Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 0
Applicability of targeting vegetative filter strips to abate fecalbacteria and sediment yield using SWAT
P.B. Parajuli a,*, K.R. Mankin b, P.L. Barnes b
aDepartment of Agronomy, Kansas State University, Manhattan, KS 66506, United StatesbDepartment of Biological and Agricultural Engineering, Kansas State University, Manhattan, KS 66506, United States
a r t i c l e i n f o
Article history:
Received 15 December 2007
Accepted 12 May 2008
Published on line 2 July 2008
Keywords:
Non-point source pollution
Best management practice
Targeted approach
Random approach
Watershed
a b s t r a c t
Quantifying and evaluating effects of best management practices (BMPs) on water quality is
necessary to maximize the effectiveness of BMPs for minimizing pollutants. Watershed-
scale evaluation of effects of BMP implementation on fecal bacteria and sediment yield can
be estimated using a watershed water quality model, and strategies for identifying critical
areas in a watershed can be pollutant specific. The soil and water assessment tool (SWAT)
model was used in the Upper Wakarusa watershed (950 km2) in northeast Kansas to explore
effectiveness of vegetative filter strip (VFS) lengths applied at the edge of fields to reduce
non-point source pollution. The Upper Wakarusa watershed is a high priority total max-
imum daily load (TMDL) designation watershed for fecal bacteria in Kansas. This study
characterizes fecal bacteria sources (human, livestock, and wildlife) and targets VFS to abate
sediment and fecal bacteria pollution from the Upper Wakarusa watershed.
Objectives of this study were to test the effectiveness of VFS lengths (0, 10, 15 and 20 m)
for removing overland process sediment and fecal bacteria concentration, rank sub-water-
sheds after determining sediment and fecal bacteria contribution of each sub-watershed,
and demonstrate the SWAT model’s ability to evaluate effectiveness of VFS application to
abate sediment and fecal bacteria using targeted and random approaches to select 10, 25 and
50% of the sub-watersheds. The 15-m VFS reasonably reduced fecal bacteria concentration
in the watershed. The greatest difference between the target and random approaches to
fecal bacteria reduction was at 50% VFS adoption; the target approach removed about 60% of
fecal bacteria, and the random approach removed about 42%. For sediment yield, the
greatest reduction was at 25% VFS adoption; the target approach removed about 63% of
sediment yield, and the random approach removed about 33%. A targeted watershed
modeling approach using SWAT was effective at reducing both fecal bacteria concentration
and sediment yield.
# 2008 Elsevier B.V. All rights reserved.
avai lab le at www.sc iencedi rec t .com
journal homepage: www.e lsev ier .com/ locate /agwat
1. Introduction
Among the five leading pollutants in U.S. water bodies,
mercury is ranked first followed by pathogens and sediment
(USEPA, 2006). In Kansas, pathogens are ranked third (15.44%)
* Corresponding author at: 2011D Throckmorton Plant Sciences CenteTel.: +1 785 532 2766; fax: +1 785 532 6094.
E-mail address: [email protected] (P.B. Parajuli).
0378-3774/$ – see front matter # 2008 Elsevier B.V. All rights reservedoi:10.1016/j.agwat.2008.05.006
among causes of water quality impairment (USEPA, 2006).
Non-point source (NPS) pollution from agricultural lands
contributes to water quality degradation. Developing total
maximum daily loads (TMDLs) require quantifying pollutant
contribution from each source and determining pollutant
r, Kansas State University, Manhattan, KS 66506, United States.
d.
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 01190
reduction from each source required to meet applicable water
quality standards. Implementing best management practices
(BMPs) can sufficiently eliminate or reduce NPS pollution to
meet water quality criteria without disturbing environmental
quality (Novotny and Olem, 1994). Vegetative filter strip (VFS)
is a popular BMP for removing sediment and fecal bacteria and
decreasing pollutant loads from manured fields and pastures
(Guber et al., 2007; Inamdar et al., 2001; Park et al., 1994; SWCS,
2001; Vennix and Northcott, 2002).
Targeting, a term for implementing stricter pollution
control in areas where it will be most cost-effective (Veith
et al., 2001) often reduces costs compared with first-come,
first-served approaches. Several studies have developed
targeting procedures that enable watershed-specific evalua-
tion of NPS pollution control, and targeting methods incor-
porating a spatially distributed hydrologic/water quality
models have been demonstrated. The BMPs effectiveness
have been researched using many available modeling tools
(Moore et al., 1992; Niu et al., 2001; Zreig, 2001; Veith et al., 2001;
Vennix and Northcott, 2002; Bracmort et al., 2004), which can
be inexpensive and less time consuming as compared to
watershed scale field evaluation (Benham et al., 2006; Corwin
and Vaughan, 1997; Dickinson et al., 1990; Heatwole et al.,
1987; Tim et al., 1992). The location of the BMP targeting can be
varied due to pollutant variation. For example, targeting for
sediment and fecal bacteria reduction may or may not be in
the same hydrologic response units (HRUs) or sub-watersheds.
Various studies, conducted in the past, determined that the
VFSeffectiveness is influenced byfactors including lengthofthe
VFS, type of vegetation, slope of strip, sediment size distribution
in the runoff, and flow concentration. Many studies consider
strip length the most important parameter affecting sediment
removal efficiency (Munoz-Carpena et al., 1999; Helmers et al.,
2006). Some studies concluded that increasing flow length
beyond 10 m does not increase VFS efficiency by large margins
(Lee et al., 2003; Zreig et al., 2004). However, Gharabhagi et al.
(2001) indicated that smaller sediments take longer to separate
out and require a longer filter. They concluded that the first 5 m
of VFS play a significant role in removing suspended solids and
aggregates larger than 40 mm. Fecal coliform bacteria (FCB) are
facultative, anaerobic, gram negative, non-spore forming, rod-
shaped bacteria smaller than 0.45 mm (Wang, 2003).
Various modeling tools have been developed to evaluate
efficiency of VFS to reduce pathogens during the 1980s (Moore
et al., 1988; Overcash et al., 1983; Springer et al., 1983; Walker
et al., 1990). Still, a wide range of opinions exists on VFS
efficiency in regard to pathogens and/or indicator organisms
(Pachepsky et al., 2006). Targeting and prioritizing the areas for
implementation of BMPs rather than random or general aerial
application is the key to the cost-effective water quality
improvement (Horan and Ribaudo, 1999; Kerr et al., 2007).
