gordon, 2018). The soil moisture content of soils underlying tree canopies have also been found to be lower than adjacent grass or other non-canopied areas (e.g., Bharati et al., 2002; Johnson and Lehmann, 2006). This gradient in soil moisture – which may be the result of reduced incipient precipitation due to canopy infiltration, spatially-focused stemflows and/or higher evapotranspiration rates associated with trees – reduces initial soil moisture conditions to promote higher initial infiltration rates, which is likely to result in increased cumulative infiltration and lower runoff potential where trees are present.

Although infiltration is highly variable and is controlled in part by soil properties, which will vary from tree to tree, it is informative to consider relative differences in infiltration between systems with and without trees as observed in studies to date. While it is difficult to fully attribute these differences to trees or to pinpoint the mechanisms responsible for these differences, studies comparing infiltration rates under trees to adjacent open spaces have reported differences ranging from 0.7 to 27 times higher (Table A-4). At the low end of this range, study authors surmised that coarse soils supported high infiltration rates under both trees and adjacent open space such that observed differences were very slight, or even lower, under trees (Zadeh and Sepaskhah, 2016). The high end of this range was observed in a column study in which edge effects and other scaling issues are likely to exaggerate differences (Bartens et al., 2008). In the middle of this range, Hart (2017) reported increases up to about 1.5 to 3.5 times in bioswales with and without trees. This is similar to the range reported by Zadeh and Sepaskhah (2016) for infiltration measurements under trees and in adjacent open space within finer (clay loam) soil textures and by Hubbart et al., (2011). Regardless of the specific mechanism(s) underpinning these observations, it is likely that root density plays a role in facilitating infiltration enhancements. Hart (2017) demonstrated a strong, positive linear correlation between seasonal infiltration rates and root mass density present in bioswale facilities. Furthermore, the positive association was stronger for subsoil root mass versus topsoil root mass, thus demonstrating the importance of deeper rooting plants (such as trees) for infiltration performance. Surface litter accumulation under tree canopies has also been associated with improved soil health and infiltration capacity. For example, Livesley et al., (2016) found soils below tree canopies were significantly less compacted than adjacent grass areas.

Page 70: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 55

Table A-4. Summary of Infiltration Studies with Urban or Open-Canopied Trees. Source Location Scale Sample size General findings

Bartens et al., (2008)

Blacksburg, VA (greenhouse)

Column scale (60-cm dia.; 1-2 cm dia. tree seedling)

N = 2 species x 2 soil types x 3 drainage regimes x 5 replicates + unplanted controls

Infiltration rates 1.5 and 27 times higher in columns with trees than unplanted controls for columns with compacted silt loam and structural soils overlying compacted silt loam, respectively

Zadeh and Sepaskhah (2016)

Badjgah, Iran Tree scale, infiltration measured w/in 0.5 m of trunk and in adjacent open space

N = 6 trees x 2 unsaturated infiltration measurements per tree

Infiltration rates under trees (136 to 233 mm hr-1; 5.4 to 10.7 in hr=-1) were .7 to 3.5 times higher than in adjacent open space

Bharati et al., (2002)

Story County, Iowa

Tree-scale N = 1 tree x 24 unsaturated infiltration measurements + 5 control treatments x 24 measurements each

Infiltration rates significantly (2-9 times) greater under trees than in adjacent cropland and pasture; up to 2 times greater than adjacent riparian grasses

Hubbart et al., (2011)

Columbia, Missouri

Tree-scale within urban riparian buffer

N = 42 infiltration measurements x forest and agricultural (control) floodplain sites

Significantly higher mean saturated infiltration in forest (380 mm hr-1) than ag (230 mm hr-1) sites

Hart (2017) Portland, Oregon

Bioswale (20 m2), 1-2 trees each

N = 10 bioswales with trees, 5 without x 58 precip events x 1 infiltration measurement per event

Infiltration rates higher in bioswales with Juncus-tree mix (40 to 70 mm hr=-1) than Carex only (20 to 40 mm hr=-1); rate correlated ( + ) with root density

Lange et al., (2009)

Near Rueschegg, Switzerland (northern Pre-Alps)

1-m2 plots, positioned 1 to 4 m from tree trunk within spruce forest

N = 13 plots (3 tree species) x 3 irrigation events

Tree root density significantly correlated with soil bulk density ( – ) and gravitational drainage ( + ); roots identified as key drivers of infiltration in clayey, periodically saturated soils

A.2.2.2 Evapotranspiration Evapotranspiration (ET) serves a critical role in restoring the water storage capacity of soils underlying trees or other vegetated systems in between rainfall events. Drier soils generally have higher initial infiltration rates, so, in addition to increasing the volume of available soil water storage, evapotranspiration also effects the initial rate of infiltration. Transpiration, which is the portion of ET attributed to soil water use by trees or other vegetation, is driven by a combination of physical factors characterizing atmospheric evaporative demand (e.g., solar radiation, temperature, humidity, wind) and biological factors characterizing water demands of the tree (e.g., species, phenology, leaf area). Transpiration rates are limited by available soil water; thus, as readily available water in the soil profile declines and soil matric potential increases, transpiration rates slow. Besides soil texture, which controls available soil water, other soil properties can influence transpiration rates. Compacted soils, which tend to hold less readily available water, have also been found to suppress tree transpiration rates (Fair et al., 2012), as have soils with slow infiltration rates (Bartens et al., 2009). Tree transpiration may be higher for trees planted over impervious surfaces, which are thought to increase long wave radiation, relative to pervious surfaces (Kjelgren and Montague, 1998). Tree spacing has also been associated with ET rates, with closer growing trees tending to have lower transpiration rates (Hagishima et al., 2007). Thus, open grown trees typical of urban street or residential plantings are likely to support higher transpiration rates than natural or managed forests in a comparable climate.

As reviewed by Cappiella et al., (2016) and Kuehler et al., (2017), transpiration rates in urban or open-

Page 71: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

56 The Water Research Foundation

canopy tree systems have been measured in at least 10 studies for over 30 tree species. Growing season transpiration rates reported in these studies ranged from 0.1 to 2.4 mm/day per unit leaf area or projected canopy cover. As an example of the impact of tree transpiration on urban hydrologic budgets, Scharenbroch et al., (2016) estimated that up to 70% of annual precipitation and runoff intercepted by a tree-bioswale system was removed from the bioswale soil media layer via transpiration. Thus, the importance of ET as a means to replenish the water storage capacity of the underlying soils is evident and a benefit to stormwater managers.

As noted by Kuehler et al., (2017), species-specific transpiration rates could be considered in selecting trees to provide greater stormwater quantity regulating benefits. Differences in species transpiration rates could result in substantial differences in the ability to dry soils for future runoff storage; for example, Riikonen et al., (2016) measured rates of 20 to 40 L day-1 and 30 to 60 L day-1 for Tilia x vulgaris (common linden) and Alnus glutinosa (common alder) street trees, respectively, under similar planting and climate conditions. Though current measurements of tree transpiration in urban environments are limited to a few studies and a handful of species, crop or water-use coefficients, which account for tree-specific controls on ET, have been developed for common fruit- and other agronomic trees (e.g., Allen et al., 1998) as well trees more typical of urban landscape plantings (e.g., Levitt et al., 1995; Niu et al., 2006). The availability of these coefficients enables use of the Penman-Monteith reference evapotranspiration method to estimate ET by urban trees (Allen et al., 2005), and thus to enumerate their contribution to regulation of urban watershed hydrology under various climatic conditions. However, as demonstrated by Riikonen et al., (2016), applying general crop coefficients can be problematic if the coefficients were developed for closed canopy conditions, a difficultly that was overcome by adding a canopy conductance and leaf area development submodel. Still, the reference ET method provides a first estimate of evapotranspiration associated with various urban tree systems, including relative differences between species.

It is important to consider the temporal distribution of tree system ET relative to that of local precipitation. Since ET rates are lower during the non-growing season, the contribution of this process to stormwater regulation will be less impactful in areas that receive the majority of storm events during the winter. Because they retain their leaves year round, evergreen trees tend to use more water throughout the year. This may be an additional hydrologic advantage of evergreens in locations that receive the majority of their precipitation during the non-growing season.

A.2.2.3 Runoff Surface runoff occurs as infiltration capacity is exceeded; therefore, the magnitude and timing of surface runoff is strongly linked to infiltration rate relative to precipitation and initial soil moisture conditions, which are influenced by evapotranspiration. Studies of runoff associated with individual urban tree systems have primarily been conducted in the context of structural green infrastructure systems designed for stormwater control, including tree pits with specialized media, tree box filters, and bioswales (Table A-5).

Although the systems have included trees, the majority of these studies have not been designed to specifically address the contribution of trees to runoff reductions in engineered green infrastructure systems through, for example, inclusion of unplanted controls. Therefore, it is difficult to ascribe a runoff reduction value to trees relative to other vegetation or soil media with high hydraulic conductivity (as is most typical of these engineered systems) on the basis of existing field data. However, as discussed in Section A.2.2.2, trees play a large role in the hydrologic budget of these systems through transpiration (e.g., up to 72% of total water budget; Scharenbroch et al., 2016) and are thus likely to improve runoff reduction by restoring available soil storage capacity between storms. Page et al., (2015)

Page 72: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 57

observed that runoff bypass from a tree planted in suspended pavement appeared to be driven by precipitation depth rather than intensity. Thus, recovery of soil storage through tree evapotranspiration would seem a critical process for reducing runoff.

The effect of urban tree systems on runoff timing has not been quantified (Kuehler et al., 2017), but lags in throughfall and stemflow provide some indication of the capacity for trees to delay peak flows. For example, throughfall initiation has been observed to lag precipitation by up to 5 hours (Asadian, 2010), and, once initiated, throughfall and stemflow were observed to continue for up to 60 minutes after the cessation of rainfall (Xiao et al., 2000). Additional work is needed to understand the net effect of canopy partitioning pathways and the interplay of storm event characteristics on runoff generation and peak flow.

Table A-5. Summary of Studies in Which Runoff Is Quantified for Individual Urban Tree Systems. Source Location (study

dates) System type and

scale Number of

observations General findings

Armson et al., 2013

Manchester City, UK (Dec 2010 – May 2011)

Tree pit (3.3 m2) with attendant asphalt watershed (5.7 m2)

N = 9 tree/asphalt plots; 9 asphalt controls

Runoff volume from tree + asphalt plots 62% lower than asphalt plots; study not design to isolate hydrologic contribution of tree

Geronimo et al., 2014

S. Korea (July 2010 – Aug 2012)

Tree box filter (3.3 m2) with attendant asphalt watershed (312 m2)

N = 1 tree x 11 storm events

Runoff volume reduced 40% up to hydraulic loading rate of 1 m/day; not designed to isolate hydrologic contribution of tree

Page et al., 2015

Wilmington, NC (Sept 2012 – June 2013)

Silva cell (405-485 m2) and attendant paved watershed (0.2-0.3 ha)

N = 2 cells x 53 storms

65% (693 mm of 1076 mm) of runoff filtered through system; 15% overflow; remainder (20%) attributed to evapotranspiration

Xiao and McPherson, 2011a

Davis, California (Feb 2007 to Oct 2008)

Bioswale (tree + structural soil; 25 m2) and attendant paved watershed (180 m2)

N = 1 bioswale + 1 control (tree in native soils) x 50 storms

3% bypass from bioswale; 50% bypass rate from control; Runoff coefficient from bioswale was 47 % lower than control

Scharenbroch et al, 2016

Morton Arboretum, Lisle, Illinois (May – Oct of 2012-2014)

Bioswale (3 m wide) and attendant paved watershed (2 ha)

Not provided Bioswale discharge accounted for 0.02 to 1.7% of total precipitation + runoff inputs

A.2.3 Water Quality and Urban Forest Systems Of recent reviews, Cappiella et al., (2016) is the only one to consider water quality implications of urban forest systems. Cappiella states that the primary mechanism for water quality regulation by urban trees is reduction of runoff volume (and thus pollutant load); however, this role is countered by nutrient contributions of leaves on impervious surfaces. Given the potential for urban trees to serve as a net sink or source of pollutants in stormwater runoff, the following sections consider conditions in which urban tree systems may sequester or contribute to stormwater pollutant loads. Here, the primary emphasis is upon nitrogen (N) and phosphorus (P) forms; however, other pollutants of concern in urban environments are summarily addressed. This section concludes with a summary of the conditions in which trees may be expected to provide an ecosystem service or disservice with respect to stormwater quality regulation.

A.2.3.1 Trees as Pollutant Sinks: Mechanisms for Runoff Quality Improvements Among the mechanisms by which trees may serve to sequester runoff pollutants are filtration of particulates and particulate-bound pollutants – both in stormwater runoff and as deposited atmospherically – and biological uptake or transformation. As suggested by Xiao et al., (2000), trees may also reduce stormwater pollutant loads by regulating runoff volume, particularly for small storms with which the majority of pollutant load has been associated (Schueler, 1987) and over which trees exert the

Page 73: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

58 The Water Research Foundation

greatest hydrologic control.

While the mechanisms for pollutant sequestration in tree systems are understood in general terms, the relative contribution of trees in mediating these processes is less so, particularly at the field scale. As reviewed by Cappiella et al., (2016), there is a small but growing number of field studies demonstrating pollutant removal potentials by structural green infrastructure systems incorporating trees (e.g., biofiltration, tree box filters, Silva Cells). Of the nine studies reviewed, percent reduction of N and P forms was highly variable, with one study reporting nutrient export (54% increase in total P, due in part to low influent concentrations) but with at least 6 studies reporting mean event or annual removals greater than 50%. While these studies indicate trees can be incorporated in green stormwater infrastructure systems that achieve nutrient reductions, they are rarely designed to understand the contribution of trees to water quality observations. Unvegetated or alternatively vegetated controls (e.g., grasses) are one approach to provide perspective to the contribution of trees in GI systems. In a column study representing biofiltration systems, Denman et al., (2016) reported significantly lower nutrient concentrations in leachate collected from columns planted with trees (range in TP undetectable to 0.6 mg P L-1; NO3- < 0.1 to 5 mg N L-1) relative to unplanted controls (range in TP 0.6 to 1.4 mg P L-1; NO3- 5 to 12 mg N L-1). Leachate water quality did not differ among tree species or evergreenness, indicating that general properties of tree root zones may mediate processes central to runoff treatment within soils. Read et al., (2008) and Payne et al., (2013) also observed lower nutrient concentrations from columns planted with trees relative to unplanted controls. Additional insights to the conditions in which trees or other vegetation may influence nutrient export from GI systems were also gained from these studies. First is the influence of tree (or vegetation) traits: following observed correlations between root mass and pollutant concentration reduction for some pollutants suggests plant roots, either through direct sorption and uptake or via enhanced microbial processing, play an important role in treating stormwater runoff, and species with greater root mass may perform better (Read et al., 2008). Species adapted a broader range in soil moisture regime may also provide greater pollutant treatment services. In a column study including replicates of 22 species (2 tree, 4 shrubs, and 16 herbaceous), Payne et al., (2013) observed that effluent nitrogen concentrations from drought tolerant species were less than those from all other species following prolonged dry periods, perhaps due to lower mortality and, hence, reduced nutrient leaching from dead root and microbial cells. Second is the influence of system design. For example, while differences in pollutant retention were observed among species examined by Payne et al., (2013), these differences were minor relative to differences among columns with and without a saturated zone. Thus, the authors conclude that, in biofiltration GI systems treating stormwater, trees and other vegetation are important but secondary to hydraulic design, and should therefore be selected to promote biodiversity and function across a range of climatic (e.g., wet, drought) conditions.

Relative to structural GI systems for stormwater treatment, the role of nonstructural urban tree systems in sequestering runoff pollutants has been less studied. Owing to their landscape position down gradient of runoff sources, riparian buffers have the potential to regulate stormwater quality. For example, Bu et al., (2016) found that the removal efficiency of particulate nutrient forms increased with tree planting density within integrated tree-grass riparian buffers intercepting cropland surface runoff, and surmised the thicker litter layer associated with higher tree densities enhanced filtration processes. Removal was significantly greater under low intensity events (< 4 mm hr-1), likely due to higher canopy interception and infiltration rates (also measured as part of the study) within higher density tree plots. Enhancement of these processes would serve to reduce pollutant load via runoff volume reductions. Biological removal pathways, namely denitrification and uptake, have also been studied in urban riparian buffers. Groffman and Crawford (2003) found that forested riparian buffers in urban areas retained similar denitrification potential in the upper soil layer relative to forested and herbaceous buffers in rural areas.

Page 74: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 59

Denitrification did not differ among forested and herbaceous buffers; rather, denitrification potential was significantly correlated with soil moisture and soil organic carbon. Thus, if riparian areas are managed to maintain soil organic carbon stores and periodic saturated conditions, nitrate removal from runoff can be expected, though other landscape controls (e.g., slope, soil type and depth) are likely to exert an overlying influence on the magnitude to which these more “natural” tree systems regulate stormwater quality (Miller et al., 2005).

