1

Research on Integrated Earth System Modeling at Global and Regional Scales

L. Ruby LeungPacific Northwest National Laboratory, Richland,

WA

2nd RASM WorkshopMonterey, CA, May 15 – 17, 2012

The needs for model coupling and new development

Understand the role of biosphere-atmosphere feedbacks on droughts in the southwestern U.S.Asses the impacts of climate change in the southeastern U.S. (e.g., hurricanes)Impacts of land-atmosphere (Amazon) and atmosphere-ocean (Atlantic) interactions on the tropical Atlantic biases

• Develop an integrated model to represent human-earth system interactions for modeling and analysis of climate change mitigation and adaptation, with a focus on the nexus of energy, water, and land use

2

3

• A regional earth system model is being developed using WRF, CLM, and ROMS, following the flux coupling approach used in CESM

Regional Earth System Model (RESM)

Atmosphere and ocean boundary conditions

POP

Flux Coupler

Atmospheric conditions

Surface fluxes

CESMCAM

Glo

bal

ROMS

Flux Coupler

Atmospheric conditions

Surface fluxes

RESMWRF

Reg

ion

al

• Consistent representations of land processes at global and regional scales

• Flexibility to model land processes using resolution or grid different from the atmospheric model

• Integrate human systems in CLM

• Facilitate air-sea coupling at regional scale

Model development

• Developed global high resolution (0.05o) input data for CLM based on MODIS

• Implemented the VIC surface/subsurface runoff and groundwater parameterizations to CLM

• Tested grid based vs subbasin based approaches• Developed a new river routing model for CLM for both

grid based and subbasin based approaches (including global input data at 6 different resolutions)

• Developing a water management model for CLM• Adding subgrid elevation classification in CLM• Applied UQ to understand model sensitivity to hydrologic

parameters• Developing WRF-ROMS coupling through CPL7

4

5

A 0.05-degree input dataset for CLM

Comparison of new and old CLM input data

6

Bare soil

Trees

Shrubs

Grass

Crop

7

Introducing VIC soil hydrology to CLM

Saturation excess runoff

Infiltration excess runoff

ARNO baseflow curve

Surface- and groundwater interactions

Hydraulic redistributionInteractions of water

movement between the root system and soil porous

media

8

Dynamic representation of surface and groundwater interactions

dtEQRpttt

tnttt

tt

t

tb

es

)()(

)(

1)()(

z

K

zD

zt

)(

)(

Change of water table depth

Change of total soil moisture in the unsaturated

zone

Net water recharge to the groundwater body

s porosityne(t) effective porosity

Change of soil

moisture

Diffusion term

Drainage term

Liang et al., JGR, 2003

9

Implementation of VICGROUND to CLM

A runtime option activated through the namelist

10

Simulated water budget at Tonzi Ranch

11

Global testing of CLMVIC

• CLM4-SP– Forcing: Qian et al. 2004– Land cover: current (i.e., 2000)– Simulation period: 1995-2004 – Resolution: standard one-degree (i.e., 0.9 x 1.25)

• CLM-CN– Forcing: CRU-NCEP– Land cover: potential vegetation (pre-industrial)– Simulation period: 1800-1900 (by randomizing 1901-

1930)– Resolution: 0.5-degree grid

12

CLM4-SP: Summer LH, 1995-2004CLM4 CLM4VIC

CLM4VIC – CLM4CLM4VIC – CLM4, global mean

13

CLM4-CN: Summer LH CLM4CN CLM4VICCN

CLM4VICCN – CLM4CN CLM4VICCN – CLM4CNglobal mean, stabilized

14

Motivation for a new runoff routing model• To provide more accurate freshwater flux to the ocean from subdaily to daily

time scales

• To provide a linkage between the human (e.g., surface water withdrawal, reservoir operation) and natural systems

• For transport of nutrients and sediments

Features

• Consistent process representation across various scales (global, regional, local)