Recent watershed water quality physically based and
spatially distributed models can account for physical and
spatial processes. The soil and water assessment tool (SWAT)
water quality model has been applied, calibrated, and
validated for one or more pollutant parameters such as
runoff, sediment yield, and nutrient losses from watersheds at
different geographic locations, conditions, and management
practices (Saleh et al., 1999; Spruill et al., 2000; Santhi et al.,
2001; Kirsch et al., 2002; Van Liew et al., 2003; White et al., 2004;
Qi and Grunwald, 2005; White and Chaubey, 2005; Wang et al.,
2006; Jha et al., 2007; Gassman et al., 2007). Parajuli et al. (2007)
reasonably calibrated and verified the SWAT model using 3
years (2004–2006) of measured daily flow and sediment data.
The model, verified at the Upper Wakarusa watershed in
Kansas, reasonably predicted total fecal bacteria concentra-
tion (coefficient of determination from 0.37 to 0.52 and Nash
Sutcliffe efficiency index from 0.24 to 0.38). Tripathi et al.
(2003) used SWAT to identify and prioritize critical areas on the
basis of average annual sediment yield and nutrient losses.
Mankin et al. (2005) and Tuppad and Mankin (2005) found that
the targeted watershed modeling approach using the SWAT
model was effective at reducing sediment loads from the
Kanopolis watershed in Kansas. However, these studies did
not consider targeting pathogens from agricultural water-
sheds. The SWAT model has not been used for targeting to
reduce fecal bacteria concentrations in agricultural water-
sheds. The SWAT model placed VFS along the edges of HRUs.
This study used the SWAT model to evaluate effectiveness
of different VFS lengths for targeting conservation programs to
reduce NPS pollution. Overall objectives were to characterize
fecal bacteria sources and target BMPs to abate sediment and
fecal bacteria pollution from the Upper Wakarusa watershed.
This study had three specific objectives: (i) test the effective-
ness of various filter strip lengths (0, 10, 15 and 20 m) for
removing overland process sediment and FCB concentration,
(ii) rank sub-watersheds after determining the sediment and
FCB contribution (base condition) of each sub-watershed by
overland process, and (iii) evaluate the effectiveness of VFS
application using both target and random approaches to select
10, 25 and 50% of the sub-watersheds.
2. Material and methods
2.1. Upper Wakarusa watershed
The Upper Wakarusa watershed (Fig. 1) is located in Douglas,
Shawnee, Osage and Wabaunsee counties in Kansas and
consists of 950 km2 with an average elevation of 304 m. The
watershed has three major land uses: grassland (57%),
cropland (28%), and woodland (9%). Silty-clay textured soils
are the predominant soil type in this watershed. The SWAT
model was verified at the Upper Wakarusa watershed.
2.2. Stream monitoring
Stream flow and bacteria data were collected at various water
quality sampling points of the watershed to validate model
results (Fig. 1). Grab samples (about 250 mL) were collected
from the midpoint of the flowing stream at each water quality
sampling point. Samples were placed immediately into an ice
chest and transferred to a laboratory refrigerator within 2–4 h
of collection. Bacteria enumeration procedures were started
within 24 h. The serial dilution method (Clesceri et al., 1998)
was applied to enumerate FCB colonies. Bacterial samples
typically required four serial dilutions to obtain reasonable
bacteria colony counts.
Flow was calculated at the time of sample collection using
Manning’s equation, as outlined by Ward and Elliot (1995).
Fig. 1 – Location map of the Upper Wakarusa watershed in northeast Kansas.
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 0 1191
Flow depth, cross-sectional area, and channel slope were
measured, and the channel roughness factor was estimated
based on channel roughness characteristics, and degree of
meandering (Cowan, 1956). The calculated flow was validated
based on ratio of the watershed area using data from the
United States Geological Survey (USGS) Richland gage station.
Calculated flow data showed very good correlation (>90%)
with weighted area flow data. This study used data collected
from January 2004 to April 2006.
2.3. Fecal bacterial source characterization
2.3.1. LivestockManure from grazing areas, feeding operations, and winter
feeding areas were major bacterial sources in this study.
Livestock population at the county and watershed level was
estimated using agricultural census/GIS layers data (USDA,
2006). The county animal census population was equally
distributed on a total land-area basis to determine the fraction
of total livestock in the study watershed. The USDA data were
compared with data from the Kansas Department of Agri-
culture (KDA, 2004a). The number of animal units (AUs) in
feedlots within the watershed were estimated using active
feedlot data (both federally permitted feedlots >1000 AUs and
state registered feedlots >300 AUs) from the Kansas Depart-
ment of Health and Environment (KDHE) (M. Jepson, personal
communication, 2005). Permitted and registered livestock
were subtracted from the total number of AUs in the
watershed to estimate the net grazed livestock population.
The field reported stocking rate of 3 ha per cow and calf pair
(KDA, 2004b) was used as the baseline value, but this value
could vary due to pasture management activities, animal
growth, and animal sales. Animal stocking rates in the
pastureland also were validated using county-wide livestock
population data (KDA, 2004a). Animals in pasturelands could
be brought from feedlots, barnyards, and leasing agreements
for grazing during the warm season (generally April to
September). However, stocking rate of the animals in pasture-
land was assumed to be constant throughout the grazing
season and decrease during the winter.
The Upper Wakarusa watershed was estimated to be
populated with 27,122 beef AUs in the pastureland (based on
stocking rate), 2982 beef AUs in the feedlots, and 627 swine
AUs and 10,849 beef AUs in the winter feeding areas (40% of
27,122), which was used in this study to represent the current
scenario of the watershed. Manure production by beef cattle
was estimated based on standard production rates (ASAE,
2000) of 26.4 kg of wet manure per day per 1000-kg AU. Actual
manure production by each AU can vary depending on dietary
habit of the animal, reflected in a reported standard deviation
of 17 kg per day for manure estimation (ASAE, 2000). FCB
concentration in manure was estimated based on ASAE (2000),
which reported a concentration of 13 � 1010 colony forming
units (cfu) per day per AU (wet-weight-basis) from the beef
manure with a standard deviation of 12 � 1010 cfu per day per
AU. Bacteria concentration was converted into model-input
units of cfu per gram of dry-weight manure using a standard
mean manure moisture content of 86% (ASAE, 2000).