A.2.3.2 Trees as Pollutant Sources: Mechanisms for Runoff Quality Degradation Although various processes are at work in tree-soil systems to sequester nutrients and other pollutants in surface runoff, these systems may pose a net nutrient sources to surface waters. Litterfall – which includes tree leaves, seeds and flowers – is the primary nutrient source associated with trees (e.g., Janke et al., 2017), though stemflow and throughfall can also deliver nutrients and other ions to surface runoff (e.g., Abas et al., 1992; Barbier et al., 2009; Johnson and Lehmann, 2006; Van Stan et al., 2012; Xiao and McPherson, 2011b). In undeveloped landscapes, nutrients supplied by tree and forest systems act as an important input to aquatic ecosystems, serving to fuel the primary productivity that underpins aquatic food webs. However, in developed landscapes, where many aquatic systems are already impaired for excess nutrient levels, nutrients leached from leaf litter may contribute further to water quality degradation. The role of trees as a nutrient source in urban areas is exasperated by the efficient hydraulic design of urban drainage systems, which by rapidly delivering runoff and associated pollutants to aquatic ecosystems in gutters and pipes, effectively short-circuit the treatment processes provided by soils-vegetation systems. The hydraulic expediency of streets and other impervious surfaces on which litterfall tends to accumulate, coupled with the soluble nature of nutrients in senesced litter (Hobbie et al., 2014) make trees a likely nutrient source in urban watersheds. At least two studies have employed experimental approaches revealing the role of trees as nutrient sources at the watershed scale. These are reviewed in Section A.2.4.

A.2.3.3 Water Quality and Urban Tree Systems: Implications The conditions in which urban trees may act as a pollutant source or sink will likely depend on the type of system under consideration. In cases where the system receives runoff from up-gradient urban land cover and litterfall has the opportunity to interact with pervious surfaces rather than accumulating on streets or other impervious surfaces, tree systems may reduce net nutrient loads (e.g., Bu et al., 2016). Litterfall in such systems has been associated with increased water storage capacity of underlying pervious surfaces, thus enhancing hydrologic benefits provided by the tree system (Bu et al., 2016). In contrast, urban tree systems that overhang impervious surfaces, particularly those that are directly connected to storm drains rather than pervious GI systems, are likely to serve as a net nutrient source to stormwater runoff. Although stormwater volume reductions achieved by trees have the potential to offset nutrient additions associated with litterfall, this offset becomes negligible at street densities representative of more densely populated urban areas (Janke et al., 2017).

Given the research to date, active management of litterfall through street sweeping or other preventative measures is essential to avoid unintended stormwater nutrient enrichment by urban tree systems. Several studies have demonstrated the potential effectiveness of leaf litter management programs (e.g., Selbig, 2016; Templer et al., 2015). It is also possible that volume regulation by urban tree systems provide more substantial nutrient offsets in locations with lower intensity precipitation or that receive the majority of precipitation during winter months after litterfall could be successfully managed with street sweeping; however, additional research is needed to determine the extent to which such offsets may exist. The nutrient content and lability of tree litter has been shown to vary among tree species (Dorney, 1986); thus, leaf nutrient characteristics could be considered alongside hydrologic characteristics in the selection of urban tree species.

Page 75: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

60 The Water Research Foundation

A.2.4 Scaling up: Hydrologic and Water Quality Implications at the Street-Scale and Beyond The preceding sections have focused primarily on hydrologic and water quality studies conducted at the scale of individual trees within urban tree systems. Studies at this scale represent the bulk of quantitative measurements on urban tree systems and are important to improve the mechanistic understanding of precipitation partitioning and/or water quality processes by urban trees. However, stormwater is managed at the watershed scale; thus, it is also important to understand how these processes translate from the scale of individual trees to larger management units. In this section, we review studies in which researchers have attempted to tease out the aggregate hydrologic (Section A.2.4.1) or water quality (Section A.2.4.2) influences of urban tree systems at scales beyond single trees.

A.2.4.1 Hydrologic Influences of Urban Tree Systems from the Yard to Watershed Scale Among this and other recent review efforts, 4 field studies were identified in which the aggregated effects of precipitation partitioning by urban trees were examined at scales greater than individual trees. As summarized in Table A-6, the measurement scale of these studies have ranged from residential yards (Inkilainen et al., 2013) to watersheds (Donovan et al., 2016).

At the yard-scale, Inkilainen et al., (2013) obtained throughfall measurements that reflect a composite of canopy processes in mixed tree-turf residential yards. Compared to measurements at the scale of individual trees, yard-scale throughfall measurements were higher (mean 89% versus 71%). The strongest variables explaining observed throughfall rates in urban residential forests were % canopy cover, tree density, structural complexity and % coniferous (all negatively correlated with throughfall). Yard-scale runoff quantity or quality was not measured as part of this study, though runoff risk would be expected to increase as throughfall rates increased. Bu et al., (2016) provide evidence that tree planting density (and associated measure of % canopy cover) does influence runoff generation. In a study of an integrated poplar-grass riparian buffer with planting densities ranging from 10 m2 to 30 m2 per tree, discharge from higher density plots was significantly less than plots planted on 6x5 m centers (223 mm versus 284 mm), indicating the role of canopy interception (measured as 242 mm versus 126 mm for planting densities of 10 m2 versus 30 m2). That the difference in plot-level runoff volume was only 20% despite a 100% difference in canopy interception likely reflects the role of infiltration in managing stemflow and throughfall. Indeed, infiltration rates did not differ with poplar planting density and were relatively high (approximately 240 to 270 mm hr-1 at saturation; 9.5 to 10.5 in hr-1). From the standpoint of stormwater management, it is informative to note the decline in hydrologic regulating capacity of the buffer with increasing precipitation intensity; for storm intensities less than 4 mm hr-1, buffers reduced runoff by 24 to 29 mm on average but by only 9 to 10 mm for rainfall intensities of 6 to 8 mm hr-1.

A handful of studies have been conducted to reveal the influence of trees on watershed-scale stormwater quality and/or quantity. Most field studies at this scale rely on measurements at the watershed outlet to infer the effects of trees and other land cover types on runoff hydrology and/or quality. The majority of studies to quantify watershed-scale effects of trees have been conducted in forested watersheds based, for example, on changes in streamflow before and after forest clear cutting. There are a limited number of studies that attempt to link urban forest patterns with watershed-scale hydrologic or water quality response (Table A-6). Donavan et al., (2016) examined changes in runoff among 34 watersheds in Portland, OR. They used differences in hydrologic response between a summer (leaf-on) and winter (leaf-off) storm event and differences in canopy cover among watersheds to infer the influence of trees on watershed runoff processes. Their analysis indicated a 1% increase in canopy cover was linked to 4,550 m3 reduction in storm sewer flow for the summer event (0.4 mm based on the average pipeshed area). However, canopy cover had no effect on runoff generation during the winter event. Boggs and Sun (2011) also provide evidence for the limited hydrologic regulating function

Page 76: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 61

provided by trees during the dormant season. In a comparison of an urbanized (56% forested/open space) and forested (99% forested) watershed in the North Carolina Piedmont, they found watershed discharge coefficients were not substantially different during the dormant season. In contrast, discharge coefficients from the forested watershed were an order of magnitude less than those from the urbanized watershed during the growing season. The study authors attributed much of this difference to ET-driven increases in available soil moisture storage in the forested watershed relative to the urban. Despite having 56% forest and open space cover, seasonal discharge coefficients in the urban watershed were not appreciably different, indicating a negligible influence of ET. Thus, ET appears to wield an important influence in regulating stormwater discharge at watershed scales, and that work is needed to better understand how to maximize this process in urban tree systems.

In addition to studies conducted in urban contexts, Cappiella et al., (2016) also reviewed watershed-scale studies of runoff processes in natural forests. As with urban studies, these studies have relied primarily upon streamflow monitoring data to infer the influence of canopy cover on hydrologic processes in closed canopy forest systems. Syntheses of these studies have indicated that, on average, a 10% decline in canopy cover results in annual water yield increases of 40 mm and 25 mm in evergreen and deciduous hardwood forests, respectively. These larger-scale studies in natural forests have identified factors such as seasonality of runoff events and characteristics of dominant subsurface flowpaths as being important to watershed hydrologic regulation (Post and Jones, 2001). However, as with studies at the scale of individual trees, it is difficult to extrapolate the findings to urban watersheds due to differences in canopy structure and, perhaps more importantly at the watershed scale, the presence of impervious surfaces and expanded, hydraulically expedient drainage networks.

The shortage of watershed-scale hydrologic studies within urban tree systems reflects the difficulty of implementing experiments at this scale (e.g., having multi-year streamflow records for co-located watersheds with tree cover characteristics suitable for a before-after-control-impact study design) and then analyzing data from such studies in such a way to isolate effects of tree cover over all the other “noise” (e.g., interannual climate variability, effects of underlying geology, unknown subsurface flowpaths, etc.) present in these environmental datasets. In light of these difficulties, hydrologic models are more commonly used to examine watershed hydrologic response to changes in urban canopy cover. While none of these models are “right,” they do provide insight to the relative influence of urban tree systems on watershed runoff hydrology. Modeling approaches include both empirical models such as the Curve Number, in which physical processes are represented by experimentally-derived parameters, and mechanistic models, in which actual physical and biological processes are represented through inclusion of measurable system attributes driving those processes in the model. Both types of models have been used to test questions related to the influence of urban tree systems on stormwater runoff and quality. For example, both Sanders (1986) and Matteo et al., (2006) used Curve Number based approaches to model the net effects of tree cover on runoff generation. Matteo et al., (2006) incorporated a spatial component with the GIS-based Generalized Watershed Loading Function model to test the relative impacts of implementing forested roadside and riparian buffers in urban and urbanizing watersheds. Although the model was not calibrated, runoff volume reductions relative to the non-buffer case (up to 9% and 15% reduction in the urban and suburban watersheds, respectively) indicate such treatments could have impact at the watershed scale. More mechanistic approaches have also been applied to model the effect of urban tree systems on runoff hydrology. For example, Gonzalez-Sosa et al., (2017) coupled a mechanistic canopy interception model with an (empirical) runoff model to examine the effects of street trees at the pipeshed scale. They reported runoff reduction of 10% to 20% and peak lag of 10 to 15 minutes at the pipeshed outlet relative to a street without trees. At the watersheds scale, models such as i-Tree Hydro (formerly UFORE-Hydro) and DHSVM have been used to examine changes in watershed discharge associated with urbanization and decreasing forest cover

Page 77: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

62 The Water Research Foundation

through time (Wissmar et al., 2004) and with various policies to increase canopy cover (Wang et al., 2008).

Table A-6. Summary of Studies Quantifying Hydrologic and/or Water Quality Benefits of Urban or Open-Canopy Tree Systems beyond the Scale of Individual Trees.

Source Location (Dates) Scale Sample size General findings Inkilainen et al., (2013)

Raleigh, NC (July – Nov. 2010)

Residential yard (497 + 255 m2 front yards; 854 +/- 810 m2 back)

N = 16 yards (total 206 measurement points) x 14 storm events

Mean yard-scale TH 88.9% (range 83.2 to 98.3%) of gross precip; Canopy cover strongest predictor of TH, followed by tree density, vertical structural complexity index, % coniferous, and LAI

Bu et al., (2016)

Yixing City, Jiangsu Province, China (Nov 2013 – Oct 2014)

Tree-grass buffers (208 m2)

N = 4 planting densities x 3 replicates x 14 storm events

Significantly greater runoff reduction, sediment and particulate P with increasing tree density

Donovan et al., (2016)

Portland, Oregon (June and Dec 2010)

Sewersheds (1,113 +/- 1,907 Ha), tree canopy cover 29 +/- 7.6

N = 34 watershed x 2 storm events

Developed statistical models relating storm sewer flow and land cover; models indicted 1% increase tree canopy cover would have reduced storm sewer flow by 4,550 m3 in June; no affect of trees or other vegetation cover in Dec.

Boggs and Sun, 2011

Piedmont region, North Carolina (2000 to 2007)

Watershed (0.7 and 2.95 km2 urban (56% tree cover) and forested (99% tree cover)

N = 2 watersheds, 7 years streamflow monitoring data

Urban watershed converted 42% of precip to discharge, forest only 24%. Stormflow 75% higher in urban watershed.

Modeling studies Source Location (model) Scale Sample size General findings Gonzalez-Sosa et al., (2017)

Queretara, Mexico (Rutter interception model + rational method, unit hydrographs)

Street (Street lengths 1.3 km and 0.9 km; 8km2 pipeshed )

N = 2 streets, 2 year simulation period

Delayed rainfall peak bt 10-15 minutes; reduced runoff 10-20% relative to treeless condition

Wang et al., 2008

Baltimore, Maryland (UFORE-Hydro, now i-tree Hydro)

Watershed (14 km2) N = 1 watershed x 3 land cover scenarios x 1 yr simulation period

Annual runoff reduced by 2.6% via 28% increase in canopy cover over pervious surfaces; 3.4% reduction for 35% increase over impervious surfaces

Wissmar et al, 2004

Cedar River, Renton, Washington (DHSVM)

Watershed (2 to 38 km2)

N = 7 watersheds x 3 design storms x 3 land cover scenarios

Watershed discharge (m yr-1) varied by about 25% across a 20% difference in forest cover in urban watersheds for 2-yr and 25-year recurrence interval storms

Matteo et al., 2006

Springfield, MA (GWLF, Curve Number)

Watershed (450 to 1230 Ha)

N = 3 watersheds, 5 yr simulation period

In suburban watershed, runoff reductions modeled for street- and riparian forest buffer policies were 5% to 12% and 5% to 9% in urban watersheds.

A.2.4.2 Water Quality Influences of Urban Tree Systems at Watershed Scales Watershed-scale field research has provided evidence that, despite their hydrologic benefits, trees may serve as nutrient sources to urban water bodies. For example, in an analysis of nutrient concentrations from over 2,000 storm events across 19 watersheds in the Minneapolis, MN metropolitan region, Janke et al., (2017) revealed strong and significant linear correlations between percent street canopy cover and total phosphorus (TP), total nitrogen (TN), and total organic nitrogen (TON) runoff concentrations. Simple linear models developed in the study indicated a 10% increase in street canopy cover was correlated with approximate increases of 15% and 10% in TP and TN, respectively. While still significant, the relationship between street canopy cover and TN and TP concentrations decreased slightly for canopy cover within a larger buffer adjacent to streets. This result suggests that canopy cover

Page 78: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 63

overhanging pervious surfaces is likely a lesser source of nutrients to stormwater than canopy that overhangs streets. This result must be weighed against modeling evidence that suggests the greatest hydrologic benefits are obtained from trees overhanging impervious surfaces (Wang et al., 2008). With respect to nutrient loads, which tend to be more important than concentrations for aquatic processes such as eutrophication, were best predicted by impervious area and, more specifically, street density. Street canopy cover was positively, but not significantly, associated with nutrient loads. This result suggests that, at least for the set of conditions associated with the data analyzed by Janke et al., (2017), volume reductions by tree canopy did not appreciably offset stormwater nutrient loads.

At the watershed scale, a weight of evidence approach is required in light of the difficulty to isolate hydrologic and water quality influences of urban tree systems from other, oft unaccounted for, watershed patterns and processes. The correlations between the spatial extent of urban canopy cover and watershed water quality measurements are supported by measurements of nutrient loads associated with litterfall at watershed outlets. Stack et al., (2013) estimated TP and TN loads associated with leaf litter of 40 kg P km2 yr-1 (0.36 lb P ac-1 yr-1) and 530 kg N ac-1 yr-1 (4.7 lb N ac-1 yr-1), respectively, as measured from leaf litter collected stormwater outfalls from an urban watershed in Maryland. These measurements are in line with nutrient yields modeled by Janke et al., (2017) as a function of canopy cover and street density. As reviewed by Cappiella et al., (2016), nutrient export from nonurban forests, whether in reference or disturbed conditions, are on the same order of magnitude for TN, and are nearer to reference conditions for TP.

Trees certainly are not the only source of nutrients to stormwater runoff. In the study by Janke et al., (2017) street canopy cover was only weakly correlated with dissolved P and bore almost no correlation with nitrate and ammonium concentrations, indicating other sources of these nutrient forms. Potential sources could include lawn fertilizers, vehicle emissions and atmospheric deposition, leaky sewers, and pet waste (e.g., Bettez et al., 2013; Fissore et al., 2012; Janke et al., 2017). However, as is consistent with national data (Tchobanoglous et al., 2004), organic nitrogen forms can comprise a substantial portion of stormwater nitrogen loads, underscoring the importance of considering trees as a source of nutrients to stormwater runoff.

A.3 Integration of Urban Tree Systems and Stormwater Management: Implications and Knowledge Gaps From the research to date, there is sufficient evidence to suggest that urban tree systems have a positive influence on stormwater hydrology as stormwater volume reductions have been measured at scales of relevance to stormwater managers (e.g., pipeshed) during the growing season. Studies up to the scale of small plots indicate that the primary mechanisms by which stormwater runoff volume reductions may be achieved are interception and, perhaps to a greater extent, the intertwined processes of infiltration and evapotranspiration.

Studies at the scale of individual trees also indicate that tree structural attributes (e.g., canopy architecture, bark roughness, leaf angle) can influence precipitation partitioning to an extent that matters for stormwater management, at least for small storms. For instance, Van Stan et al., (2015) provide an example of how structural differences between two deciduous tree species gave rise to a 20% difference in canopy interception capacity and, in effect, their potential to regulate stormwater runoff.