• Easy to be coupled with water management model

• Easy to be coupled with other fluxes

15

River Transport Model (RTM) in CLM 4.0

• Study area divided into cells

• Flow direction is determined by D8 algorithm

• Cell-to-cell routing with a linear advection model

Limitations

• Over-simplification of river network• Over-simplification of physical processes

• Global constant channel velocity (0.35m/s)• No account for sub-grid heterogeneity

16

Grid-based approach

This hierarchical dominant river tracing method preserves the baseline high resolution

hydrography (flow direction, flow length, upstream drainage area) at any coarse

resolution (Wu et al. 2011)

Subbasin representation preserves the natural boundaries of runoff

accumulation and river system organization

Model for Scale-Adaptive River Transport (MOSART)

Subbasin-based approach

17

Grid-based approach

Model for Scale-Adaptive River Transport (MOSART)

Subbasin-based approachConceptualized network

Hillslope routingSub-network routing

Main channel routing

• Hillslope routing to account for event dynamics and impacts of overland flow on soil erosion, nutrient loading, etc.

• Sub-network routing: scale adaptive across different resolutions to reduce scale dependence

• Main channel routing: explicit estimation of in-stream status (velocity, water depth, etc.)

18

Inputs and Parameters

• Daily runoff generation from UW VIC at 1/16o resolution for the Columbia River Basin

• Spatial delineation and network based on HydroSHEDS– DRT algorithm for grid-based representation 1/16, 1/8,

¼ and ½ degree resolutions (available globally)– ArcSWAT package for subbasin-based representation

(average size ~109km2)

• Manning’s roughness for hillslope and channel routing set to 0.4 and 0.05, respectively

• Evaluate against monthly naturalized streamflow data at selected major stations

Improved streamflow simulations

19 Large drainage area Small drainage area

20

Water Resource Management Model: Conceptual Design

• For full coupling in an earth system model:– Assume no knowledge of future inflow – Use generic operating rules

• Two components: – Regulation module: extraction of water at the reservoir

• Storage: stores water over extended period of time• Regulation: Follows monthly operating rules for flood

control, environmental flow, irrigation and hydropower• Constrained extraction: Daily partitioning of reservoir

releases for irrigation water supply, other consumptive uses and environmental constraints. It includes the distribution across demanding units.

– Local surface water extraction module: extracts water at the unit

• Hillslope surface runoff: represent irrigation retention ponds• Unit main stem if unpounded by an upstream reservoir

21

CLM-MOSART-WRM coupling

Loop over PFTs

Irrigated fraction found

Need irrigation

Routing + reservoir model (T- Δt )

YES YES

NO

End of loop

Updated ET, runoff, baseflow, irrigation demand

YES

CLM (t)

PFTs: vegetation types

Local surface water (t- Δt) contribution to irrigation demand (t). Remaining

demand?

Extraction from main stem if not impounded. Remaining demand?

Extraction from reservoir release to complement the local supply

NO

Aggregated demand (t- Δt)

Aggregated supply (t)

YES

NO

CLM Routing model

Natural flow in each units; irrigation demand

WRM

Generated runoff, Agg. irrigation demand

Regulated flow; Irrigation supply at each unit

Agg. irrigation supply

22

Data Preprocessing

Create a “unit-reservoir dependency database”:- Local approach: independent tributaries, elevation

constraint, constrained distance-based buffer- Global approach: elevation constraint and distance-based

buffer

Distribute the demand across the reservoir based on the dependency database and maximum storage capacity of each dependent dam

Downscaling CCSM Simulations

23

• WRF-CLM is being used to downscale CMIP5 CCSM historical, RCP4.5, and RCP8.5 simulations from 1975 - 2100

24

Uncertainty quantification framework

25

Ranks of significance of input parameters over 10 Flux Tower Sites

Larger sensitivity to parameters of subsurface processes

Effects of Barrier Layers on TC Intensification

26

Balaguru et al. 2012 PNAS (in revision)

TC intensification rate is higher by 20% for TC that passes over BL than over non-BL

27