Pastureland was simulated under two major grass-type
management conditions, which represent typical field condi-
tions: grazed (80%, typically native prairie) and non-grazed
(20%, typically smooth brome and tall fescue). Native prairie
grass typically is not fertilized, but tall fescue is fertilized with
70–15–0 (NPK) (W. Boyer, personal communication, 2006). It was
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 01192
estimated that about 1.81 kg per ha per day dry weight of
manure is applied in the pasturelands due to grazing operation
during the growing season and cattle are grazed for 153 days in
the pastureland (W. Boyer, personal communication, 2006). It is
possible that actual animal density varies in the watershed
every day due to animalgrowth and the pattern of incoming and
outgoing animals. It was estimated that about 20% of the air-dry
biomass is trampledevery day and about 341 kgof air-dry forage
is required for an AU for 30 days (Paul and Watson, 1994). In this
watershed, grazing starts about a month earlier in tall fescue
grasslands than in native prairie grass (W. Boyer, personal
communication, 2006). The entire native prairie is grazed, but
only 80% of the tall fescue is grazed; the remaining tall fescue is
used for haying or the Conservation Reserve Program (CRP).
About 3.7 Mg per ha of hay is harvested annually from the un-
grazed area, but biomass is not removed from the CRP land (W.
Boyer, personal communication, 2006). Because cattle do not
graze pastureland from October to March, no biomass uptake
from the pastureland occurred, with no grass trampling and no
manure deposition on the soil during this period.
All source loads due to livestock in confined animal feedlots
were modeled to be land-applied as grazing operations in
pasturelands of the sub-watershed where the active permitted
feedlots were located. Winter feeding areas were modeled
assuming that all livestock were confined within 40% of the
grazed land of the watershed based on observed animal
congregating behavior (W. Boyer, personal communication,
2006). Animals in feedlots and winter-feeding areas contrib-
uted fecal bacteria for 212 days during the dormant season of
the year (generally October to March). It was estimated that
about 4.52 kg per ha per day dry weight of cattle manure (2.5
times greater than regular pastureland operation) was applied
in the respective pasturelands of the sub-watersheds due to
winter-feeding operations.
2.3.2. HumanDigital orthophoto quarter quadrangles of the watershed from
2002 were digitized depending on the physical context, roads,
and type of houses to represent each septic system in the
watershed. Each rural house was assumed to have one septic
system, resulting in a total of 2137 septic systems in the Upper
Wakarusa watershed. About 40% of the estimated septic
systems (n = 855) were assumed failing in the watershed
(KDHE, 2000). Each septic system was assumed to be used by
three persons in the household that can contribute about
0.32 m3 of sewage effluent per household per day (USEPA,
2001). Failing septic systems in the watershed were modeled
using two techniques: effluent was either land-applied in the
non-grazing grassland areas or input as a direct-daily point
load to the outlet of the each sub-basin. The FCB concentration
in failing septic system effluent was taken as 6.3 � 106 cfu per
100 mL (Overcash and Davidson, 1980).
2.3.3. WildlifeNo comprehensive wildlife inventory was available for the
Upper Wakarusa watershed. Wildlife population density was
estimated based on information received from the Kansas
Department of Wildlife and Parks (KDWP). The 2002 summer
road-kill indices survey data (M. Peek, personal communica-
tion, 2005) for Kansas were used to estimate small mammal
populations in the watershed. Information included data for
various wild animal species: raccoon, opossum, striped skunk,
coyote, badger, bobcat, red fox, gray fox, swift fox, beaver,
mink, muskrat, river otter, spotted skunk, weasel, armadillo,
woodchuck, and porcupine. The population of raccoon,
opossum, striped skunk, and coyote constituted about 81%
of the total surveyed small mammals in Kansas (Timm et al.,
2007). Population of the predominate large mammal (white-
tailed deer) in the watershed was estimated based on expert
opinion from the KDWP big-game coordinator (L. Fox, personal
communication, 2006). Similar data were collected for the
predominate indigenous avian species (turkey) from the
KDWP small game coordinator (J. Pitman, personal commu-
nication, 2006), and for migratory birds (duck, geese, sandhill
crane) from the KDWP waterfowl biologist (M. Kraft, personal
communication, 2006).
To estimate the AUs of each wildlife species in the
watershed, population data were first distributed over the
potential habitat for each species. Small mammals and turkey
population data were counted from a road survey. Most of the
small mammals were counted dead at the road shoulder. Sight
distances of 5 m for small mammals and 50 m for turkey from
each side of the road were assumed, and the population density
of each species was estimated as number of animals per unit
area, using total length of the road driven during survey. For
deer, the number of deer harvested in northeastern Kansas was
estimated and equally distributed in the total land area of
northeastern Kansas to determine deer population density.
Indigenous turkey population was equally distributed in the
woodland, cropland, and pastureland areas of the watershed as
data was collected by rural mail carriers from those land use
areas (J. Pitman, personal communication, 2006). Migratory
birds are present in the watershed for about seven months of
the year (January–March and September–December). The
Kansas population of these wildlife species was equally
distributed over the water surface area and wetland area to
determine population density. Population density in the
watershedwasestimatedbased onavailablewatersurface area.
Animal weights were estimated based on information
received from ‘‘Mammals of Kansas’’ (Timm et al., 2007) and
personal communication (J. Pitman, personal communication,
2006). Population in 1000-kg AUs of about 456 migratory birds,
760 turkey AUs, 304 deer AUs, and 89 small mammal AUs was
used to represent the current scenario of the Upper Wakarusa
watershed. All wildlife-generated manure was applied to the
woodland, cropland, and pastureland areas depending on
wildlife habitats to be considered as the baseline scenario.
2.4. Crops and vegetative filter strips
Corn and soybeans were major warm-season crops, and
winter wheat was the primary cool-season crop grown in 3
year’s rotation in the watershed (W. Boyer, personal commu-
nication, 2006). The warm season crop was planted on May 1
and harvested on October 1. The cool-season crop was planted
on October 20 and harvested on July 30. These dates represent
typical planting and harvesting dates in the watershed. Crop
residue was left on the cropland between crop periods. The
conservation tillage system is the most widely adopted system
for corn, soybean, and wheat in the watershed. Croplands of
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 0 1193
the watershed were simulated with a 5-m wide filter strip at
the edge of HRUs, which is smaller than Natural Resources
Conservation Service (NRCS) guidelines but selected to provide
an average sediment removal rate of 59%, consistent with local
field buffer strip data (Ngandu, 2004).