Although stemflow tends to comprise the smallest component of canopy partitioning, it is our opinion that it should not be counted as trivial to stormwater management. Since stemflow focuses precipitation at the base of the tree, ensuring the infiltration capacity of the soils at the tree base

Page 79: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

64 The Water Research Foundation

presents another opportunity to remove precipitation from the runoff budget. If the infiltration capacity is sufficient, stemflow could be treated as an advantageous hydrologic process for stormwater management, and urban tree systems could be designed to promote this process through selection of species with stemflow-promoting traits (e.g., high branch angles, smooth bark and multiple secondary leaders; Schooling and Carlyle-Moses, 2015). Based on the range in stemflow rates within individual studies reviewed herein, stemflow generation between species could vary by a factor of 2 of 3 and account for over 15% of total precipitation (Xiao et al., 2000; Sadeghi et al., 2016), highlighting the opportunity to mitigate runoff generation through stemflow processes.

Since it potentially affects canopy architecture, tree management practices should also be considered when integrating tree systems with stormwater management goals. While only one study was identified specifically addressing the effects of canopy management practices on precipitation partitioning, it appears that management can at least influence stemflow generation (Kaushal et al., 2017). If management practices increase stemflow, it would be important to ensure that the base of the tree has adequate infiltration capacity to capitalize on the potential to limit stormwater runoff.

The literature is clear that canopy architecture is not the only factor that influences the role of trees in regulating stormwater runoff. Atmospheric conditions – whether considered at the scale of seasonal climate or intra-event meteorological conditions – also influence canopy partitioning and subsequent runoff potential. At the scale of seasonal climate, the overall effectiveness of urban trees in stormwater hydrologic regulation is likely controlled to some extent by the rainfall distribution of a particular city. For example, several studies reviewed herein reported higher rates of canopy interception under low intensity rainfall; thus, trees likely intercept a greater proportion of total precipitation in regions typified by low intensity events. For cities in which the majority of precipitation occurs during the dormant season, as in Mediterranean climates, evergreens are likely to be more effective for stormwater volume control than deciduous trees (Xiao and McPherson, 2016).

While consensus is relatively clear to the positive impact of urban tree systems on runoff hydrology, the effects on water quality are still subject to some debate. Efforts to link water quality and the extent of urban trees at the watershed scale with field data suggest trees serve as a net nutrient source to urban runoff (Stack et al., 2013; Janke et al., 2017). That trees contribute nutrients to runoff is not surprising; estimates of nutrient load export from partially canopied urban systems are on par with loads presented by Cappiella et al., (2016) for full canopy forested systems. The evidence to date suggests that leaf litter management programs must be advanced alongside of efforts to incorporate trees as part of the green stormwater infrastructure. Given the high infiltration capacity of tree systems relative to other pervious surfaces, implementing street trees and other urban tree systems such that they are hydrologically connected to impervious surfaces to promote infiltration of runoff could also help mitigate nutrient additions of litterfall. It should be noted that studies addressing water quality impacts of urban tree systems have been conducted in the Midwest and eastern US. Therefore, more research is needed to understand if there are sets of conditions (e.g., within areas with low intensity rainfall or high proportion of evergreen species) in which the hydrologic regulating benefits provided by urban tree systems can offset nutrient additions of litterfall.

Based on the literature to date, one could begin to envision a “designer” tree species with characteristics that maximize interception and, potentially, stemflow for the purpose of maximizing stormwater hydrologic benefits. However, fixation on a single or small number of trees is probably not advisable. Rather, a diverse mix of species adapted to a particular site is likely preferable, both because mixed-species systems tend to be more resilient to environmental stressors (e.g., Beniston et al., 2007) and, as observed by Kermavnar and Vilhar (2017), may provide equal or greater hydrologic benefits. In

Page 80: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 65

considering stormwater quality and quantity regulation beyond the scale of individual trees, examination of the hydrologic and/or water quality impacts of structurally diverse urban tree systems is also worth consideration.

A.3.1 Knowledge Gaps of Relevance to Stormwater Management One of the most frequently cited research needs in the literature to date was the need to better understand how hydrologic and water quality effects of urban tree systems scale to pipeshed- and watershed scales of relevance to stormwater managers. This knowledge gap is particularly vexing as it hinders decisions regarding the conditions under which trees can be effectively integrated with stormwater management efforts. Filling this gap likely requires more “opportunistic” analyses of available watershed discharge data with canopy cover patterns in urban and urbanizing watersheds, as well as studies designed to specifically target the influence of canopy cover through, for example, establishing before-after-control-impact watershed treatments and collecting new streamflow data as necessary. Isotopes or other environmental tracers could be integrated with streamflow measurements to provide an additional line of evidence for the hydrologic and water quality influence of urban tree systems upon receiving aquatic systems. Although not treated extensively in this review, hydrologic models will continue to play an important role in providing further insight to watershed-scale influences of urban tree systems. Mechanistic models seem preferable to empirical models to develop this insight; however, every model is subject to simplifying assumptions that are “largely untested in the field” (Berland et al., 2017). Among these simplifying assumptions are the use of constant interception and stemflow rates and a lack of consideration for some of the finer canopy architecture attributes known to affect canopy partitioning. Thus, future research should include collection of data to validate models and their underlying assumptions, as well as to determine when the tradeoff between more accurate models and their complexity is appropriate for stormwater management applications.

The need for understanding at the watershed scale does not preclude continued work at the tree- and plot scales. Indeed, studies at these scales are needed to inform understanding of the hydrologic and water quality mechanisms that drive observed watershed responses. Among the knowledge gaps that can be addressed, at least partially, at these smaller spatial scales are: how canopy partitioning processes in urban tree systems affect runoff peak timing and magnitude under a range of canopy architectures and climatic conditions; how tree canopy cover influences runoff generation over pervious and impervious surfaces; the specific mechanisms controlling stormwater infiltration into underlying soils; and how hydrologic connectivity between impervious surfaces and urban tree systems may influence nutrient export from urban stormwater. Ideally, these studies will be conducted in open canopied systems most representative of urban tree systems and, when possible, incorporate controls to better isolate the hydrologic and water quality effects of trees relative to alternative land covers (e.g., turf, grasses, or impervious).

To approach the potential role of urban tree systems in stormwater management from a purely biophysical standpoint, as was done herein, is, admittedly, an oversimplification of the system. Social, policy, and cultural drivers also play a role in the structure of urban tree systems, from the scale of individual trees at a site to the broader urban forest at the community and city scales. To date, there is not a clear understanding of how sociocultural factors affect the structure and management of tree assemblages (Berland and Hopton, 2014). Thus, future research should also consider the interaction among human biophysical factors upon urban tree systems and how this ultimately impacts the potential contribution of urban tree systems to stormwater quantity and quality regulation.

Page 81: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD
Page 82: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 67

APPENDIX B

Meta-analysis Supporting Information B.1 Influence of Other Factors on Event-Based Precipitation Partitioning As discussed in Section 3.2.1, regression models were developed to predict precipitation capture and/or throughfall for urban tree systems. The collection of studies from which data used in these regression models were extracted are summarized in Table B-1.

Table B-1. Studies Included in Precipitation Partitioning Meta-analysis. Citation Tree species Climate

Classification Temporal scale

Asadian and Weiler, 2009 Seudotsuga menziesii, Thuja plicata Continental Annual, event Attarod et al., 2015 Cupressus arizonica, Pinus eldarica Semi-arid Annual David et al., 2006 Quercus ilex Semi-arid Annual, event Hassan et al., 2017 Quercus ilex, Q. pyrenaica Semi-arid Annual, event Kaushal et al., 2017 Morus alba Mediterranean Annual Livesley et al., 2016 Eucalyptus nicholii, E. saligna Continental Annual, event Motahari et al., 2013 Pinus eldarica Semi-arid Annual Nytch et al., 2019 Albizia procera, Calophyllum antillanum Tropical Annual, event Park and Cameron, 2008 Acacia mangium, Gliricidia sepium, Guazuma

ulmifolia, Ochroma pyramidale, Pachira quinata Tropical Annual, event

Pereira et al., 2009 Quercus ilex Mediterranean Annual Sadeghi et al., 2016 Cupressus arizonica, Fraxinus rotundifolia, Pinus

eldarica, Robinia pseudoacaci Semi-arid Annual, event

Staelens et al., 2008 Fagus sylvatica Continental Annual Van Stan et al., 2015 Fagus grandifolia, Liriodendron tulipifera Continental Annual Véliz-Chávez et al., 2014 Ficus benjamina Semi-arid Annual, event Xiao and McPherson, 2011b Citrus limon, Ginko biloba, Liquidambar styraciflua Mediterranean Annual, event Xiao et al., 2000 Pyrus calleryana, Quercus suber Mediterranean Annual, event Zabret et al., 2018 Betula pendula, Pinus nigra Continental Annual

The majority of variability in observed data was explained by precipitation depth and tree evergreen-ness (i.e., evergreen versus deciduous for annual scale and evergreen versus deciduous leaf on versus deciduous leaf off periods for event scale). LAI was found to be the next best predictor and, even though it explained a small portion of observed variability in partitioning measurements relative to rainfall depth and tree leaf type, a set of secondary equations incorporating LAI was developed to allow model users to elucidate the role of dense versus sparse tree canopies in rainfall capture if they so choose. These models represent a balance between simplicity and explanatory power, and are not intended to suggest that other variables do not provide strong controls on precipitation fate in urban tree canopies. In contrast, unexplained variability in these models indicates that other factors do indeed play an important role in precipitation partitioning. At present, our ability to include these factors in a statistical model was constrained by data availability as a limited subset of precipitation partitioning studies reported specific meteorological conditions during rain events. However, analysis of this small subset of studies suggests that rainfall intensity (Figure B-1) and climate (Figure B-2) are important. With respect to intensity, which is the depth of rainfall per unit time, a threshold of around 4 mm hr-1 was observed at which rainfall capture and throughfall rates became relatively constant. We were only able to extract detailed intensity data from two studies (encompassing six tree species), and were not able to map

Page 83: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

68 The Water Research Foundation

these data to total precipitation depth as would be needed to incorporate and potentially strengthen the predictive equations presented in Section 3.2.1 or to adequately test the effects of other tree-specific variables.

Figure B-1. Relationship between Event Rainfall Throughfall or Capture and Mean Event Intensity.

Exponential decay (capture) and growth (throughfall) curves provided the best model for this relationship, explaining 25% of the observed variability in precipitation fate.

Figure B-2. Relationship between Event Rainfall Throughfall or Capture and Study Area Climate.

General climate regions are indicative of other important meteorological controls. For example, tropical regions (“Am”) are generally typified by intense rainfall and low evaporative demand, while evaporative demand is

generally high in arid (“Bs”) and Mediterranean (“Cs”) climates. The continental climate category (“Cf”) in this dataset was dominated by data from the Pacific northwest, which may be skewed toward lower intensity rainfall

events and, hence, high capture/low throughfall.

In addition to intensity, existing data suggest the importance of other dynamic meteorological variables such as wind speed and vapor pressure deficit in determining relative rates of throughfall and capture (e.g., Van Stan et al., 2015). While we were not able to tease these relationships out in the data complied herein, we did find that overall climate classification of each study location had an effect on throughfall and capture rates (Figure B-2). For example, precipitation partitioning studies conducted in tropical climates tended to have higher rates of throughfall/lower rates of capture than studies conducted across the same range in event size but in drier Mediterranean climates. Studies conducted in arid regions, which are also typified by high evaporative demand, displayed similar throughfall and

Page 84: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 69

capture rates to Mediterranean locations. Studies conducted in continental climate regions exhibited the highest rates of rainfall capture. However, this climate category was dominated by data collected in the Pacific Northwest, and so may be skewed by the effects of low intensity rainfall (for which this region is known) on precipitation partitioning.

B.2 Supporting Documentation for Transpiration Analysis Table B-2 provides a summary of studies meeting inclusion criteria for the urban tree transpiration meta-analysis.

Table B-2. Summary of Studies Included in the Tree Transpiration Meta-analysis. Citation Tree species

Climate Classification

Meteorological and/or tree trait

variables reported1

Temporal scale

Bartens et al., 2009 Fraxinus pennsylvanica, Quercus bicolor Continental DBH Growing

season Chen et al., 2012

Cedrus deodara, Euonymus bungeanus, Metasequoia glyptostroboides, Zelkova schneideriana Continental DBH, T, VPD, Rs

Growing season

David et al., 2006 Quercus ilex Mediterranean DBH, LAI, T, VPD, Rs

Growing season,

Daily Dawson, 1996 Acer saccharum Continental DBH Growing

season

Pataki et al., 2011; Litvak et al., 2017

Brachychiton discolor, B. populneus, Eucalyptus grandis, Ficus microcarpa, Gleditsia triacanthos, Jacaranda chelonian, J. mimosifolia, Koelreuteria paniculata, Lagerstroemia indica, Pinus canariensis, Platanus

hybrida, Platanus racemosa, Sequoia sempervirens, Ulmus parvifolia

Mediterranean DBH, T, VPD, Rs Growing season,

Daily

Peters et al., 2010

Fraxinus pennsylvanica, Julgans nigra, Picea species, Pinus species, Tilia Americana, Ulmus species Continental DBH, LAI, T, VPD, Rs

Growing season,

Daily Riikonen et al., 2016 Alnus species, Tilia americana Continental DBH, LAI, T, VPD, Rs

Growing season

Wang et al., 2012 Aesculus chinensis Continental DBH, LAI, T, VPD, Rs

Growing season

1Variables abbreviated as DBH (diameter at breast height), LAI (leaf area index), T (temperature), VPD (vapor pressure deficit), Rs (Solar radiation)

Each of the tree species listed in Table B-2 was classified according to xylem anatomy as differences in the size and distribution of pores through which water is transported within tree sapwood has been shown to control transpiration rates, particularly at high VPD. Wood anatomy was classified as:

• Conifer. Water transport occurs in small conduits known as tracheids, which are distributed relatively evenly throughout the sapwood. Although diffuse and ring porous trees also contain tracheids, the majority of water transport occurs within larger vessels.

• Diffuse porous. Vessels in which water is transported through the tree are distributed relatively evenly within each growth ring in terms of pore size. Examples of diffuse porous trees include maple, linden, and birch. Diffuse porous trees have been associated with higher water use because, whereas other trees tend to close their stomates at higher vapor pressure deficits, diffuse porous woods are able to continue transpiring (Litvak et al., 2012).

• Ring porous. Vessels pore sizes are unevenly distributed within each growth ring, with larger pores forming early in the growing season and decreasing in size to the outside of each growth ring. As reviewed by Peters et al., (2010) ring porous trees are more sensitive to cavitation and thereby

Page 85: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

70 The Water Research Foundation

mitigate this risk by down-regulating transpiration as evaporative demand increases by closing stomates. Examples include oak, elm and ash trees.

• Semi-ring porous. Pore size decreases gradually from the inside to outside of each wood growth rings, thus bearing some characteristics of ring porous wood.

• Brachichiton. Characteristic of Brachichitons, a drought-tolerant genus native to Australia (two of which were included in the dataset herein), the wood of these trees is considered separately as it consists of distinctive spongy sapwood with wide vessels.

Results of tree transpiration as averaged over the growing season (coinciding with the measurement period reported in the collection of studies analyzed herein) are summarized in the following points.

B.2.1 Effects of Meteorological Variables and Climate Tree transpiration was significantly correlated with environmental variables at study sites with a continental climate classification, but not at study sites characterized as Mediterranean. We note that, although approximately half of tree transpiration rates were measured in a Mediterranean climate (with irrigation to provide adequate water), these data points came from only two of the eight studies, with nearly all coming from a single study (Pataki et al., 2011). So this apparent difference in transpiration response by climate zone may actually be caused by some other study effect (e.g., methodological approaches for measuring transpiration and/or physiological drought adaptations of trees in the study). Notwithstanding, we developed prediction equations based only on climate factors since some potential users of these equations may lack more specific information about the trees in their system of interest. Prediction equations are presented in Table B-3 and illustrated in Figure B-3. Use of these equations should be limited to continental temperate climates (or applied within the range of vapor pressure deficit, solar radiation and temperature for which the equation was developed). As discussed in the following sections, accounting for the effects of tree characteristics improves transpiration predictions; therefore we would only recommend using this metrological-based equation if tree characteristics are not known.

B.2.2 Effects of Tree Functional Variables: Wood Anatomy Although they were not designed specifically to test for differences in water use among wood anatomies, two of the studies included in this dataset (Pataki et al., 2011; Peters et al., 2010) measured transpiration rates in trees with different xylem types exposed to similar environmental conditions, but did not observe differences in transpiration rates. In this meta-analysis, the effects of tree wood type on growing season average transpiration rates were somewhat mixed. Seasonal transpiration rates were not significantly different among conifer, ring porous or diffuse porous species (Figure B-4). The data do suggest, however, that different wood anatomies differ in their transpiration response across a range in environmental variables. This was demonstrated in correlation analyses between transpiration rates and individual environmental variables (Table B-3) as well as when interactions between environmental and tree factors were considered (Figure B-5 and B-6), though not all relationships were statistically significant. For example, ring-porous species exhibited a strong positive relationship between average transpiration rate and VPD while diffuse porous exhibited a negative (though insignificant) correlation. This result was the opposite of what we expected – physiologically, diffuse porous species are able to continue transpiration at high rates (provided adequate soil water) across a wide range in vapor pressure deficit as their xylem vessels are less susceptible to cavitation (Peters et al., 2010). Accordingly, higher rates of water use have been documented for diffuse porous species relative to their ring porous counterparts in natural forests (e.g., Ford et al., 2011; Taneda and Sperry, 2008). However, other studies have found associations between cavitation risk, transpiration regulation, and rooting depth with shallow rooting depths associated with greater vulnerability to cavitation (as reviewed by Peters et al., 2010). While factors such as this could be at play in our dataset, we also suspect that the loss in

Page 86: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 71

resolution that occurred by averaging transpiration rates (and associated environmental drivers) from a relatively small number of studies across seasonal time scales may also mask the influence of this functional tree characteristic. At the event-scale, differences in response to climatic variables, namely VPD, were also detected for different xylem anatomies.