This study used the previously calibrated and validated
SWAT model in the Upper Wakarusa watershed. The VFS was
considered one of the BMPs in this study. Various VFS lengths
(0, 5, 10 and 20 m) were evaluated for effectiveness at reducing
FCB transport through the overland flow process. After the
SWAT model was tested for various VFS lengths for the whole
watershed, the recommended VFS length (15 m) was applied
in selected (0, 10, 25, 50 and 100%) sub-watersheds with both
target and random methods.
2.5. SWAT model
The SWAT model uses geospatially referenced data to satisfy
the necessary input parameters. USGS (1999) 30 m � 30 m
elevation data was used to delineate the watershed bound-
aries and topography. The Soil Survey Geographic Database
was used to create a soil database (USDA, 2005). The Gap
Analysis Program (GAP) 2001 land cover data that depicts 20
general land cover classes for the state of Kansas (KARS, 2001)
was used. Wardlow and Egbert (2003) evaluated GAP and
National Land Cover Data (NLCD) land use data (USGS, 2000)
for the state of Kansas. Kansas GAP data provided better
discrimination of most land cover classes compared with
NLCD. Accuracy assessment found an overall accuracy of 87%
for GAP and 81% for NLCD, and GAP had higher accuracies for
most individual land cover classes. The Kansas GAP and NLCD
land cover products were comparable in terms of characteriz-
ing broad scale land cover patterns, but the Kansas GAP land
cover map appears to be more appropriate for localized
applications that require detailed and accurate land cover
information.
Land use classes were re-classified into eight classes
(grazedland, non-grazedland/hay, cropland, woodland, CRP,
water, urban areas, and quarry) based on field-verified land
use conditions (Mankin and Koelliker, 2001; Mankin et al.,
2003). The stream threshold area was defined as 950 ha, which
is about 10% of the total watershed area (950 km2). The SWAT
model delineated 53 sub-basins ranging in area from
0.003 km2 to 66.709 km2. Watershed parameters for each
HRU in each watershed were defined on the basis of soil,
land use, and topographic characteristics of the watershed as
described in the SWAT 2005 documentation (Neitsch et al.,
2005).
The microbial survival and transport sub-model was added
to the SWAT model in 2000 and modified in 2005. The SWAT
model microbial component considers the fate and transport
of organisms for bacterial concentration. The microbial sub-
model uses Chick’s Law, as revised by Moore et al. (1989), to
model fecal bacteria die-off and re-growth. Chick’s law, a first
order decay equation, determines the quantity of bacteria that
are removed or added by die-off and re-growth as described by
Sadeghi and Arnold (2002):
Ct ¼ C0 � e�K20t uðT�20Þ(1)
where Ct = bacteria concentration at time t, count/100 mL;
C0 = initial bacteria concentration, count/100 mL; K20 = first-
order die-off rate at 20 8C, day�1; t = exposure time, days;
u = temperature adjustment factor; T = temperature, 8C.
In-stream bacteria die-off is the only process modeled in
SWAT (Neitsch et al., 2005). SWAT calculates loading of
pathogens and indicator bacteria for pathogens from land
areas in the watershed. The VFS are generally provided at the
edge of the field and can be defined in an HRU. Sediment and
bacteria loads in surface runoff from overland flow process are
reduced as the surface runoff passes through the filter strip.
Filter strip trapping efficiency for fecal bacteria (Moore et al.,
1988) is calculated by
trapef;bact ¼11:8þ 4:3 widthfiltstrip
100(2)
where trapef,bact is the fraction of the bacteria loading trapped
by the filter strip, and widthfiltstrip is the width of the filter strip
(m).
Looking at Eq. (2), 20.5 m is the maximum VFS length that
can be used in the model to remove 100% of fecal bacteria.
Eq. (2) is recommended not to exceed more than 75% of
bacteria removal, and the equation is still being researched.
VFS trapping efficiency for sediment is calculated by
trapef ¼ 0:367 ðwidthfiltstripÞ0:2967 (3)
where trapef is the fraction of the constituent loading
trapped by the filter strip, and widthfiltstrip is the width of
the filter strip (m).
2.6. Weather and hydrologic data
Weather data such as daily precipitation and daily ambient
temperatures were extracted from the National Climatic
Data Center, and records maintained by the Kansas state
climatologist also were used. Daily climate data were used
from nine weather stations near the watershed: Clinton
Lake, Auburn, Silver Lake, Overbrook, Eskridge, Pomona
Lake, Lawrence, Lecompton, and Topeka (Fig. 1) but only five
weather stations had complete daily precipitation data. The
Silver Lake weather database was used for daily solar
radiation, daily wind speed, and daily relative humidity
data. Missing data were adjusted using SWAT database
simulation. For simulation, the SWAT model uses data from
the Ottawa weather station (Franklin County), which is
located about 24 km south-east from the nearest point of
the watershed. Average annual rainfall data for 2004 to
2006 was measured for Overbrook (939 mm), Lecompton
(955 mm), Topeka (1008 mm), Auburn (935 mm) and Eskridge
(942 mm), which had complete daily precipitation data, are
given in Table 1.
3. Results and discussion
The SWAT model-predicted daily flow and sediment reason-
ably matched measured values during previous calibration
and verification studies for the Upper Wakarusa watershed
(Parajuli et al., 2007).
Table 1 – Description of weather stations rainfall data used in the study
Station name Yeara Total(mm per year)
Peak rainfall(mm per day)
Growingseason (%)b
No. ofrainfall-eventsc
No. of potentialrunoff-eventsd
Overbrook 2004 1126 69 69 83 29
2005 1180 81 77 70 23
2006 512 53 72 44 12
Lecompton 2004 1206 83 69 102 28
2005 1029 87 72 81 23
2006 629 40 68 59 18
Topeka 2004 1013 62 67 91 25
2005 1239 142 74 82 24
2006 772 64 81 62 20
Auburn 2004 1146 114 70 93 27
2005 1013 82 71 76 19
2006 646 60 73 59 17
Eskridge 2004 960 67 73 85 25
2005 1185 125 75 87 28
2006 681 58 75 65 16
a Year 2006 rainfall data from January to October only.b % of rainfall between April to September.c Rainfall greater than 1 mm.d Rainfall events greater than 14 mm.
Fig. 2 – Watershed sub-basins and annual average overland flow fecal bacteria response.
Fig. 3 – Watershed sub-basins and annual average overland sediment yield response.
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 01194
Fig. 4 – Distribution of watershed sub-basins for fecal coliform bacteria BMP selection: target method (a) 10%, (b) 25%, (c) 50%
and random method (d) 10%, (e) 25%, and (f) 50%.