B.2.3 Effects of Tree Structural Variables We also observed a relatively strong correlation between transpiration rate and leaf area index (LAI; Figure B-7). As reviewed by Wullschleger et al., (2001) such relationships have also been observed in natural forests, particularly under non-limiting water conditions. As noted previously, urban trees grown in isolated conditions have more exposure to climatic drivers of transpiration (e.g., solar radiation, wind, higher VPD) along their vertical profile. Therefore, a structural tree attribute such as LAI (which is the total one-sided leaf area divided by the ground area underlying the tree canopy) that indicates the vertical distribution of leaf area would be expected to correlate well with urban tree transpiration measurements. While transpiration rates were predicted relatively well with LAI as the only predictor variable (R2 = 0.74, indicating LAI described 74% of observed variability in transpiration rates as reported across the 11 data points for which LAI measurements were available), the fit was improved when temperature effects on transpiration were also added (R2 = 0.9; Figure B-8). In this model, increasing temperature exerted a negative influence on transpiration rate, and we believe this relationship indicates the role of temperature (and likely related environmental stressors such as vapor pressure deficit) as a constraint to transpiration. Within this limited dataset, we did not have an adequate number of different wood anatomies to include the effects of this functional tree attribute with LAI. Both LAI-based equations are presented in Table B-4. The primary caveat with the use of these equations is that they only represent transpiration rates from studies in which LAI was reported (about half of the transpiration dataset), the majority of which were conducted in relatively humid continental climates. Therefore, as discussed in Section 3.2.2, if the equations are to be applied to semi-arid or arid environments, we would recommend the use of the wood-based annual average and daily time scale equations as these are the most representative of climate zones included in the data set and include predictor variables that are both statistically significant and that “make sense” from a mechanistic/physiological standpoint.

B.2.3.1 Limitations of Transpiration Analysis Caveats associated with these seasonal-scale transpiration equations include they were developed for a relatively small sample size over a range in VPD and solar radiation characteristic of temperate climates (as opposed to semi-arid). Although tree transpiration rates in semi-arid climates were included in this analysis, we were not able to adequately predict transpiration based on available meteorological variables, and LAI was only reported for one tree within the semi-arid dataset, thus limiting exploration of LAI as a predictive factor across the higher temperatures, VPDs and solar radiation characteristic of semi-arid climes. The dataset was expanded substantially when daily-scale dynamics of transpiration and meteorological factors were considered. However, the data comprising this dataset came from three general locations in three studies (St. Paul, MN; Los Angeles, CA; Lisbon Portugal) and so may be more susceptible to within study biases than a larger dataset would be.

Page 87: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

72 The Water Research Foundation

Table B-3. Correlations between Average Annual Transpiration Rates and Tree, Climate Variables. Correlations are shown across all trees in the dataset (n = 35) and as separated by the general climate

characterization of their respective study sites. Explanatory variables averaged over the period by which transpiration measurements were reported (i.e., growing season).

Variable

Correlation coefficienta : with transpiration on canopy area basis By climate category (canopy areas basis) By xylem structure

All study sites (n=35)

Continental (n= 17)

Semi-arid (n=18)

Diffuse porous (n = 15)

Ring-porous (n = 10)

Conifer (n=8)

Diameter at breast height (cm) -0.17 -0.52** -0.13 -0.06 -0.16 -0.05 Leaf area index 0.71** 0.75** NAb 0.77* NAc 0.5 Temperature (C) -0.31* -0.68** -0.38 -0.36 0.67* -0.74** Vapor pressure deficit (kPa) -0.07 -0.025 -0.31 -0.13 0.86** -0.21 Solar radiation (W/m2) 0.27 0.47* 0.26 0.37 0.74** 0.11 aSpearman’s rho is a non-parametric correlation coefficient, used here since transpiration data were not normally distributed. It indicates the direction of the correlation, but not the shape (e.g., it does not assume a linear relationship); p-values less than 0.05 denoted with ** to indicate statistical significance of correlations. p-values between 0.05 and 0.1 denoted with *. bLAI values were only reported for 1 tree in a semi-arid climate; thus it was not possible to calculate a correlation coefficient. cLAI values only reported for 2 of the ring-porous trees in the dataset; thus a reliable correlation coefficient could not be determined.

Figure B-3. Tree Transpiration as Predicted by Meteorological Variables for All Data Points (n = 35; left) and within Continental Temperate Climates Only (n = 18; right) Compared to Measured Values.

Predictor variables included solar radiation, vapor pressure deficit (VPD) and temperature. Data were separated based on climate zone since ability to predict transpiration based on meteorological variables for trees in

Mediterranean climates was so poor. Transpiration rates reflect growing season averages (in mm/day/m2 canopy area) as reported across 8 studies in urban environments (see Table B-2 for citation information).

Page 88: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 73

Figure B-4. Box Plots Depicting Range in Growing Season Average Daily Transpiration Rates by Wood Anatomy Type Reported for Urban Trees.

Wood anatomies represented in the dataset included Brachichiton (Br, n = 2), conifer (C, n = 9), diffuse porous (DP, n = 15) and ring porous (RP, n = 9) while study sites were characterized as continental (n=6; closed circles) or

Mediterranean (n=2; open squares). Boxes represent 25%, median, and 75% data quantiles; bars indicate 10% and 90%. Overall mean indicated by gray line. Studies from which data extracted cited in Table B-2.

Figure B-5. Wood Anatomy Effects on Transpiration Response to Range of Meteorological Variables. In this dataset, average growing season transpiration by diffuse porous (DP) species exhibited a significantly greater positive response to increasing solar radiation and a greater (but statistically insignificant) negative

response to increasing temperature and vapor pressure deficit relative to conifers (C) and ring porous (RP). Solid markers indicate data points from temperate climates; open markers indicate data from semi-arid climates.

Studies from which data extracted cited in Table B-2.

Page 89: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

74 The Water Research Foundation

Table B-4. Predictive Equations for Tree Transpiration. Equations are presented according to the types of predictive variables required: meteorological only, wood

anatomy (based on tree species) + meteorological, and tree leaf area index (LAI). All predictive variables were significant at the p < 0.05 level. Data from which regression equations developed cited in Table B-2.

aR2 used to indicate portion of variability in transpiration explained by each predictive equation. Number of data points n included in the equation given in parentheses.

Figure B-6. Influence of Meteorological Variables on Transpiration Rates by Wood Type. 95% confidence intervals are also shown (red dashed lines) along with overall all mean transpiration (blue dashed

line). Solid markers indicate data points from temperate climates; open markers indicate data from semi-arid climates. Studies from which data presented in plots were extracted are cited in Table B-2.

Diffu

s

e porous T = 0.048*Rs – 0.43*T + 0.55

Ring porous T = 0.031*Rs + 0.21*T – 10.81 Conifer

T = 0.023*Rs – 3.20*VPD – 0.39

Equation type Predictive equation: average daily growing season transpiration, Tr Equation fita

Meteorological variables only (continental climates only) Tr = 0.036*Rs – 4.95*VPD – 0.20*T + 3.4 R2 = 0.77 (n=13)

Wood anatomy-based Diffuse porous: Tr = 0.048*Rs – 0.43*T + 0.55 Ring porous: Tr = 0.031*Rs + 0.21*T – 10.81

Conifer: Tr = 0.023*Rs – 3.20*VPD – 0.39

R2 = 0.58 (n = 12) R2 = 0.79 (n = 8) R2 = 0.58 (n = 8)

LAI-based Tr = 0.22*LAI + 0.31 Tr = 0.26*LAI – 0.16T + 3.16

R2 = 0.74 (n=11) R2 = 0.90 (n=11)

Equation terms: Transpiration (Tr) is in mm/day/m2 canopy area; Rs is average daily solar radiation in W/m2; VPD is average daily vapor pressure deficit in kPa; T is temperature in degrees Celsius; LAI is the leaf area index of tree(s), ideally as averaged over the period of interest. Meteorological-based equation appropriate for solar radiation values ranging from 140 W/m2 to 260 W/m2 and VPD < 1.2 kPa.

Equation type Predictive equation: daily tree water use Tr Equation fita

Daily water use (mm d-1) Diffuse porous: Tr = 0.0003*Rs +0.37*VPD

Ring porous: Tr = 0.006*Rs - 0.24*VPD + 0.38 Conifer: Tr = 0.001*Rs + 0.68*VPD

R2 = 0.28 (n = 235) R2 = 0.24 (n = 313) R2 = 0.40 (n = 241)

Equation terms: Transpiration (Tr) is in mm/day/m2 canopy area; Rs is average daily solar radiation in W/m2; VPD is average daily vapor pressure deficit in kPa; T is temperature in degrees Celsius; Equations most appropriate for solar radiation values ranging

from 40 W/m2 to 350 W/m2 and VPD up to 3 kPa.

Page 90: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 75

Figure B-7. Relationship between Tree Transpiration Rates and LAI. 95% confidence intervals are indicated by red dashed lines; overall mean transpiration rate shown by blue

horizontal line. Marker colors indicate wood anatomy: ring porous (blue), diffuse porous (green), conifer (red) ; open and closed symbols indicate study conducted in semi-arid or temperate climate, respectively. Studies from

which data presented in plots were extracted are cited in Table B-2.

Figure B-8. Relationship between Tree Transpiration Rates, LAI and Temperature. 95% confidence intervals are also shown (red dashed lines) along with overall all mean transpiration rate (blue horizontal line). Marker colors indicate wood anatomy: ring porous (blue), diffuse porous (green), conifer (red); open and closed symbols indicate study conducted in semi-arid or temperate climate, respectively. Studies from

which data presented in plots were extracted are cited in Table B-2.

Page 91: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

76 The Water Research Foundation

B.3 Water Quality Analysis Supporting Documentation Studies from which water quality-related data were extracted are summarized in Table B-5.

Table B-5. Summary of Studies Reporting Nutrient Data Associated with Urban Tree Systems. Citation Process Tree system type Nitrogen forms

reporteda Phosphorus forms

reportedb Bettez & Groffman, 2013

Canopy washoff / leaching

Urban and suburban deciduous forests NH4, NO3 --

Chiwa et al., 2003 Suburban evergreen forests NH4, NO3 -- Decina et al., 2018 Urban deciduous forests NH4, NO3, Total SRP, TP Fang et al., 2011 Urban and suburban deciduous forests;

Urban and suburban evergreen forests NH4, NO3 --

Juknys et al, 2007 Urban and suburban evergreen forests NH4, NO3 -- Izquita-Rojano et al., 2016

Suburban evergreen forests NH4, NO3, TN --

Kimura et al., 2009 Suburban deciduous and evergreen forests NH4, NO3 -- Michopoulos et al., 2007

Urban evergreen forests NO3 --

Neal et al., 2003 Suburban evergreen forests -- SRP Ponette-Gonzalez et al., 2017

Urban deciduous forests NH4, NO3 --

Tulloss & Cadenesso, 2015

Urban deciduous forests NH4, NO3 --

VanStan et al., 2012 Suburban deciduous forests NO3 -- Xiao & McPherson, 2011

Urban deciduous trees NH4, NO3, TN SRP, TP

Zhang, 1999 Suburban deciduous trees NH4, NO3 TP Bratt et al., 2017

Urban tree litter decomposition

Urban deciduous trees (A. saccharum; Q. palustris)

TN TP

Enloe et al., 2015 Urban deciduous (Quercus sp.) and evergreen (Pinus sp.) forests

TN TP

Fu et al., 2018 Urban deciduous (Ginko) TN TP Hobbie et al., 2014 Urban deciduous trees (A. platanoides, F.

pennsylvanica, Quercus sp., Tilia) TN TP

Hutmacher et al., 2015 Urban riparian forest (Parkinsonia microphylla, Acacia greggii)

TN --

Nikula et al., 2010 Urban Populus tremula forest TN -- Pavao-Zuckerman & Coleman, 2005

Urban forest (Quercus sp.) TN --

Pouyat & Carreiro, 2003 Urban forest (Quercus sp.) TN -- Sun & Zhao, 2016 Urban deciduous (Robinia pseudoacacia) and

evergreen (Pinus sp.) forests TN TP

Groffman et al., 2009 Belowground processes (nutrient

leaching studies)

Urban remnant deciduous forests; fertilized turfgrass

NO3 --

Amador et al., 2007 Urban evergreen and deciduous trees; fertilized turfgrass

NO3 --

Groffman et al, 2006 Urban remnant deciduous forests; unfertilized turfgrass

NO3 --

Nidzgorski & Hobbie, 2016

Urban evergreen and deciduous trees; unfertilized turfgrass

NO3, TDN SRP, TDP

Erickson et al., 2005; 2008

Urban fertilized turfgrass NO3 TDP

Qin et al., 2013 Urban fertilized turfgrass NO3, TDN SRP Strahm et al., 2005 Evergreen forest NO3, TDN anitrogen forms include ammonium (NH4), nitrate (NO3), total dissolved nitrogen (TDN) and total nitrogen (TN) bphosphorus forms include soluble reactive phosphorus (SRP), total dissolved phosphorus (TDP) and total phosphorus (TP)

Page 92: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 77

Nutrient mass dynamics were highly variable across and within studies included in the litter decomposition dataset. In an attempt to understand these dynamics and provide guidance to stormwater managers to better predict them, we examined nutrient mass losses for individual species and as categorized by litter nutrient or other tree traits. On a species basis, exponential decay functions described nitrogen mass losses for 6 of 10 tree species and phosphorus losses for 5 of 7 species at a significance level of 0.1 or less (Table B-6). On average, phosphorus loss rates (1.59 yr-1) were higher than the average rates of nitrogen (0.65 yr-1) or total leaf mass (0.90 yr-1) loss, and exhibited greater variation across species. It was observed that P mass losses were not constant through time, but followed a rapid loss period immediately following litter fall. Table B-7 presents exponential decay constants determined by species for a “quick” and “slow” P mass change periods. A similar analysis was conducted for nitrogen mass (Table B-8), which displayed the opposite behavior from P.

Table B-6. Mass Decay Rate Constants k for Litter Dry Mass, Total Nitrogen (TN) and Total Phosphorus (TP) Fit to Individual Tree Species.

The proportion of data variability explained by the exponential decay function is indicated by the R2 value. Bold values indicate the relationship was significant at the 0.05 level; regular black font indicates significance at the 0.1

level; gray font were not statistically significant. Studies included in analysis cited in Table B-5. Speciesa Dry mass decay Nitrogen decay Phosphorus decay

k p-value R2 k p-value R2 k p-value R2 Acacia greggii 0.27 0.002 0.93 0.07 0.2 0.37 -- -- --

Acer platanoides 1.5 0.05 0.7 0.36 0.0094 0.39 0.07 0.6 0.02 Acer sacharum -1.58b 0.004 0.78 -0.45b 0.17 0.29 5.6 0.012 0.66

Fraxinus pennsylvanica 1.23 0.05 0.8 0.53 0.049 0.25 0.39 0.08 0.19 Ginkgo biloba 1.28 0.02 0.76 1.72 0.05 0.66 3.2 0.23 0.33

Parkinsonia microphylla 0.48 0.03 0.75 0.53 0.026 0.75 -- -- -- Pinus species 0.27 <0.0001 0.88 0.2 0.44 0.03 0.52 0.02 0.25

Populus tremula 0.6 0.06 0.99 0.4 0.17 0.93 -- -- -- Quercus species 0.44 <0.0001 0.77 0.2 0.013 0.13 0.31 0.1 0.08

Robinia pseudoacacia 0.28 0.0001 0.85 0.3 0.1 0.30 -0.25 0.26 0.14 Tilia americana 1.3 0.05 0.9 0.56 0.0094 0.39 0.99 0.003 0.47

Averagec 0.55 0.6 1.56 aAll species datasets were obtained from a single study with the exception of Pinus (pine) and Quercus (oak), which represent complied loss rates reported in 2 and 4 studies, respectively. bnegative value indicates mass gain. The study from which these data were obtained was conducted in a street gutter during winter over a short time period, which may explain this otherwise uncommon result (Bratt et al., 2017). caverage values only include those k values significant at the 0.1 level or less.

Page 93: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

78 The Water Research Foundation

Table B-7. Phosphorus Mass Change by Species for Full Monitoring Period, “Quick” Period (< 0.2 years) and “Slow” Period (0.2 to 1.5 years).

The majority of species exhibited phosphorus mass loss (as indicated by a positive value of k) over a period of a few weeks following deposition (i.e., “quick” leaching period); however, this relationship was significant only for 2 species, likely due to relatively small number of data points when data were divided by species and time period. P

dynamics over longer time periods (i.e., > 0.2 years after leaf fall) were more variable, with some species exhibiting P mass gain (negative k values) and some loss, but relationships between time and P mass change were not

significant. Studies included in analysis cited in Table B-5.