Fig. 5 – Distribution of watershed sub-basins for sediment BMP selection: target method (a) 10%, (b) 25%, (c) 50% and random
method (d) 10%, (e) 25%, and (f) 50%.
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 0 1195
Table 2 – Fecal bacteria and sediment yield ranking fortargeting
Rank Fecal bacteria Sediment yield
Sub-basin
Bacteria(cfu per 100 mL)
Sub-basin
Sediment(Mg per ha)
1 49 5460 48 13.22
2 44 5407 46 10.18
3 34 5163 30 8.42
4 21 5011 38 8.25
5 17 4954 32 7.83
6 48 4573 39 7.73
7 43 4532 35 7.44
8 50 4456 16 6.58
9 36 4397 36 6.47
10 46 4293 18 5.86
11 35 4074 31 3.11
12 45 3936 44 1.65
13 42 3693 34 1.53
14 32 3626 1 1.48
15 41 3523 42 1.42
16 40 3144 5 1.25
17 37 3008 23 1.19
18 47 2834 3 1.12
19 16 2724 22 1.07
20 24 2723 20 1.04
21 25 2629 4 1.03
22 39 2610 27 0.96
23 30 2580 2 0.88
24 33 2519 17 0.85
25 26 2387 29 0.84
26 1 2185 21 0.82
27 31 2059 19 0.79
28 38 1989 14 0.73
29 3 1872 47 0.72
30 4 1667 40 0.69
31 5 1599 52 0.66
32 20 1397 49 0.64
33 19 1361 26 0.64
34 22 1334 51 0.61
35 23 1202 33 0.61
36 2 1153 25 0.57
37 18 864 24 0.56
38 6 770 53 0.54
39 11 636 6 0.51
40 51 630 50 0.50
41 10 561 7 0.48
42 7 539 43 0.46
43 8 535 10 0.45
44 28 514 37 0.45
45 9 483 15 0.43
46 27 375 45 0.43
47 53 366 28 0.37
48 13 346 11 0.31
49 14 345 12 0.25
50 29 313 41 0.18
51 15 313 9 0.17
52 52 219 8 0.13
53 12 189 13 0.10
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 01196
3.1. Vegetative filter strip length
Fig. 2 shows that increasing the length of a VFS in model
simulation reduces average overland annual FCB concentra-
tion from the watershed. The first 10 m of VFS reduced about
57% of FCB. The additional 5 m of a 15-m VFS removed up to
80% of the FCB, which was reasonable. The 20-m VFS removed
about 100% of FCB, which was anticipated based on Eq. (2).
Fig. 3 shows a similar trend. About 73% of sediment yield was
reduced with a 10-m VFS. The 15-m VFS removed up to 82% of
sediment yield, and the 20-m VFS removed up to 89% of
sediment yield, which was anticipated based on Eq. (3). Figs. 2
and 3 showed more noticeable results using Eqs. (2) and (3) in
the model. A 20-m VFS removed 100% of the FCB whereas 100%
reduction of sediment needed 30 m of VFS.
Parajuli et al. (2005) tested different lengths of VFS (0–25 m)
using SWAT 2000 and suggested that 20-m VFS were most
effective at reducing FCB concentration in the agricultural
watershed. They did not find any further reduction in FCB
concentration using 25-m VFS. However, using Eq. (2), which
was changed slightly in SWAT 2005, was recommended for up
to 75% bacteria removal efficiency. The study of Moore et al.
(1988) also recommends 15-m VFS.
For the target method, sub-basins were selected separately
for FCB concentration and sediment contribution by overland
flow based on ranking (Figs. 4 and 5). Sub-watersheds were
ranked separately for sediment and FCB concentration
prediction because model-identified top ranking sub-water-
sheds for sediment and fecal bacteria were different (Table 2),
which was anticipated. The SWAT model uses the modified
universal soil loss equation (Williams, 1995) to estimate
sediment yield. The slope and slope length parameters in
this equation can directly affect sediment yield estimates.
Generally, average slope of each HRU/sub-watershed influ-
enced sediment prediction, and total water yield from each
HRU/sub-basin influenced FCB concentration prediction.
3.2. Fecal bacteria and sediment yields
3.2.1. Overland flow outputRemoval of annual average FCB concentration and sediment
yields by overland flow process was affected due to VFS
adoption. When VFS were applied in 100% of sub-basins in the
watershed, reduction in annual average FCB concentration
was about 79%, from 2134 to 438 cfu per 100 mL. Reduction in
FCB concentration due to 10% VFS adoption was 20% for the
target approach and about 12% for the random approach.
The greatest difference between the target and random
approach to FCB concentration reduction was at 50% VFS
adoption; the target approach removed about 60% of FCB
concentration, and the random approach removed about
42% (Fig. 6). In this watershed, greater amounts of FCB
came from grazedlands than from non-grazedlands or
cropland. Grazedlands were fairly equally distributed in
the watershed. Because many sub-watersheds were ranked
(Table 2) closely with regard to FCB concentration, the
greatest reduction occurred only when 50% of the sub-
watersheds adopted VFS. Annual average sediment yield
removal by overland flow process also was affected by VFS
adoption. When VFS were applied in 100% of sub-basins in
the watershed, reduction in annual average sediment yield
was about 82%, from 2.17 to 0.39 Mg per ha.
Reduction in sediment yield due to 10% VFS adoption was
46% for the target approach and about 28% for the random
approach. The greatest difference between the target and
random approach to sediment yield reduction was at 25% VFS
Fig. 6 – Watershed sub-basins and annual average
overland flow fecal bacteria response.
Fig. 7 – Watershed sub-basins and annual average
overland flow sediment yield response.
Fig. 8 – Watershed sub-basins and annual average
overland and watershed outlet fecal bacteria response.
Fig. 9 – Watershed sub-basins and annual average
overland flow and watershed outlet sediment yield
response.
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 0 1197
adoption; the target approach removed about 63% of the
sediment yield, and the random approach removed about 33%
(Fig. 7). Sediment yield was mostly dependent on slope of the
HRU/sub-basin, and percentage of cropland area in the sub-
basins. Generally, longer VFS can reduce more sediment and
FCB than shorter VFS.
3.2.2. Watershed outlet flow outputApplying VFS to the edge of HRUs did not yield measurable
differences in annual average FCB concentration at the outlet
of the watershed (Richland outlet). Applying VFS had good
results only in the overland flow process. Because estimated
direct point loads from livestock (8.5%), human (10%), and
wildlife (10%) sources were input in the model, direct point
loads dominated total FCB prediction at the outlet of the
watershed. Fecal bacteria source loads applied on the land
were determined to have no or less sensitivity in a previous
study (Parajuli et al., 2007). Direct point loads were more
sensitive to channel flow process in the model. Interestingly,
contribution of overland flow process FCB concentration at 0%
VFS adoption (log transformed value = 3.3) was less than 1% of
total FCB concentration (log transformed value = 5.79) pre-
dicted at the outlet of the watershed (Richland) where direct
point source loads were input (Fig. 8). The NPS loads of FCB had
chances of reduction through die-off and sorption during the
overland flow process, but point loads had chances of
reduction through only the channel flow process.
Annual average sediment yield removal at the outlet of the
watershed (compared with the Richland outlet) was about
36% when VFS were applied in 100% of the sub-basins.
Reduction in sediment yield due to 10% VFS adoption was 12%
for the target approach and about 2% for the random
approach. The greatest difference between the target and
random approach to sediment yield reduction was at 50% VFS
adoption; the target approach removed about 22% of the
sediment yield, and the random approach removed about 7%
(Fig. 9).
4. Conclusions
The concept of identifying, selecting, and targeting critical
areas of point and NPS pollution for pollutant reduction has
been widely recognized and studied. In this study, a watershed
modeling approach was used to quantify effects of imple-
menting a single BMP on incremental increases in watershed
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 01198
sub-basins and evaluate effectiveness of a targeted approach
compared with a random approach at reducing estimated
pollutant loadings, both overland and at the watershed outlet
(Richland). Priority areas for the targeted approach were
selected based on model-predicted sediment yield and FCB
concentration. A targeted watershed modeling approach
using SWAT was found to be effective at reducing sediment
yield load both overland and at the watershed outlet, but this
approach was effective at reducing FCB only from overland
flow. This study also used SWAT to evaluate effectiveness of
various VFS lengths in removing FCB concentration and
sediment yield from the agricultural watershed. The model
determined that applying VFS using a target approach can be
more cost-effective than a random approach. Other BMPs,
such as stream fencing to reduce cattle access to the stream or
rotational grazing, and variable percentages of direct point
loads could be tested in future studies.
Acknowledgements
This material is based on work supported by the USDA
Cooperative State Research, Education, and Extension Service
under agreement no. 2003-04949, 2003-51130-02110. We
acknowledge the contributions of Mr. Will Boyer, extension
watershed specialist at Kansas State University; Mr. Mark
Jepson, environmental scientist at the Kansas Department of
Health and Environment (KDHE); and Mr. Matt Peek, furbearer
biologist, Dr. Lloyd Fox, big game coordinator, Mr. Jim Pitman,
small game coordinator, and Mr. Marvin Kraft, waterfowl
research biologist, at the Kansas Department of Wildlife and
Parks (KDWP). Contribution no. 08-284-J from the Kansas
Agricultural Experiment Station, Manhattan, KS.
r e f e r e n c e s
ASAE, 2000. Standard D384.1: Manure production andcharacteristics. St. Joseph, MI.
Benham, B.L., Zeckoski, R.W., Mankin, K.R., Pachepsky, Y.A.,Sadeghi, A.M., Brannan, K.M., Soupir, M.L., Habersack, M.J.,2006. Modeling bacteria fate and transport in watersheds tosupport TMDLs. Trans. ASABE 49 (4), 987–1002.
Bracmort, K.S., Engel, B.A., Frankenberger, J.R., 2004. Evaluationof structural best management practices 20 years afterinstallation: Black Creek Watershed, IN. J. Soil WaterConserv. 59 (5), 659–667.
Clesceri, L., Greenberg, A., Eaton, A., 1998. Standard Methods forthe Examination of Water and Wastewater. Am. PublicHealth Assoc./United Book Press, Inc., Baltimore, MD.
Corwin, D.L., Vaughan, P.J., 1997. Modeling non-point sourcepollutants in the vadose zone with GIS. Environ. Sci.Technol. 31 (8), 2157–2175.
Cowan, W.L., 1956. Estimating hydraulic roughness coefficients.Agr. Eng. (July).
Dickinson, W.T., Rudra, R.P., Wall, G.J., 1990. Targeting remedialmeasures to control non-point source pollution. WaterResour. Bull. 26 (3), 499–507.
Gassman, P.W., Reyes, M.R., Green, C.H., Arnold, J.G., 2007. Thesoil and water assessment tool: historical development,applications, and future research directions. Trans. ASABE50 (4), 1211–1250.
Gharabhagi, B., Rudra, R.P., Whiteley, H.R., Dickinson, W.T.,2001. Sediment removal efficiency of vegetative filter strips.ASAE Meeting Paper no. 012071. ASAE, St. Joseph, MI.
Guber, A.K., Pachepsky, Y.A., Sadeghi, A.M., 2007. Evaluatinguncertainty in E. coli retention in vegetated filter strips inlocations selected with SWAT simulations. Watershedmanagement to meet water quality standards and TMDL(total maximum daily load) Proceedings. ASABE Publicationno. 701P0207. ASAE, St. Joseph, MI.
Heatwole, C.D., Bottcher, A.B., Baldwin, L.B., 1987. Modelingcost-effectiveness of agricultural non-point pollutionabatement programs on two Florida basins. Water Resour.Bull. 23 (1), 127–131.
Helmers, M.J., Isenhart, T., Dosskey, M., Dabney, S., Strock, J.,2006. Buffers and vegetative filter strips. Available at: http://www.epa.gov/msbasin/taskforce/2006symposia/4BuffersVegHelmers.pdf (accessed April 23, 2007).
Horan, R., Ribaudo, M., 1999. Policy objectives and economicincentives for controlling agricultural sources of nonpointpollution. J. Am. Water Resour. Assoc. 35 (6), 1023–1035.
Inamdar, S.P., Mostaghimi, S., McClellan, P.W., Brannan, K.M.,2001. BMP impacts on sediment and nutrient yields from anagricultural watershed in the coastal plain region. Trans.ASAE 44 (5), 1191–1200.
Jha, M.K., Gassman, P.W., Arnold, J.G., 2007. Water qualitymodeling for the Raccoon River watershed using SWAT.Trans. ASABE 50 (2), 479–493.