Species Phosphorus mass change (time 0 to

1.5 years) P "quick" period (time < 0.2

years) P "slow" period (time 0.2 to 1.5

years) k (yr-1) p-value R2 k (yr-1) p-value R2 k (yr-1) p-value R2

Acer platanoides 0.07 0.6 0.02 2.18 0.28 0.36 -0.11 0.64 0.03

Acer sacharum 5.6 0.012 0.66 8.39 0.42 0.33 3.17 0.22 0.6

Fraxinus pennsylvanica 0.39 0.08 0.19 5.7 0.009 0.92 -0.07 0.82 0.006

Ginkgo biloba 3.2 0.23 0.33 16.7 0.4 0.65 -0.63 0.79 0.11

Pinus species 0.52 0.02 0.25 -1.36 0.33 0.19 0.66 0.07 0.25

Quercus species 0.31 0.1 0.08 5 0.005 0.47 0.18 0.57 0.02

Robinia pseudoacacia -0.25 0.26 0.14 1.62 0.68 0.23 -0.45 0.14 0.33

Tilia americana 0.99 0.003 0.47 3.88 0.33 0.31 0.83 0.11 0.26

Page 94: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 79

Table B-8. Nitrogen Mass Change by Species for Full Monitoring Period, “Quick” Period (< 0.2 years) and “Slow” Period (0.2 up to 2.5 years).

The majority of species did not exhibit nitrogen mass losses (as indicated by positive values of k) until litter had been on the ground over 10 weeks, while N mass gain tended to be more common within the first 10 weeks (0.2 years) of litter fall (as indicated by negative values of k). However, none of these relationships were significant,

likely due to the relatively small number of datapoints when data were divided by species and time period. Studies included in analysis cited in Table B-5.

Species Nitrogen mass change - full

dataset N "quick" period (time < 0.2

years) N "slow" period (time 0.2 to 1.5

years) k (yr-1) p-value R2 k (yr-1) p-value R2 k (yr-1) p-value R2

Acacia greggii 0.07 0.2 0.37 -- -- -- 0.03 0.07 0.66

Acer platanoides 0.36 0.0094 0.39 -0.73 0.3 0.34 0.32 0.18 0.19

Acer sacharum -0.45a 0.17 0.29 -0.23 0.87 0.02 0 -- --

Fraxinus pennsylvanica 0.53 0.049 0.25 -0.02 0.99 0 0.76 0.13 0.23

Ginkgo biloba 1.72 0.05 0.66 5.3 0.36 0.71 0.86 0.4 0.65

Parkinsonia microphylla 0.53 0.026 0.75 -- -- -- 0.53 0.12 0.61

Pinus species 0.2 0.44 0.03 -1.66 0.63 0.05 0.64 0.05 0.29

Populus tremula 0.4 0.17 0.93 -- -- -- 0.58 -- --

Quercus species 0.2 0.013 0.13 1.1 0.2 0.1 0.23 0.05 0.13

Robinia pseudoacacia 0.3 0.1 0.3 -0.33 0.85 0.05 0.61 0.006 0.74

Tilia americana 0.56 0.0094 0.39 -0.28 0.87 0.01 0.85 0.03 0.44

Figure B-9. TN and TP Mass Losses by Tree Species. Tree species are indicated by number: Ginkgo biloba (1); Acer saccharum (2); Tilia americana (3); Fraxinus

pennsylvanica (4); Pinus sp. (5); Quercus sp. (6); Acer platanoides (7); Populus tremula (8); Parkinsonia microphylla (9); Acacia greggii (10). Dashed fit lines and gray font indicate exponential decay function fit was not significant.

The positive slope of fit lines represents indicates net nutrient mass loss through time (Equation 3-1). Studies included in analysis cited in Table B-5.

Page 95: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD
Page 96: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 81

APPENDIX C

iTree-Hydro Model Calibration and Performance Assessment Input files to i-Tree Hydro include a digital elevation model of the watershed, daily meteorological data (including precipitation, temperature, and evapotranspiration rates) and a combination of physical and semi-empirical watershed hydrologic parameters used to model the processes by which precipitation is infiltrated, evapotranspired, is translated to surface runoff or is recharged (Table C-1). While values to some of these parameters can be estimated (for example, soil infiltration parameters can be estimated based on soil types and/or infiltration testing in the watershed, while leaf on and leaf off periods can be determined from leaf bud and fall observations from the study area), others are determined through the calibration process (for example, coefficients for soil surface and subsurface routing).

To calibrate the model, streamflow and weather data measured in each watershed from March to October 2013 were input to the model. Model parameters were then adjusted, first using the auto-calibration feature built in to i-Tree Hydro and then by fine tuning individual parameters to ensure realistic values were obtained, so that the streamflow predicted by the model matched (as closely as possible) the streamflow that was measured at the watershed outlet for a given set of precipitation and other meteorological conditions. The set of parameter values obtained through the calibration process with 2013 data were then applied to predict streamflow runoff given weather data measured during 2011 as a further assessment of model performance. For both the calibration and assessment periods, model performance was characterized using the Nash-Sutcliffe Efficiency (NSE) ratio and visual inspection of residual plots and streamflow hydrographs. The model was calibrated and assessed across a range of frequent (small, 25 mm and less) and more infrequent (large, up to 100 mm) storms. In general, residual plots were randomly distributed, indicating that model errors were random as opposed to being biased toward over or under prediction, which would likely indicate a shortcoming in representation of a physical runoff process(es) in the model. NSE coefficients for selected storm events (event basis) and overall calibration period (continuous basis) in the three study and reference watersheds are presented in Table C-2. An NSE value of 1 indicates a perfect match between predicted and observed streamflow values, while values less than 0 indicate the fit is rather awful. By convention, NSE values greater than 0.5 are generally considered acceptable for hydrologic models (Engel et al., 2007). The models created herein tend to have higher NSE values (indicating better fit) for larger storms, though model performance was acceptable or near-to-acceptable for smaller storms (e.g., 10 mm to 25 mm as observed on 8/6/2013 and 9/1/2013. Model performance in two of the study watersheds (Urb2 and Urb3) was quite poor for a runoff event recorded on 10/30/2013 that consisted of multiple rainfall (and stormflow) peaks. As indicated in Figures C-1 through C-3, the lower NSE values obtained through the calibration process result from models either missing the timing and/or magnitude of the runoff peak. Despite poor fits between predicted and observed streamflow for some of the individual events, the overall NSE for the three developed watersheds approached an acceptable level with values of 0.66, 0.52 and 0.58 in model watersheds Urb1, Urb2, and Urb3, respectively (Table C-2).

Page 97: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

82 The Water Research Foundation

Table C-1. Model Watershed Hydrologic Parameters and Values Obtained through Calibration Process. Model Parameters Urb1 Urb2 Urb3 Ref

Annual Average Flow of Project Area (cms): 0.582 0.045 0.015 0.003 Soil characteristics Wetting Front Suction (m): 0.34 0.07 0.03 .39 Wetted Moisture Content (m): 0.111 0.297 0.7 0.7 Surface Hydraulic Conductivity (cm/h): 0.194 0.4 0.765 2.5 Depth of Root Zone (m): 0.1 0.5 0.5 0.97 Initial Soil Saturation Condition (%): 65 9.257 10.22 15 Vegetation and landcover parameters Leaf Transition Period (days): 28 28 28 28 Leaf On Day (Day of year 1-365)a: 97 97 97 97 Leaf Off Day (Day of year 1-365)a: 297 297 297 297 Tree Bark Area Index: 1.5 1.7 1.7 1.7 Shrub Bark Area Index 0.5 0.5 0.5 0.5 Leaf Storage (mm): 0.1 0.1 0.1 0.1 Pervious Depression Storage (mm): 0.6 0.1 3.0 3.0 Impervious Depression Storage (mm): 0.955 0.102 2 1 % directly connected impervious area 40 55 40 0 % canopy cover over impervious area 10 15 19 1 % canopy cover over pervious area 90 85 81 99 Surface and subsurface routing parameters Scale Parameter of Power Function: 2 2 2 2 Scale Parameter of Soil Transmissivity: 0.023 0.023 0.91 0.03

Transmissivity at Saturation (m^2/h): 0.12 0.24 3.27 0.1 Unsaturated Zone Time Delay (h): 10 39.9 129.8 118 Soil Macropore Percentage (%): 0 0 0 0 Soil Surface Flow Routing: B (h): 0.77 1.28 0.65 1.0 Time Constant for Surface Flow: Alpha (h): 106 239 124 143

Time Constant for Surface Flow: Beta (h): 0.1 0.4 0.49 0.3 Watershed Area Where Rainfall Rate can Exceed Infiltration Rate (%): 90 41 67 70

aLeaf on and off day of year based on observations from Carter (2015)

Table C-2. Nash-Sutcliffe Efficiency (NSE) Ratios for Model Calibration and Assessment Periods. Selected precipitation

events, 2013 Rainfall depth (mm) Nash-Sutcliffe Efficiency value

Urb1 Urb2 Urb3 Urb1 Urb2 Urb3

5/26 to 5/28 51.8 34.5 48.0 0.544 0.880 0.860 6/14 to 6/16 38.6 25 50.0 0.895 0.832 0.600

8/5 to 8/7 27.4 39.6 30.0 0.708 0.460 0.540 8/31 to 9/2 10.2 25.4 14.2 0.380 0.546 0.690

9/18 to 9/20 51.8 61 30.0 0.807 0.966 0.690 10/29 to 10/31 78.2 44.7 105.2 0.647 -0.560 0.120

Overall calibration period 0.66 0.52 0.58 Overall assessment period 0.71 0.54 0.48

Page 98: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 83

Figure C-1. Urb1 - Example Hydrographs Comparing Observed Streamflow in m3 (black line) and Total Flow Predicted by i-Tree Hydro with Parameter Values in Table C-1 (red line).

Corresponding precipitation measurements are shown along the right y-axis (mm/hr) in blue. Streamflow and precipitation observations were obtained at the watershed outlet from the Johnson County Stormwatch sensor

network.

Page 99: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

84 The Water Research Foundation

Figure C-2. Urb2 - Example Hydrographs Comparing Observed Streamflow in m3 (black line) and Total Flow Predicted by i-Tree hydro with Parameter Values in Table C-1 (red line).

Corresponding precipitation measurements are shown along the right y-axis (mm/hr) in blue. Streamflow and precipitation observations were obtained at the watershed outlet from the Johnson County Stormwatch sensor

network.

Page 100: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 85

Figure C-3. Urb3 - Example Hydrographs Comparing Observed Streamflow in m3 (black line) and Total Flow Predicted by i-Tree Hydro with Parameter Values in Table C-1 (red line).

Corresponding precipitation measurements are shown along the right y-axis (mm/hr) in blue. Streamflow and precipitation observations were obtained at the watershed outlet from the Johnson County Stormwatch sensor

network.

Page 101: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD
Page 102: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 87

APPENDIX D

Stormwater and Forestry Survey: Responses from Stormwater Utility Managers and Municipal Arborists

Page 103: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

88 The Water Research Foundation

Stormwater Utilities Survey Report Incorporating Forestry into Stormwater Management Programs July 16th 2018, 1:22 pm MDT

Q1 - Does your municipality account for the effect of existing wooded areas and other landscape trees when planning for stormwater quantity and quality issues?

# Field Minimum Maximum Mean Std Deviation Variance Count

1

Does your municipality account for the effect of existing wooded areas and other landscape trees

when planning for stormwater quantity and quality issues?

1.00 2.00 1.39 0.49 0.24 51

# Answer % Count

1 Yes 60.78% 31

2 No 39.22% 20

Total 100% 51

Page 104: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 89

Q2 - Does your municipality actively promote the protection of existing trees or the establishment and management of new wooded areas or other landscape trees for their effect on stormwater quantity and quality issues?

# Field Minimum Maximum Mean Std Deviation Variance Count

1

Does your municipality actively promote the protection of existing trees or the establishment and

management of new wooded areas or other landscape trees for their effect on stormwater quantity and

quality issues?

1.00 2.00 1.37 0.48 0.23 52

# Answer % Count

1 Yes 63.46% 33

2 No 36.54% 19

Total 100% 52

Page 105: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

90 The Water Research Foundation

Q3 - Does your department incorporate trees into engineered stormwater structures?

# Field Minimum Maximum Mean Std Deviation Variance Count

1 Does your department incorporate trees into engineered stormwater structures? 1.00 2.00 1.40 0.49 0.24 52

# Answer % Count

1 Yes 59.62% 31

2 No 40.38% 21

Total 100% 52

Q4 - Does your municipality have regulations regarding trees in riparian (stream side) set-backs?

Page 106: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 91

# Field Minimum Maximum Mean Std Deviation Variance Count

1 Does your municipality have regulations regarding trees in riparian (stream side) set-backs? 1.00 2.00 1.43 0.50 0.25 51

# Answer % Count

1 Yes 56.86% 29

2 No 43.14% 22

Total 100% 51

Q5 - Does your municipality have regulations regarding trees in flood plains?

# Field Minimum Maximum Mean Std Deviation Variance Count

1 Does your municipality have regulations regarding trees in flood plains? 1.00 2.00 1.67 0.47 0.22 51

# Answer % Count

1 Yes 33.33% 17

2 No 66.67% 34

Total 100% 51

Page 107: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

92 The Water Research Foundation

Q6 - Do staff in your department collaborate with the city forester or municipal arborist on tree management issues?

# Field Minimum Maximum Mean Std Deviation Variance Count

1 Do staff in your department collaborate with the city

forester or municipal arborist on tree management issues?

1.00 2.00 1.33 0.47 0.22 52

# Answer % Count

1 Yes 67.31% 35

2 No 32.69% 17

Total 100% 52

Page 108: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 93

Q7 - When planning a new stormwater project, do you consider the projected amount of tree canopy or wooded areas in the development watershed?

# Field Minimum Maximum Mean Std Deviation Variance Count

1 When planning a new stormwater project, do you consider the projected amount of tree canopy or

wooded areas in the development watershed? 1.00 2.00 1.56 0.50 0.25 48

# Answer % Count

1 Yes 43.75% 21

2 No 56.25% 27

Total 100% 48

Q7a - If yes, what methods or models do you utilize to estimate the influence of various land uses, including tree canopy or wooded areas?

If yes, what methods or models do you utilize to estimate the influence of various land uses, including tree canopy or wooded areas?

VGIN Land Cover Layer

GIS-based canopy coverage data See appendix N: https://doee.dc.gov/sites/default/files/dc/sites/ddoe/page_content/attachments/FinalGuidebook_changes%20accepted_Appendices%20A-U_07_29_2013_compressed.pdf

Tree canopy reflected in the runoff curve number for areas with "enough" trees; individual plantings are not counted.

Typically we measure caliper of existing trees and the canopy, as well as protection for existing trees that are able to remain undisturbed, I do not perform hydraulic modeling so I am unsure if this information is reflected in project modeling. SCS runoff curve number

Page 109: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

94 The Water Research Foundation

Q8 - If offered free of charge, how likely would you be to participate in an educational module (read a bulletin for 20 minutes, view a webinar for 1 hour) on the effect trees and wooded areas have on stormwater quantity and quality?

# Field Minimum Maximum Mean Std Deviation Variance Count

1

If offered free of charge, how likely would you be to participate in an educational module (read a bulletin

for 20 minutes, view a webinar for 1 hour) on the effect trees and wooded areas have on stormwater

quantity and quality?

1.00 5.00 1.79 0.97 0.93 39

# Answer % Count

1 Extremely likely 46.15% 18

2 Somewhat likely 38.46% 15

3 Neither likely nor unlikely 7.69% 3

4 Somewhat unlikely 5.13% 2

5 Extremely unlikely 2.56% 1

Total 100% 39

Page 110: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 95

Q9 - What locations of trees do you consider when assessing stormwater effects? Check all that apply

# Answer % Count

1 Street trees 16.19% 17

2 Trees on private property 11.43% 12

3 Park trees with mown grass groundcover 11.43% 12

4 Trees in riparian (streamside) areas 23.81% 25

5 Natural wooded areas with understory plants and a ground cover of fallen leaves 20.95% 22

6 Trees in floodplains 16.19% 17

Total 100% 105

Page 111: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

96 The Water Research Foundation

Q10 - What values and benefits do you attribute to the location of trees above? Check all that apply

Page 112: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 97

# Question

Reduce stormwater runoff quantity

Increase soil

infiltration rates

Improve water

quality of stormwater runoff

Erosion

reduction

Carbon

sequestration

Air quality

improvement

Shading and

summer temperature reduction of streams

Summer air temperature reduction

Total

1 Street trees 13.16% 2

0 12.50% 19 11.84% 1

8 8.55% 13 11.84% 1

8 15.13% 23 9.21% 1

4 17.76% 27 152

2 Trees on

private property

15.24% 25 10.98% 1

8 12.80% 21 14.63% 2

4 10.98% 18 13.41% 2

2 6.71% 11 15.24% 2

5 164

3

Park trees with mown

grass groundcov

er

12.58% 19 11.92% 1

8 10.60% 16 11.92% 1

8 11.92% 18 15.23% 2

3 9.93% 15 15.89% 2

4 151

4

Trees in riparian

(streamside areas)

11.36% 20 10.80% 1

9 13.07% 23 15.34% 2

7 11.36% 20 12.50% 2

2 14.77% 26 10.80% 1

9 176

5

Natural wooded

areas with understory plants and

a ground cover of

and fallen leaves

13.25% 22 14.46% 2

4 12.05% 20 12.65% 2

1 11.45% 19 12.65% 2

1 10.84% 18 12.65% 2

1 166

6 Trees in floodplains 13.86% 2

3 12.05% 20 13.86% 2

3 13.86% 23 11.45% 1

9 11.45% 19 12.65% 2

1 10.84% 18 166

Page 113: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

98 The Water Research Foundation

Q11 - What regulations, policies or programs are actively used in your municipality or state to promote tree planting and retention of tree cover?

What regulations, policies or programs are actively used in your municipality or state to promote tree planting and retention of tree cover?

Minimum tree canopy in zoning ordinance

Tree credits, street tree policy, design guidelines, tree and sidewalk program.

Native plant protections, sod limitations, native plant requirements for new development, Tree City, Vegetation Advisory Committee

Hillside planting to stabillise slips and sub-surface erosion ("under-runners" in loess)

Statewide natural areas have been recognizes as significant for stormwater infrastructure, and conservation and management are approved for stormwater funding investments. There are 88 municipalities in LA County and locally many of them have protections for certain species with significant calipers, but waivers to remove them are easy to secure anyway. There are also periodically funding allocations for street tree programs in certain cities, and parks and rec and street tree services.

Zoning Regulations (green area ratio and tree overlay districts), stormwater regulations, "special" tree laws, MS4 permit

Parks department has a tree mitigation for trees removed in park areas, ratio changes depending on site & type

No specific regulations

Large trees must be replaced. Looking at managing forests for groundwater recharge

Executive Order by the Mayor applicable to projects undertaken by the city. There is a Branch Out Columbus program to promote the planting of 300,000 new trees. NPDES permit requirements, TMDLs, Portland Watershed Management Plan, Urban Forest Management Plan, climate change action and prepreparation plans, Portland Plan, Central City 2035 Plan, and others Phase II Level 4 stormwater permit (Creek Buffer zones, Riparian Buffers, Land Disturbance permit, ESC BMPs, PC BMPs, etc.); Texas Water Law (runoff quantity); Tree City USA ordinance;

Tree ordinance, tree protection plans, street tree rebate program

Tree Ordinance

Tree protection & mitigation regulations within my City's Unified Development Code

impervious cover drainage fee, environmentally sensitive area code, tree code

Local Climate Action Plan; No Fee Street Tree Replacement Permits; Municipal Land Development Code

Agencies provide minimal benefit to using trees in this way. Most of the use of trees is because private parties desire trees. Technically, they should be considered as a source of runoff reduction due to evapotranspiration. I work a lot with Golf Courses where their ET is a major aspect of my design work.

Currently climate action planning is the driving force behind tree preservation and new planting

Municipal: Riparian Action Program, Urban Forest Strategic Plans, BiodiverCity Strategy; Provincial Watershed Restoration and Resiliency Program, etc.

local tree advisory board, tree ordinances, timber harvest regulations

Page 114: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 99

Q12 - 50 States, D.C. and Puerto Rico

# Field Minimum Maximum Mean Std Deviation Variance Count 1 50 States, D.C. and Puerto Rico 5.00 53.00 24.43 16.95 287.43 37

# Answer % Count

1 Alabama 0.00% 0

2 Alaska 0.00% 0

3 Arizona 0.00% 0

4 Arkansas 0.00% 0

5 California 16.22% 6

6 Colorado 5.41% 2

7 Connecticut 0.00% 0

8 Delaware 0.00% 0

9 District of Columbia 2.70% 1

10 Florida 16.22% 6

11 Georgia 0.00% 0

12 Hawaii 0.00% 0

13 Idaho 0.00% 0

14 Illinois 2.70% 1

15 Indiana 0.00% 0

16 Iowa 0.00% 0

17 Kansas 2.70% 1

18 Kentucky 0.00% 0

19 Louisiana 0.00% 0

20 Maine 0.00% 0

21 Maryland 2.70% 1

22 Massachusetts 0.00% 0

23 Michigan 5.41% 2

24 Minnesota 2.70% 1

25 Mississippi 0.00% 0

Page 115: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

100 The Water Research Foundation

26 Missouri 2.70% 1

27 Montana 0.00% 0

28 Nebraska 2.70% 1

29 Nevada 0.00% 0

30 New Hampshire 0.00% 0

31 New Jersey 0.00% 0

32 New Mexico 0.00% 0

33 New York 5.41% 2

34 North Carolina 0.00% 0

35 North Dakota 0.00% 0

36 Ohio 2.70% 1

37 Oklahoma 0.00% 0

38 Oregon 2.70% 1

39 Pennsylvania 0.00% 0

40 Puerto Rico 0.00% 0

41 Rhode Island 0.00% 0

42 South Carolina 0.00% 0

43 South Dakota 0.00% 0

44 Tennessee 0.00% 0

45 Texas 16.22% 6

46 Utah 0.00% 0

47 Vermont 0.00% 0

48 Virginia 2.70% 1

49 Washington 0.00% 0

50 West Virginia 0.00% 0

51 Wisconsin 0.00% 0

52 Wyoming 0.00% 0

53 I do not reside in the United States 8.11% 3

Total 100% 37

Page 116: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 101

Q13 - What is the population size category of your municipality?

Page 117: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

102 The Water Research Foundation

Municipal Arborists Survey Report Incorporating Urban Forestry into Stormwater Management Programs July 16th 2018, 1:25 pm MDT

Q1 - Does your municipality account for the effect of existing wooded areas and other landscape trees when planning for stormwater quantity and quality issues?

# Answer % Count

5 Yes 22.22% 2

6 Maybe 44.44% 4

7 No 33.33% 3

Total 100% 9

Page 118: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 103

Q2 - Does your municipality actively promote the protection of existing trees or the establishment and management of new wooded areas or other landscape trees?

# Answer % Count

1 Yes 88.89% 8

2 No 11.11% 1

Total 100% 9

Q3 - Does your department incorporate trees into engineered stormwater structures?

Page 119: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

104 The Water Research Foundation

# Answer % Count

5 Yes 44.44% 4

6 Maybe 33.33% 3

7 No 22.22% 2

Total 100% 9

Q4 - Does your municipality have regulations regarding trees in riparian (stream side) set-backs?

# Answer % Count

1 Yes 77.78% 7

2 No 22.22% 2

Total 100% 9

Page 120: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 105

Q5 - Does your municipality have regulations regarding trees in flood plains?

# Answer % Count

1 Yes 66.67% 6

2 No 33.33% 3

Total 100% 9

Q6 - Do staff in your department collaborate with the stormwater utility department on tree management issues?

Page 121: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

106 The Water Research Foundation

# Answer % Count

1 Yes 88.89% 8

2 No 11.11% 1

Total 100% 9

Q7 - When your municipality is planning a new stormwater project, is your department consulted by the planning or stormwater departments about the projected amount of tree canopy or wooded areas in the development watershed?

# Answer % Count

4 Yes 11.11% 1

5 Maybe 77.78% 7

6 No 11.11% 1

Total 100% 9

Q7a - If yes, what methods or models do you utilize to estimate the influence of various land uses, including tree canopy or wooded areas?

If yes, what methods or models do you utilize to estimate the influence of various land uses, including tree canopy or wooded areas?

We use the 2011 City of Renton Tree Canopy Assessment Report

Urban Tree Canopy assessment using satellite imagery and LIDAR.

Page 122: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 107

Q8 - If offered free of charge, how likely would you be to participate in an educational module (read a bulletin for 20 minutes, view a webinar for 1 hour) on the effect trees and wooded areas have on stormwater quantity and quality?

# Answer % Count

1 Extremely likely 42.86% 3

2 Somewhat likely 28.57% 2

3 Neither likely nor unlikely 14.29% 1

4 Somewhat unlikely 14.29% 1

5 Extremely unlikely 0.00% 0

Total 100% 7

Page 123: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

108 The Water Research Foundation

Q9 - What locations of trees do you consider when assessing stormwater effects? Check all that apply

# Answer % Count

1 Street trees 14.71% 5

2 Trees on private property 11.76% 4

3 Park trees with mown grass groundcover 17.65% 6

4 Trees in riparian (streamside) areas 20.59% 7

5 Natural wooded areas with understory plants and a ground cover of fallen leaves 17.65% 6

6 Trees in floodplains 17.65% 6

7 None 0.00% 0

Total 100% 34

Page 124: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 109

Q10 - What values and benefits do you attribute to the location of trees above? Check all that apply

Page 125: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

110 The Water Research Foundation

# Question

Reduce stormwate

r runoff quantity

Increase soil

infiltration rates

Improve water

quality of stormwate

r runoff

Erosion

reduction

Carbon

sequestration

Air quality

improvement

Shading and summer

temperature reduction of streams

Summer air temperature reduction

Total

1 Street trees 15.91% 7 13.64% 6 15.91% 7 6.82% 3 13.64% 6 15.91% 7 2.27% 1 15.91% 7 44

2 Trees on

private property

13.21% 7 13.21% 7 13.21% 7 13.21% 7 13.21% 7 13.21% 7 7.55% 4 13.21% 7 53

3

Park trees with mown

grass groundcove

r

13.73% 7 13.73% 7 13.73% 7 9.80% 5 13.73% 7 13.73% 7 9.80% 5 11.76% 6 51

4

Trees in riparian

(streamside areas)

13.21% 7 9.43% 5 13.21% 7 13.21% 7 13.21% 7 13.21% 7 13.21% 7 11.32% 6 53

5

Natural wooded

areas with understory

plants and a ground

cover of and fallen leaves

12.96% 7 12.96% 7 12.96% 7 11.11% 6 12.96% 7 12.96% 7 11.11% 6 12.96% 7 54

6 Trees in floodplains 12.73% 7 12.73% 7 12.73% 7 12.73% 7 12.73% 7 12.73% 7 12.73% 7 10.91% 6 55

Page 126: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 111

Q11 - What regulations, policies or programs are actively used in your municipality or state to promote tree planting and retention of tree cover?

What regulations, policies or programs are actively used in your municipality or state to promote tree planting and retention of tree cover? Our Urban Forest Master Plan has several strategies including regulations, policies, and programs for tree preservation and planting. Development Regulations for existing developed parcels and new development. Fee in lieu of planting program. Arbor Day plantings. Tree City USA. Tree planting budget. Landmark tree designation. Critical areas regulations. Shoreline regulations (rivers, streams, lakes).

Regular planting programs, Tree protection during construction, Cyclic Trimming program

Requirements for tree preservation and landscaping during development. Floodplain and Resource Protection Area protections from development and removal of trees and vegetation.

Austin's Tree Preservation Code

Q12 - 50 States, D.C. and Puerto Rico

# Answer % Count

1 Alabama 0.00% 0

2 Alaska 0.00% 0

3 Arizona 0.00% 0

4 Arkansas 0.00% 0

5 California 0.00% 0

6 Colorado 0.00% 0

7 Connecticut 0.00% 0

8 Delaware 0.00% 0

9 District of Columbia 0.00% 0

10 Florida 0.00% 0

11 Georgia 0.00% 0

12 Hawaii 0.00% 0

13 Idaho 0.00% 0

14 Illinois 14.29% 1

15 Indiana 0.00% 0

16 Iowa 0.00% 0

17 Kansas 0.00% 0

18 Kentucky 0.00% 0

Page 127: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

112 The Water Research Foundation

19 Louisiana 0.00% 0

20 Maine 0.00% 0

21 Maryland 0.00% 0

22 Massachusetts 0.00% 0

23 Michigan 0.00% 0

24 Minnesota 0.00% 0

25 Mississippi 0.00% 0

26 Missouri 14.29% 1

27 Montana 0.00% 0

28 Nebraska 0.00% 0

29 Nevada 0.00% 0

30 New Hampshire 0.00% 0

31 New Jersey 0.00% 0

32 New Mexico 0.00% 0

33 New York 0.00% 0

34 North Carolina 14.29% 1

35 North Dakota 0.00% 0

36 Ohio 0.00% 0

37 Oklahoma 0.00% 0

38 Oregon 0.00% 0

39 Pennsylvania 0.00% 0

40 Puerto Rico 0.00% 0

41 Rhode Island 0.00% 0

42 South Carolina 0.00% 0

43 South Dakota 0.00% 0

44 Tennessee 0.00% 0

45 Texas 14.29% 1

46 Utah 0.00% 0

47 Vermont 0.00% 0

Page 128: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 113

48 Virginia 28.57% 2

49 Washington 14.29% 1

50 West Virginia 0.00% 0

51 Wisconsin 0.00% 0

52 Wyoming 0.00% 0

53 I do not reside in the United States 0.00% 0

Total 100% 7

Q13 - What is the population size category of your municipality?

# Answer % Count

1 Greater than 100,000 71.43% 5

2 Greater than 75,000 but less than 100,000 14.29% 1

3 Greater than 50,000 but less than 75,000 0.00% 0

4 Greater than 25,000 but less than 50,000 14.29% 1

5 Greater than 10,000 but less than 25,000 0.00% 0

6 Less than 10,000 0.00% 0

Total 100% 7

Page 129: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD
Page 130: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 115

Glossary Closed canopy. Closed canopy refers to condition in which the canopies of adjacent trees overlap. Deciduous. Trees that lose their leaves for part of the year. Diameter at breast height (DBH). Diameter of the trunk of a tree which is measured at the height of an adult’s breast. Conventionally, it is measured at 1.3 m above the ground. Diffuse porous. Trees that have vessels almost evenly distributed during an annual ring. In other words, the vessels don’t vary greatly in size from early in the growing season to later in the season. Evergreen. Trees which retain their leaves throughout the full year. Interception. The amount of rainfall which is intercepted by tree canopy or stems and mostly evaporates back to the atmosphere. Leaf Area Index (LAI). A dimensionless factor which characterizes the vertical distribution of leaf surfaces within the tree canopy. It is calculated as one-sided green leaf area per unit ground surface area. Open canopy. Tree systems in which trees are spaced far enough apart that canopies do not overlap. Phenology. Cyclic and seasonal natural phenomena. In the specific context of trees, includes seasonal leaf on and leaf off periods, flowering. Ring porous. Trees that have vessels different in sizes which means there is transition between the size of the pores shaped early in the growing season and those shaped later in the season. Solar radiation. Electromagnetic radiation which is created by the emitted energy from the sun. Stemflow. The portion of rainfall that flows down the stems and trunk to the ground at the tree’s base. Throughfall. The amount of rainfall which falls on the ground through the canopy. It falls whether as free throughfall (not touching the canopy) or dripping from the canopy. Transpiration. The process through which plants extract water from the soil and release it to the atmosphere in support of plant growth and cooling needs. Urban tree system. Defined as tree(s) and their underlying ground cover (e.g., vegetation, mulch) and soil/structural support systems growing in an urban area. Urban tree systems may occur as remnant upland and riparian forests, plantings in parks, along streets or other landscape areas, or within stormwater control measures. Vapor pressure deficit. The difference between the vapor pressure at saturation and actual vapor pressure. Represents part of the evaporative demand of the atmosphere and is a driver of transpiration. Water Quality Storm. The rainfall event for which 85% to 90% of storm events at a given location are

Page 131: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

116 The Water Research Foundation

equal to or smaller than. For many locations in the U.S., the water quality storm ranges in magnitude from 0.9 to 1.2 inches. Because it is associated with the majority of runoff volume and pollutant load, the water quality storm serves as the basis for sizing many green stormwater infrastructure practices. Xylem. Vascular tissue of the plant which transport water from the roots to the leaves.

Page 132: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 117

References Abas, M.R., Ahmad-Shah, A., and Awang, M.N. 1992. “Fluxes of anions in precipitation, throughfall and stemflow in an urban forest in Kuala Lumpur, Malaysia.” Environmental Pollution, 75(2): 209-213. Adair E.C., Hobbie S.E., and Hobbie R.K. 2010. “Single-pool exponential decomposition models: potential pitfalls in their use in ecological studies.” Ecology, 91(4):1225–1236 Allen, R.G., Pereira, L.S., Raes, D., and Smith, M. 1998. Crop evapotranspiration – guidelines for computing crop water requirements. Rome: Food and Agriculture Organization of the United Nations. Allen, R.G., Walter, I.A., Elliott, R., Howell, T., Itenfisu, D., and Jensen, M. 2005. The ASCE standardized reference evapotranspiration equation. Report prepared by Task Committee on Standardization of Reference Evapotranspiration of the Environmental and Water Resources Institute. Amador J.A., Hull, R.J., Patenaude, E.L., Bushoven, J.T., and Gorres, J.H. 2007. “Potential nitrate leaching under common landscaping plants.” Water Air Soil Pollut, 185(1-4): 323-333. Armson, D., Stringer, P., and Ennos, A.R. 2013. “The effect of street trees and amenity grass on urban surface water runoff in Manchester, UK.” Urban Forestry & Urban Greening, 12: 282-286. Asadian, Y. 2010. Rainfall interception in an urban environment. Master of Science Thesis, the University of British Columbia. Vancouver. BC. Asadian, Y., and Weiler, M. 2009. “A new approach in measuring rainfall intercepted by urban trees in coastal British Columbia.” Water Quality Research J. Can., 44: 16-25. Attarod, P., Sadeghi, S.M.M., Pypker, T.G., Bagheri, H., Bagheri, M., and Bayramzadeh, V. 2015. “Needle-leaved trees impacts on rainfall interception and canopy storage capacity in an arid environment.” New Forests, 46(3): 339–355. Barbier S., Balandier P., and Gosselin, F. 2009. “Influence of several tree traits on rainfall partitioning in temperate and boreal forests: a review.” Annals of Forest Science, 66: 602-612. Bartens, J., Day, S.D., Harris, J.R., Dove, J.E., and Wynn, T.M. 2008. “Can urban tree roots improve infiltration through compacted subsoils for stormwater management?” Journal of Environmental Quality, 37: 2048-2057. Bartens, J., Day, S.D., Harris, J.R., Wynn, T.M., and Dove, J.E. 2009. “Transpiration and root development of urban trees in structural soil stormwater reservoirs.” Environmental Management, 44: 646-657. Beck, H.E., Zimermann, N.E., McVicar, T.R., Vergopolan, N., Berg, A., and Wood, E.F. 2018. “Present and future Koppen-Geiger climate classification maps at 1-km resolution.” Scientific Data, 5: 180214. Beniston, M., Stephenson, D.B., Christensen, O.B., Ferro, C.A.T., Frei, C., Goyette, S., Halsnaes, K., Holt, T., and Woth, K. 2007. “Future extreme events in European climate: an exploration of regional climate model projections.” Climatic Change, 81: 71-95.