KARS (Kansas Applied Remote Sensing), 2001. GAP AnalysisProgram (GAP). Available at http://www.kansasgis.org/catalog/catalog.cfm (accessed March 29, 2006).
KDA (Kansas Department of Agriculture), 2004a. Kansasagricultural statistics. Kansas farm facts 2004. Available athttp://www.nass.usda.gov/ks/ffacts/2004/pdf/ffpdf.htm(accessed May 30, 2006).
KDA (Kansas Department of Agriculture), 2004b. Kansasagricultural statistics services. Bluestem Pasture. SpecialPress Release. April 22, 2004. Topeka, KS.
KDHE (Kansas Department of Health and Environment), 2000.Bureau of water. Clean water needs survey. Kansas Non-point Source Need Inventory, Methodology andAssumptions, Topeka, KS.
Kerr, J., Quyang, D., Bartholic, J., 2007. Targeting watershedinterventions for reduction of non-point source pollution.Available at http://www.iwr.msu.edu/rusle/doc/stony.htm(accessed May 21, 2007).
Kirsch, K., Kirsch, A., Arnold, J.G., 2002. Predicting sediment andphosphorus loads in the Rock River basin using SWAT.Trans. ASAE 45 (6), 1757–1769.
Lee, K.H., Isenhart, T.M., Schultz, R.C., 2003. Sediment andnutrient removal in an established multi-species riparianbuffer. J. Soil Water Conserv. 58 (1), 1–8.
Mankin, K.R., Koelliker, J.K., 2001. Clinton lake water qualityassessment project—final report. KDHE contract numberNPS 98-059. Kansas Department of Health andEnvironment, Bureau of Water, Topeka, KS.
Mankin, K.R., Wang, S.H., Koelliker, J.K., Huggins, D.G.,deNoyelles, F., 2003. Watershed-lake water qualitymodeling: verification and application. J. Soil WaterConserv. 58 (4), 188–197.
Mankin, K.R., Tuppad, P., Devlin, D.L., McVay, K.A., Hargrove,W.L., 2005. Strategic targeting of watershed managementusing water quality modeling. In: Brebbia, C.A., Antunes doCarmo, J.S. (Eds.), River Basin Management III, Bologna,Italy. WIT Press, Southampton, UK, pp. 327–337.
Moore, J.A., Smyth, J., Baker, S., Miner, J.R., 1988. Evaluatingcoliform concentrations in runoff from various animalwaste management systems. Special Report 817.Agricultural Experiment Station, Oregon State University,Corvallis, OR.
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 0 1199
Moore, J.A., Smyth, J.D., Baker, E.S., Miner, J.R., Moffitt, D.C.,1989. Modeling bacteria movement in livestock manuresystems. Trans. ASAE 32 (3), 1049–1053.
Moore, L.W., Chew, C.Y., Smith, R.H., Sahoo, S., 1992. Modelingof best management practices on North Reelfoot Creek, TN.Water Environ. Res. 64 (3), 241–247.
Munoz-Carpena, R., Parsons, J.E., Gilliam, J.W., 1999. Modelinghydrology and sediment transport in vegetative filter strips.J. Hydrol. 214, 111–129.
Neitsch, S.L., Arnold, J.G., Kiniry, J.R., Williams, J.R., 2005. Soiland Water Assessment Tool (SWAT), TheoreticalDocumentation. Blackland Research Center, Grassland, Soiland Water Research Laboratory, Agricultural ResearchService, Temple, TX.
Ngandu, D.M., 2004. Grass-shrub riparian buffer removal ofsediment, phosphorus and nitrogen from surface runoff.Master Thesis, Department of Biological and AgriculturalEngineering, Kansas State University, Manhattan, KS.
Niu, Z., Sun, G., McNulty, S.G., Xie, M., Byne, W., 2001. ApplyingANSWERS-2000 to simulate BMP effects on sediment andrunoff from two watersheds in the Three Gorges area,Southern China. In: Proceedings of the Conference of SoilErosion Research for 21st Century, Honolulu, HI.
Novotny, V., Olem, H., 1994. Water Quality: Prevention,Identification, and Management of Diffuse Pollution. VanNostrand Reinhold, New York, NY.
Overcash, M.R., Davidson, J.M., 1980. Environmental Impact ofNon-point Source Pollution. Ann Arbor Science Publishers,Inc., Ann Arbor, MI.
Overcash, M.R., Reddy, K.R., Khaleel, R., 1983. Chemicalprocesses and transport of animal waste pollutants. In:Schaller, F.W., Bailey, G.W. (Eds.), Agricultural Managementand Water Quality. Iowa State University Press, Ames, IA,pp. 109–125.
Pachepsky, Ya.A., Sadeghi, A.M., Bradford, S.A., Shelton, D.R.,Guber, A.K., Dao, T.H., 2006. Transport and fate of manure-borne pathogens: modeling perspective. Agric. WaterManage. 86, 81–92.
Parajuli, P., Mankin, K.R., Barnes, P.L., 2005. EvaluatingBacteria Removal from Land Application of Manure usingSWAT. ASABE Meeting Paper no. MC05-203. ASABE, St.Joseph, MI.
Parajuli, P., Mankin, K.R., Barnes, P.L., 2007. New methods inmodeling source-specific bacteria at watershed scale usingSWAT. Watershed Management to meet Water QualityStandards and TMDLs (Total Maximum Daily Load)Proceedings. ASABE Publication no. 701P0207. ASABE, St.Joseph, MI.
Park, S.W., Mostaghimi, S., Cooke, R.A., McClellan, P.W., 1994.BMP impacts on watershed runoff, sediment and nutrientyields. Water Resour. Bull. 30 (6), 1011–1023.
Paul, D., Watson, S.L., 1994. Stocking Rate and GrazingManagement, MF-1118. Kansas State UniversityAgricultural Experiment Station and Cooperative ExtensionService, Manhattan, KS.
Qi, C., Grunwald, S., 2005. GIS-based hydrologic modeling in theSandusky watershed using SWAT. Trans. ASAE 48 (1), 160–180.
Sadeghi, A.M., Arnold, J.G., 2002. A SWAT/Microbial sub-modelfor predicting pathogen loadings in surface andgroundwater at watershed and basin scales. TotalMaximum Daily Load (TMDL) Environmental Regulations.ASAE Publication no. 701P0102. ASAE, St. Joseph, MI.