Page 133: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

118 The Water Research Foundation

Benfield, E.F. 1997. “Comparison of litterfall input to streams.” Journal of the North American Benthological Society, 16(1): 104-108. Berland, A., Shiflett, S.A., Shuster, W.D., Garmestani, A.S., Goddard, H.C., Herrmann, D.L., and Hopton, M.E. 2017. “The role of trees in urban stormwater management.” Landscape and Urban Planning, 162: 167-177. Berland, A., and Hopton, M.E. 2014. “Comparing street tree assemblages and associated stormwater benefits among communities in metropolitan Cincinnati, Ohio, USA.” Urban Forestry & Urban Greening, 13(4): 734-741. Bettez, N.D., and Groffman, P.M. 2012. “Denitrification potential in stormwater control structures and natural riparian zones in an urban landscape.” Environmental Science & Technology, 46: 10909-10917. Bettez, N.D., and Groffman, P.M. 2013. “Nitrogen deposition in and near an urban ecosystem.” Environmental Science & Technology, 47: 6047-6051. Bettez, N.D., Marino, R., Howarth, R.W., and Davidson, E.A. 2013. “Roads as nitrogen deposition hot spots.” Biogeochemistry, 114 (1−3): 149−163. Bharati, L., Lee, K.-H., Isenhart, T.M., and Schultz, R.C. 2002. “Soil-water infiltration under crops, pasture, and established riparian buffer in Midwestern USA.” Agroforestry Systems, 56: 249-257. Blair, J.M. 1988. “Nitrogen, sulfur, and phosphorus dynamics in decomposing deciduous leaf litter in decomposing deciduous leaf litter in the southern Appalachians.” Soil Biol Biochem. 20: 693-701. BMP Database (International Stormwater BMP Database). n.d. “International Stormwater BMP Database.” http://www.bmpdatabase.org/. Boggs, J.L., and Sun, G. 2011. “Urbanization alters watershed hydrology in the piedmont of North Carolina.” Ecohydrology, 4: 256-264. Bratt, A., Finlay, J.C., Hobbie, S.E., Janke, B.D., Worm, A.C., and Kemmitt, K.L. 2017. “Contribution of leaf litter to nutrient export during winter months in an urban residential watershed.” Environmental Science & Technology, 51(6): 3138-3147. Brok, J., Thorlund, K., Gluud, C., and Wetterslev, J. 2008. “Trial sequential analysis reveals insufficient information size and potentially false positive results in many metaanalyses.” J. Clin. Epidemiol., 61: 763–769. Bu, X., Xue, J., Zhao, C. Wu, Y., Han, F., and Zhu, L. 2016. “Sediment and nutrient removal by integrated tree-grass riparian buffers in Taihu Lake watershed, eastern China.” J. Soil Water Conserv., 71(2): 129-136. Cappellato R., and Peters, N.E. 1995. “Dry deposition and canopy leaching rates in deciduous and coniferous forests of the Georgia piedmont—an assessment of a regression-model.” J Hydrol., 169(1–4): 131–150.

Page 134: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 119

Cappiella, K., Claggett, S., Cline, K., Day, S., Galvin, M., MacDonagh, P., Sanders, J., Whitlow, T., and Xiao, Q. 2016. Recommendations of the expert panel to define BMP effectiveness for urban tree canopy expansion. http://www.chesapeakebay.net/documents/Urban_Tree_Canopy_EP_Report_WQGIT_approved_final.pdf. Cappiella, K., Schueler, T., Tomlinson, J., and Wright, T. 2006b. Urban watershed forestry manual. Part 3: Urban tree planting guide. Newtown Square, PA: USDA Forest Service. Cappiella, K., Schueler, T., and T. Wright. 2005. Urban watershed forestry manual. Part I: Methods for increasing forest cover in a watershed. Newtown Square, PA: USDA Forest Service. Cappiella, K., Wright, T., and Schueler, T. 2006a. Urban watershed forestry manual. Part 2: Conserving and planting trees at development sites. Ellicott City, MD: Center for Watershed Protection. Carter, J.M. 2015. Phenology and physiology of white ash in relation to climate extremes. PhD Dissertation, University of Kansas, Lawrence, KS. https://kuscholarworks.ku.edu/handle/1808/21701. Chen, L., Zhang, Z., and Ewers, B.E. 2012. “Urban tree species show the same hydraulic response to vapor pressure deficit across varying tree size and environmental conditions.” PLOS ONE, 7(10): e47882. Chiwa, M., Kim, D.H., and Sakugawa, H. 2003. “Rainfall, stemflow, and throughfall chemistry at urban- and mountain-facing sites at Mt. Gokurakuji, Hiroshima, Western Japan.” Water, Air, and Soil Pollution, 146: 93-109. Cornwell, W.K., Cornelissen, J.H.C., Amatangelo, K., Dorrepal, E., Eviner, V.T., Godoy, O., Hobbie, S.E., Hoorens, B., and Kurokawa, H. 2008. “Plant species traits are the predominant control on litter decomposition rates within biomes worldwide.” Ecology Letters, 11: 1065-1071. Crockford, R.H., and Richardson, D.P. 2000. “Partitioning of rainfall into throughfall, stemflow and interception: effect of forest type, ground cover and climate.” Hydrological Processes, 14, 2903–2920. Dawson, T.E. 1996. “Determining water use by trees and forests from isotopic, energy balance and transpiration analyses: the roles of tree size and hydraulic lift.” Tree Physiology, 16(1–2): 263–272. David, T.S., Gash, J.H.C., Valente, F., Pereira, J.S., Ferreira, M.I., and David, J.S. 2006. “Rainfall interception by an isolated evergreen oak tree in a Mediterranean savannah.” Hydrological Processes, 20(13), 2713–2726. Decina, S.M., Templer, P.H., and Hutyra, L.R. 2018. “Atmospheric inputs of nitrogen, carbon, and phosphorus across an urban area: unaccounted fluxes and canopy influences.” Earth’s Future, 6: 134-148. Denman, E.C., May, P.B., and Moore, G.M. 2016. “The potential role of urban forests in removing nutrients from stormwater.” J. Environ. Qual., 45: 207-214.

Page 135: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

120 The Water Research Foundation

Donovan, G.H., Butry, D.T., and Mao, M.Y. 2016. “Statistical analysis of vegetation and stormwater runoff in an urban watershed during summer and winter storms in Portland, Oregon, U.S.” Arboriculture & Urban Forestry, 42(4): 318-328 Dorendorf, J., Wilken, A., Eschenbach, A., and Jensen, K. 2015. “Urban-induced changes in tree leaf litter accelerate decomposition.” Ecological Processes, 4 (1): 1–16. https://doi.org/doi:10.1186/s13717-014-0026-5. Dorney, J.R. 1986. “Leachable and total phosphorus in urban street tree leaves.” Water, Air, and Soil Pollution, 28: 439-443. Engel, B., Storm, D., White, M., Arnold, J., and Maxdak, A. 2007. “A hydrologic/water quality model application protocol.” J. American Water Res. Assoc., 43(5): 1123–1136. Enloe, H.A., Lockaby, G., Zipperer, W.C., and Somers, G.L. 2015. “Urbanization effects on leaf litter decomposition, foliar nutrient dynamics and aboveground net primary productivity in the subtropics.” Urban Ecosyst. DOI 10.1007/s11252-015-0444-x. EPA (United States Environmental Protection Agency). 2018. “Opti-tool: EPA Region 1’s stormwater management optimization tool.” Accessed May 28th, 2018. https://www.epa.gov/tmdl/opti-tool-epa-region-1s-stormwater-management-optimization-tool. Erickson, J.E., Cisar, J.L., Snyder, G.H., and Volin, J.C. 2005. “Phosphorus and potassium leaching under contrasting residential landscape models established on a sandy soil.” Crop Science, 45:546-552. Erickson, J.E., Cisar, J.L., Snyder, G.H., Park, D.M., and Williams, K.E. 2008. “Does a mixed- species landscape reduce inorganic- nitrogen leaching compared to a conventional St. Augustine grass lawn?” Crop Science, 48:1586-1594. Fair, B.A., Metzger, J.D., and Vent. J. 2012. “Characterization of physical, gaseous, and hydrologic properties of compacted subsoil and its effects on growth and transpiration of two maples grown under greenhouse conditions.” Arboriculture & Urban Forestry, 38(4): 151-159. Fang, Y., Yoh, M., Koba, K., Zhu, W., Takebayashi, Y., Xiao, Y., Lei, C., Mo, J., Zhang, W., and Lu, X. 2011. “Nitrogen deposition and forest nitrogen cycling along an urban–rural transect in southern China.” Global Change Biology, 17: 872-885. Fissore, C., Hobbie, S.E., King, J.Y., McFadden, J.P., Nelson, K.C., and Baker, L.A. 2012. “The residential landscape: Fluxes of elements and the role of household decisions.” Urban Ecosyst., 15(1): 1−18. Ford, C.R., Hubbard, R.M., and Vose, J.M. 2011. “Quantifying structural and physiological controls on variation in canopy transpiration among planted pine and hardwood species in the Southern Appalachians.” Ecohydrology, 4(2): 183-195. Fu, W., He, X., Xu, S., Chen, W., Li, Y., Li, B., Su, L., and Ping, Q. 2018. “Changes in nutrients and decay rate of Ginkgo biloba leaf litter exposed to elevated O3 concentration in urban area.” PeerJ., 5(6): e4453. doi: 10.7717/peerj.4453.

Page 136: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 121

Gash, J.H.C. 1979. “An analytical model of rainfall interception by forests.” Q. J. Roy. Meteorol. Soc., 105: 43–45. Geronimo, F.K.F., Maniquiz-Redillas, M.C., Tobio, J.A.S., and Kim, L.H. 2014. “Treatment of suspended solids and heavy metals from urban stormwater runoff by a tree box filter.” Water Science and Technology, 69(12): 2460-7. Gonzalez-Sosa, E., Braud, I., Pina, R.B., Loza, C.A.M., Salinas, N.M.R., and Chavez, C.V. 2017. “A new methodology to quantify ecohydrological services of street trees.” Ecohydrology & Hydrobiology. http://dx.doi.org/10.1016/j.ecohyd.2017.06.004. Groffman, P.M., Boulware, N.J., Zipperer, W.C., Pouyat, RV., Band, L.E., and Colosimo, M.F. 2002. “Soil nitrogen cycle processes in urban riparian zones.” Environmental Science & Technology, 36: 4547-4552. Groffman, P.M., and Crawford, M.K. 2003. “Denitrification potential in urban riparian zones.” J. Environ. Qual., 32: 1144-1149. Groffman, P.M., Pouyat, R.V., Cadenasso, M.L., Zipperer, W.C., Szlavecz, K., Yesilonis, I.D., Band, L.E., and Brush, G.S. 2006. “Land use context and natural soil controls on plant community composition and soil nitrogen and carbon dynamics in urban and rural forests.” Forest Ecology and Management, 236:177–192. doi: 10.1016/j.foreco.2006.09.002. Groffman, P.M., Williams, C.O., Pouyat, R.V., Band, L., and Yesilonis, I. 2009. “Nitrate leaching and nitrous oxide flux in urban forests and grasslands.” J Environ Qual., 38: 1848–1860. doi: 10.2134/jeq2008.0521. Guevara-Escobar, A., González-Sosa, E., Véliz-Chávez, C., Ventura-Ramos, E., and Ramos-Salinas, M. 2007. “Rainfall interception and distribution patterns of gross precipitation around an isolated Ficus benjamina tree in an urban area.” Journal of Hydrology, 333: 532-541. Gusewell, S., and Gessner, M.O. 2009. “N:P ratios influence litter decomposition and colonization by fungi and bacteria in microcosms.” Functional Ecology, 23(1): 211-219. Hagishima, A., Narita, K.I., and Tanimoto, J. 2007. “Field experiment on transpiration from isolated urban plants.” Hydrological Processes, 21: 1217-1222. Hart, T.D. 2017. Root-enhanced infiltration in stormwater bioretention facilities in Portland, Oregon. PhD Dissertation, Portland State University, Portland, OR. Hassan, S.M.T., Ghimire, C.P., and Lubczynski, M.W. 2017. “Remote sensing upscaling of interception loss from isolated oaks: Sardon catchment case study, Spain.” Journal of Hydrology, 555: 489–505. Hobbie, S.E., Baker, L.A., Buyarski, C., Nidzgorski, D., and Finlay, J.C. 2014. “Decomposition of tree leaf litter on pavement: Implications for urban water quality.” Urban Ecosyst., 17(2): 369–385. Hubbart, J.A., Muzika, R., Huang, D., and Robinson, A. 2011. “Hardwood forest influence on soil water consumption in an urban floodplain: Potential to improve flood storage capacity and reduce stormwater runoff.” Watershed Science Bulletin, Fall: 34-43.

Page 137: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

122 The Water Research Foundation

Hutmacher A.M., Zaimes G.N., Martin J., and Green D.M. 2015. “Vegetative litter decomposition along urban ephemeral streams in Southeastern Arizona.” Urban Ecosyst, 18: 431–448. doi: 10.1007/s11252-014-0405-9. Inkiläinen, E.N.M, McHale, M.R., Blank, G.B., and James, A.L. 2013. “The role of the residential urban forest in regulating throughfall: a case study in Raleigh, North Carolina, USA.” Landscape and Urban Planning, 119: 91-103. Izquieta-Rojano, S., Garcia-Gomez, H., Aguillaume, L., Santamaria, J.M., Tang, Y.S., Santamaria, C., Valino, F., Lasheras, E., Alonso, R., Avila, A., Cape, J.N., and Elustondo, D. 2016. “Throughfall and bulk deposition of dissolved organic nitrogen to holm oak forests in the Iberian Peninsula: Flux estimation and identification of potential sources.” Environmental Pollution, 210: 104-112. Janke, B., Finlay, J., and Hobbie, S. 2017. “Trees and streets as drivers of urban stormwater nutrient pollution.” Env. Sci. Tech., 51: 9569-9579 Janke, B.D., Finlay, J.C., Hobbie, S.E., Baker, L.A., Sterner, R.W., Nidzgorski, D., and Wilson, B.N. 2014. “Contrasting influences of stormflow and baseflow pathways on nitrogen and phosphorus export from an urban watershed.” Biogeochemistry, 121(1): 209–228. Johnson, M.S., and Lehmann, J. 2006. “Double-funneling of trees: stemflow and root-induced preferential flow.” Ecoscience, 13:324–33. Juknys, R., Zaltauskaite, J., and Stakenas, V. 2007. “Ion fluxes with bulk and throughfall deposition along an urban–suburban–rural gradient.” Water Air Soil Pollut., 178: 363–372. Kaushal, R., Kumar, A., Alam, N.M., Mandal, D., Jayaparkash J., Tomar, J.M.S., Patra, S., Guta, A.K., Mehta, H., Panwar, P., Chaturvedi, O.P., and Mishra, P.K. 2017. “Effect of different canopy management practices on rainfall partitioning in Morus alba.” Ecological Engineering, 102: 374-382. Kermavnar, J., and Vilhar, U. 2017. “Canopy precipitation interception in urban forests in relation to stand structure.” Urban Ecosystems. DOI 10.1007/s11252-017-068907. Kimura, S.D., Saito, M., Hara, H., Xu, Y.H., and Okazaki, M., 2009. “Comparison of nitrogen dry deposition on cedar and oak leaves in the Tama hills using foliar rinsing method.” Water Air Soil Pollut., 202: 369–377. Kjelgren, R., Beeson, R.C., Pittenger, D.R., and Montague, D.T. 2016. “Simplified landscape irrigation demand estimation: slide rules.” Applied Engineering in Agriculture, 32(4): 363-378. Kjelgren, R., and Montague, T. 1998. “Urban tree transpiration over turf and asphalt surfaces.” Atmospheric Environment, 32: 35-41. Kuehler, E., Hathaway, J., and Tirpak, A. 2017. “Quantifying the benefits of urban forest systems as a component of the green infrastructure stormwater treatment network.” Ecohydrology, 10(3): e1813. Lange, B., Luescher, P., and Germann, P.F. 2009. “Significance of tree roots for preferential infiltration in stagnic soils.” Hydrology and Earth System Sciences, 13: 1809-1821.