Saleh, A., Arnold, J.G., Gassman, P.W., Hauck, L.W., Rosenthal,W.D., Williams, J.R., MacFarland, A.M.S., 1999. Applicationof SWAT for the Upper North Bosque River watershed.Trans. ASAE 43 (5), 107–1087.
Santhi, C., Arnold, J.G., Williams, J.R., Hauck, L.M., Dugas, W.A.,2001. Application of a watershed model to evaluate
management effects on point and non-point sourcepollution. Trans. ASAE 44 (6), 1559–1570.
Springer, E.P., Gifford, G.F., Windham, M.P., Thelin, R., Kress, M.,1983. Fecal Coliform Release Studies and Development of aPreliminary non-point Source Transport Model for IndicatorBacteria. Utah Water Research Laboratory, Utah StateUniversity, Logan, UT.
Spruill, C.A., Workman, S.R., Taraba, J.L., 2000. Simulation ofdaily and monthly stream discharge from smallwatersheds using the SWAT model. Trans. ASAE 43 (6),1431–1439.
SWCS (Soil and Water Conservation Society), 2001. Realizing thepromise of conservation buffer technology. Soil and WaterConservation Society, Ames, IA.
Tim, U.S., Mostaghimi, S., Shanholtz, V.O., 1992. Identificationof critical nonpoint pollution source areas using geographicinformation systems and water quality modeling. WaterResour. Bull. 28 (5), 877–887.
Timm, R.M., Pisani, G.R., Slade, N.A., Choate, J.R., Kaufman, G.A.,Kaufman, D.W., 2007. Mammals of Kansas. Available athttp://www.kS.ku.edu/libres/Mammals_of_Kansas(accessed February 17, 2007).
Tripathi, M.P., Panda, R.K., Raghuwanshi, N.S., 2003.Identification and prioritization of critical sub-watershedsfor soil conservation management using the SWAT Model.Biosyst. Eng. 85 (3), 365–379.
Tuppad, P., Mankin, K.R., 2005. Cost-effective watershedmodeling approach to achieve TMDL targets. Watershedmanagement to meet water quality standards andemerging TMDL (Total Maximum Daily Load) Proceedings.ASAE Publication no. 701P0105. ASAE, St. Joseph, MI.
USDA (U.S. Department of Agriculture), 2005. Soil Data Mart.Natural Resources Conservation Service. Available at http://soildatamart.nrcs.usda.gov/Default.aspx (accessed May 29,2006).
USDA (U.S. Department of Agriculture), 2006. National Atlas ofthe United States. Census of the United States. Available athttp://www.nationalatlas.gov/atlasftp.html (accessed May29, 2006).
USEPA, 2001. Handbook for Developing Watershed Plans toRestore and Protect Our Waters. EPA841-B-05-005. U.S.Environmental Protection Agency, Washington, D.C.
USEPA, 2006. Total Maximum Daily Loads. National Section303(d) list fact sheet. U.S. Environmental Protection Agency,Washington, D.C. Available at http://oaspub.epa.gov/waters/national_rept.control (accessed July 19, 2006).
USGS (U.S. Geological Survey), 1999. National Elevation Dataset.Available at http://www.kansasgis.org/catalog/catalog.cfm(accessed March 29, 2006).
USGS (U.S. Geological Survey), 2000. NLCD Land Cover ClassDefinitions, USGS EROS Data Center, Sioux Falls, SouthDakota. Available at http://eros.usgs.gov/products/landcover/nlcd.html (accessed August 5, 2003).
Van Liew, M.W., Arnold, J.G., Garbrecht, J.D., 2003. Hydrologicsimulation on agricultural watersheds choosing betweentwo models. Trans. ASAE 46 (6), 1539–1551.
Veith, T.L., Wolfe, M.L., Heatwole, C.D., Bosch, D.J., 2001.Watershed level optimization of BMP placement for cost-effective NPS pollution reduction. ASAE Meeting Paper no.013126. ASAE, St. Joseph, MI.
Vennix, S.A., Northcott, W.J., 2002. Prioritizing vegetative bufferstrips within agricultural watershed. ASAE Meeting Paperno. 022135. ASAE, St. Joseph, MI.
Walker, S.E., Mostaghimi, S., Dillaha, T.A., Woeste, F.E., 1990.Modeling animal waste management practices: impacts onbacteria levels in runoff from agricultural lands. Trans.ASAE 33, 807–817.
Wang, L., 2003. Fecal bacteria survival in cow manure andEscherichia coli release and transport through porous media.
a g r i c u l t u r a l w a t e r m a n a g e m e n t 9 5 ( 2 0 0 8 ) 1 1 8 9 – 1 2 0 01200
PhD dissertation. Kansas State University, Departmentof Biological and Agricultural Engineering, Manhattan,KS.
Wang, X., Melesse, A.M., Yang, W., 2006. Influences of potentialevapotranspiration estimation methods on SWAT’shydrologic simulation in a northwestern Minnesotawatershed. Trans. ASABE 49 (6), 1755–1771.
Ward, A.D., Elliot, W.J., 1995. Environmental Hydrology. LewisPublishers, New York.
Wardlow, B.D., Egbert, S.L., 2003. A comparative analysis of theGAP and NLCD land cover data sets for Kansas.Photogramm. Eng. Remote Sens. 69 (12), 1387–1397.
White, K.L., Chaubey, I., 2005. Influence of hydrologic responseunit (HRU) distribution on SWAT model flow and sedimentpredictions. Watershed Management to Meet Water QualityStandards and Emerging Total Maximum Daily Load (TMDL)
Proceedings. ASAE Publication no. 701P0105. ASAE, St.Joseph, MI.
White, K.L., Chaubey, I., Haggard, B.E., Matlock, M.D., 2004.Comparison of two methods for modeling monthly TP yieldfrom a watershed. ASAE Meeting Paper no. 042162. ASAE, St.Joseph, MI.
Williams, J.R., 1995. Chapter 25: the EPIC model. In: Singh, V.P.(Ed.), Computer Models of Watershed Hydrology. WaterResource Publications, Littleton, CO, pp. 909–1000.
Zreig, M.A., 2001. Factors affecting sediment trapping invegetated filter strips: simulation study using VFSMOD.Hydrol. Process. 15, 1477–1488.
Zreig, M.A., Rudra, R.P., Lalonde, M.N., Whiteley, H.R., Kaushik,N.K., 2004. Experimental investigation of runoff reductionand sediment removal by vegetated filter strips. Hydrol.Process. 18, 2029–2037.