Page 138: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 123

Levitt, D.G., Simpson, J.R., and Tipton, J.L. 1995. “Water use of two landscape tree species in Tucson, Arizona.” Journal of the American Society for Horticultural Science, 120(3): 409-416. Li, Y.C., Alva, A.K., Calvert, D.V., and Zhang, M. 2008. “Chemical composition of throughfall and stemflow from citrus canopies.” Journal of Plant Nutrition. https://doi.org/10.1080/01904169709365339. Litvak, E., McCarthy, H.R., and Pataki, D.E. 2012. “Transpiration sensitivity of urban trees in a semi-arid climate is constrained by xylem vulnerability to cavitation.” Tree Physiology, 32: 373-388. Litvak, E., McCarthy, H.R., and Pataki, D.E. 2017. “A method for estimating transpiration of irrigated urban trees in California.” Landscape and Urban Planning, 158: 48-61. Livesley, S.J., Baudinette, B., and Glover, D. 2014. “Rainfall interception and stem flow by eucalypt street trees – The impacts of canopy density and bark type.” Urban Forestry and Urban Greening, 13(1): 192-197. Livesley, S.J., Ossola, A., Threlfall, C.G., Hahs, A.K., and Williams, N.S.G. 2016. “Soil carbon and carbon/nitrogen ratio change under tree canopy, tall grass and turf grass areas of urban green space.” J. Environ. Quality, 45(1): 215-223. MARC (Mid-America Regional Council). 2013. Natural resources inventory 2.0. Kansas City, MO: MARC. http://www.marc.org/Environment/Natural-Resources/Natural-Resources-Inventory/NRI-Resources. MARC (Mid-America Regional Council). 2014. The nature of nature in greater Kansas City: Using natural resource inventory data in planning. Kansas City, MO: MARC. http://www.marc.org/Environment/Natural-Resources/pdf/NRI-2013/TheNatureOfNatureInGreaterKansasCity_eResOptmzd.aspx. Matteo, M., Randhir, T., and Bloniarz, D. 2006. “Watershed-scale impacts of forest buffers on water quality and runoff in urbanizing environment.” Journal of Water Resources Planning and Management, 132(2): 144-15. McDonnell, M., Pickett, S., Groffman, P., Bholen, P., Pouyat, R., Zipperer, W., Parmelee, R., Carreiro, M., and Medley, K. 1997. “Ecosystem processes along an urban-to-rural gradient.” Urban Ecosystems, 1(1): 21–36. https://doi.org/doi:1014359024275. McPherson, G., Simpson, J.R., Peper, P.J., Maco, S.E., and Xiao, Q. 2005. “Municipal forest benefits and costs in five US cities.” Journal of Forestry, 103(8): 411-416. Michopoulos, P., Baloutsos, G., Economou, A., Voulala, M., and Bourletsikas, A. 2007. “Bulk and throughfall deposition chemistry in three different forest ecosystems.” Fresenius Environmental Bulletin, 16(1): 91–98. https://doi.org/10.1007/978-94-007-1363-5_16. Miller, W.W., Johnson, D.W., Denton, C., Verburg, P.S.J., Dana, G.L., and Walker, R.F. 2005. “Inconspicuous nutrient laden surface runoff from mature forest Sierran watersheds.” Water, Air, Soil Pollut., 163(1–4): 3–17.

Page 139: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

124 The Water Research Foundation

Moore, T.L., Gulliver, J.S., Stack, L., and Simpson, M.H. 2016. “Stormwater management and climate change: vulnerability and capacity for adaptation in urban and suburban contexts.” Climatic Change, 138(3): 491–504. https://doi.org/10.1007/s10584-016-1766-2. Moore, J., Macrellis, A., and Bailey, K. 2014. Stormwater benefits of trees. Vermont Department of Forests, Parks and Recreation. https://vtcommunityforestry.org/sites/default/files/pictures/waterqualitytreebenefits.pdf. Motahari, M., Attarod, P., Pypker, T.G., Etemad, V., and Shirvany, A. 2013. “Rainfall interception in a Pinus eldarica plantation in a semi-arid climate zone: an application of the gash model.” Journal of Agricultural Science and Technology, 15(5): 981-994. Nadelhoffer, K.J., Aber, J.D., and Melillo, J.M. 1983. “Leaf-litter production and soil organic matter dynamics along a nitrogen-availability gradient in Southern Wisconsin, U.S.A.” Can. J. For. Res., 13: 12-21. NAS (National Academies of Sciences, Engineering and Medicine). 2014. Long-term performance and lifecycle costs of stormwater best management practices. Washington, DC: The National Academies Press. https://doi.org/10.17226/22275. Neal, C., Reynolds, B., Neal, M., Hill, L., Wickham, H., and Pugh, B. 2003. “Nitrogen in rainfall, cloud water, throughfall, stemflow, stream water and groundwater for the Plynlimon catchments of mid-Wales.” Sci. Total Environ., 314–316: 121–151. Nidzgorski, D.A., and Hobbie, S.E. 2016. “Urban trees reduce nutrient leaching to groundwater.” Ecological Applications, 26(5): 1566-1580. Nikula, S., Vapaavuori, E., and Manninen, S. 2010. “Urbanization-related changes in European aspen (Populus tremula L.): Leaf traits and litter decomposition.” Environmental Pollution, 158(6): 2132–2142. https://doi.org/10.1016/j.envpol.2010.02.025. Niu, G., Rodriguez, D.S., Cabera, R., McKenney, C., and Mackay, W. 2006. “Determining water use and crop coefficients of five woody landscape plants.” J. Environ. Hort., 24(3): 160-165. Nowak, D.J., Bodine, A.R., Hoehn, R.E., Crane, D.E., Ellis, A., Endreny, T.A., Yang, Y., Jacobs, T., and Shelton, K. 2013. Assessing urban forest effects and values: the greater Kansas City region. Newtown Square, PA: U.S. Department of Agriculture, Forest Service, Northern Research Station. https://doi.org/10.2737/NRS-RB-75. NRCS (Natural Resources Conservation Service). 1986. Urban hydrology for small watersheds. Washington, DC: USDA NRCS. Nytch, C.J., Meléndez-Ackerman, E.J., Pérez, M.-E., and Ortiz-Zayas, J.R. 2019. “Rainfall interception by six urban trees in San Juan, Puerto Rico.” Urban Ecosystems, 22(1): 103–115. Page, J.L., Winston, R.J., and Hunt, W.F., III. 2015. “Soils beneath pavements: An opportunity for stormwater control and treatment.” Ecological Engineering, 82: 40-48.

Page 140: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 125

Park, A., and Cameron, J.L. 2008. “The influence of canopy traits on throughfall and stemflow in five tropical trees growing in a Panamanian plantation.” For. Ecol. Manag., 255: 1915–1925. Pataki, D.E., McCarthy, H.R, Litvak, E., and Pincetl, S. 2011. “Transpiration of urban forest in the Los Angeles metropolitan area.” Ecological Applications, 21(3): 661-677. Pavao-Zuckerman, M.A., and Coleman, D.C. 2005. “Decomposition of chestnut oak (Quercus prinus) leaves and nitrogen mineralization in an urban environment.” Biol. Fertil. Soils, 41: 343-349. Payne, E., Pham, T., Cook P., Fletcher, T.D., Hatt, B.E., and Deletic, A. 2013. “Biofilter design for effective nitrogen removal from stormwater – influence of plant species, inflow hydrology and use of a saturated zone.” Water Sci. & Tech., 69(6): 1312-1319. Pereira, F.L., Gash, J.H.C., David, J.S., David, T.S., Monteiro, P.R., and Valente, F. 2009. “Modelling interception loss from evergreen oak Mediterranean savannas: Application of a tree-based modelling approach.” Agricultural and Forest Meteorology, 149(3): 680–688. Peters, E.B., McFadden, J.P., and Montgomery, R.A. 2010. “Biological and environmental controls on tree transpiration in a suburban landscape.” Biogeosciences, 115(G4). Ponette-González, A.G., Weathers, K.C., and Curran, L.M. 2017. “Tropical land-cover change alters biogeochemical inputs to ecosystems in a Mexican montane landscape.” Ecol. Appl., 20: 1820–1837. Post, D.A., and Jones, J.A. 2001. “Hydrologic regimes of forested, mountainous, headwater basins in New Hampshire, North Carolina, Oregon, and Puerto Rico.” Advances in Water Resources, 24: 1195-1210. Pouyat, R.V., and Carreiro, M.M. 2003. “Controls on mass loss and nitrogen dynamics of oak leaf litter along an urban-rural land-use gradient.” Oecologia, 135(2): 288–98. Qin, Z., Shober, A.L., Beeson, R.C., and Wiese, C. 2013. “Nutrient leaching from mixed-species Florida residential landscapes.” J. Environ. Qual., 42: 1534–1544. doi: 10.2134/jeq2013.04.0126. Rahman, M.A., Moser, A., Anderson, M., Zhang, C. Rotzer, T., and Pauleit, S. 2019. “Comparing the infiltration potentials of soils beneath the canopies of two contrasting urban tree species.” Urban Forestry & Urban Greening, 38: 22-32. Read, J., Wevill, T., Fletcher, T., and Deletic, A. 2008. “Variation among plant species in pollutant removal from stormwater in biofiltration systems.” Water Research, 42(4-5): 893-902. Riikonen, A., Jarvi, L., and Nikinmaa, E. 2016. “Environmental and crown related factors affecting street tree transpiration in Helsinki, Finland.” Urban Ecosystems, 19: 1693-1715. Rohatgi, A. 2019. “WebPlotDigitizer.” https://automeris.io/WebPlotDigitizer. Sadeghi, S., Attarod, P., Van Stan, J.T., and Pypker, T.G. 2016. “The importance of considering rainfall partitioning in afforestation initiatives in semiarid climates: a comparison of common planted tree species in Tehran, Iran.” Science of the Total Environment, 568: 845-855.

Page 141: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

126 The Water Research Foundation

Sanders, R.A. 1986. “Urban vegetation impacts on the hydrology of Dayton, Ohio.” Urban Ecology, 9(3-4): 361-376. Scharenbroch, B.C., Morgenroth, J., and Maule, B. 2016. “Tree species suitability to bioswales and impact on the urban water budget.” Journal of Environmental Quality, 45: 199-206. Schechter, S.P., Canfield, T.J., and Mayer, P.M. 2013. A meta-analysis of phosphorous attenuation in best management practices (MBP) and low impact development (LID) practices in urban and agricultural areas. U.S. Environmental Protection Agency, National Risk Management Research Laboratory. Accessed September 1, 2017. https://archive.epa.gov/ada/web/pdf/p100jdot.pdf. Schooling, J.T., and Carlyle-Moses, D.E. 2015. “The influence of rainfall depth class and deciduous tree traits on streamflow production in an urban park.” Urban Ecosystems, 18: 1261-1284. Schueler, T. 1987. Controlling urban runoff: a practical manual for planning and designing urban BMPs. Washington, DC: Metropolitan Washington Council of Governments. Schueler, T. 2003. Impacts of impervious cover on aquatic systems. Ellicott City, MD: Center for Watershed Protection. Selbig, W.R. 2016. “Evaluation of leaf removal as a means to reduce nutrient concentrations and loads in urban stormwater.” Sci. Tot. Environ., 571: 124-133. Song, X.P., Tan, P.Y., Edwards, P., and Richards, D. 2018. “The economic benefits and cost of trees in urban forest stewardship: a systematic review.” Urban Forestry & Urban Greening, 29: 162-170. Stack, B., Law, N., Drescher, S., and Wolinski, B. 2013. Gross solids characterization study in the Tred Avon Watershed Talbot County, MD. Prepared for Talbot County Department of Public Works. Staelens, J., De Schrijver, A., Verheyen, K., and Verhoest, N.E.C. 2008. “Rainfall partitioning into throughfall, stemflow, and interception within a single beech (Fagus sylvatica L.) canopy: Influence of foliation, rain event characteristics and meteorology.” Hydrological Processes, 22: 33–45. Strahm, B.D., Harrison, R.B., Terry, T.A., Flaming, B.L., Licata, C.W., and Petersen, K.S. 2005. “Soil solution nitrogen concentrations and leaching rates as influenced by organic matter retention on a highly productive douglas-fir site.” Forest Ecology and Management, 218: 74–88. Sun, Y., and Zhao, S. 2016. “Leaf litter decomposition in urban forests: test of the home-field advantage hypothesis.” Annals of Forest Science, 73: 1063–1072. doi: 10.1007/s13595-016-0577-y. Taneda, H., and Sperry, J.S. 2008. “A case-study of water transport in co-occurring ring-versus diffuse-porous trees: contrasts in water-status, conducting capacity, cavitation and vessel refilling.” Tree Physiol. 28: 1641-1651. Tchobanoglous, G., Burton, F.L., and Stensel, H.D. 2004. Wastewater engineering treatment and reuse, 4th Ed. Boston, MA: McGraw Hill.

Page 142: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

Incorporating Forestry into Stormwater Management Programs 127

Templer, P.H., Toll, J.W., Hutyra, L.R., and Raciti, S.M. 2015. “Nitrogen and carbon export from urban areas through removal and export of litterfall.” Environmental Pollution, 197: 256-261. Tulloss, E.M., and Cadenasso, M.L. 2015. “Nitrogen deposition across scales: hotspots and gradients in a California savanna landscape.” Ecosphere, 6(9): 1-12. Van Stan, J.T., and Gordon, D.A. 2018. “Mini-review: stemflow as a resource limitation to near-stem soils.” Frontiers in Plant Science, 9. https://doi.org/10.3389/fpls.2018.00248. Van Stan, J.T., Levia, D.F., Inamdar, S.P., Lepori-Bui, M., and Mitchelle, M.J. 2012. “The effects of phenoseason and storm characteristics on throughfall solute washoff and leaching dynamics from a temperate deciduous forest canopy.” Science of the Total Environment, 430: 48–58. Van Stan, J.T., II, Levia, D.F., Jr., and Jenkins, R.B. 2015. “Forest canopy interception loss across temporal scales: Implications for urban greening initiatives.” The Professional Geographer, 67: 41–51. Van Stan, J.T., Siegert, C.M., Levia, D.F., and Scheick, C.E. 2011. “Effects of wind-driven rainfall on stemflow generation between codominant tree species with differing crown characteristics.” Agricultural and Forest Meteorology, 151: 1277–86. Véliz-Chávez, C., Mastachi-Loza, C.A., González-Sosa, E., Becerril-Piña, R., and Ramos-Salinas, N.M. 2014. “Canopy storage implications on interception loss modeling.” American Journal of Plant Sciences, 05(20): 3032–3048. Wang, J., Endreny, T.A., and Nowak, D.J. 2008. “Mechanistic simulation of tree effects in an urban water balance model.” Journal of the American Water Resources Association, 44(1): 75–85. https://doi.org/doi:10.1111/j.1752-1688.2007.00139.x. Wang, H., Wang, X., Zhao, P., Zheng, H., Ren, Y., Gao, F., and Ouyang, Z. 2012. “Transpiration rates of urban trees, Aesculus chinensis.” Journal of Environmental Sciences (China), 24(7): 1278–1287. Wissmar, R.C., Timm, R.K., and Logsdon, M.G. 2004. “Effects of changing forest and impervious land covers on discharge characteristics of watersheds.” Environmental Manage., 34(1): 91-98. Wullschleger, S.D., Hanson, P.J., and Todd, D.E. 2001. “Transpiration from a multi-species deciduous forest as estimated by xylem sap flow techniques.” Forest Ecology and Management, 143: 205-213. Xiao, Q., McPherson, E.G., Ustin, S.L., and Grismer, M.E. 2000. “A new approach to modeling tree rainfall interception.” Journal of Geophysical Research, 105(D23): 29, 173-29,188. Xiao, Q., and McPherson, E.G. 2011a. “Performance of engineered soil and trees in a parking lot bioswale.” Urban Water Journal, 8(4): 241-253. Xiao, Q., and McPherson, E.G. 2011b. “Rainfall interception of three trees in Oakland, California.” Urban Ecosystems, 14: 755-769. Xiao, Q., and McPherson, E.G. 2016. “Surface water storage capacity of twenty tree species in Davis, California.” Journal of Environmental Quality, 45: 188–198.

Page 143: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

128 The Water Research Foundation

Yang, Y., and Toor, G.S. 2017. “Sources and mechanisms of nitrate and orthophosphate transport in urban stormwater runoff from residential catchments.” Water Research, 112 (April): 176–184. https://doi.org/doi:10.1016/j.watres.2017.01.039. Zabret, K., Rakovec, J., and Sraj, M. 2018. “Influence of meteorological variables on rainfall partitioning for deciduous and coniferous tree species in urban area.” Journal of Hydrology, 558: 29-41. Zadeh, M.K., and Sepaskhah, A.R. 2016. “Effect of tree roots on water infiltration rate into the soil.” Iran Agricultural Research, 35: 13-20. Zhang, P. 1999. Nutrient inputs from trees via throughfall, stemflow and litterfall in an intercropping system. Thesis, University of Guelph, Ottawa, Canada.

Page 144: Incorporating Forestry into Stormwater …...Incorporating Forestry into Stormwater Management Programs iii Acknowledgments Research Team Principal Investigators: Trisha Moore, PhD

1199 North Fairfax Street, Suite 900Alexandria, VA 22314-1445

6666 West Quincy AvenueDenver, CO 80235-3098

www.waterrf.org | [email protected]

advancing the science of water®