Science of The Total Environment

Science of the Total Environment 669 (2019) 512–526
Contents lists available at ScienceDirect
Science of the Total Environment
journal homepage: www.elsevier.com/locate/scitotenv
Review
A review of pesticide fate and transport simulation at watershed level

using SWAT: Current status and research concerns
Ruoyu Wang a, Yongping Yuan b, Haw Yen c, Michael Grieneisen a, Jeffrey Arnold c, Dan Wang d,
Chaozi Wang e,⁎, Minghua Zhang a,⁎
a
Department of Land, Air and Water Resources, University of California, Davis, CA 95616, United States
b
USEPA/ORD/NERL, Research Triangle Park, NC 27711, United States
c
Blackland Research and Extension Center, Texas A&M University, Temple, TX 76502, United States
d
Department of Pesticide Regulation, California Environmental Protection Agency, Sacramento, CA 95812, USA
e
College of Water Resources and Civil Engineering, China Agricultural University, Beijing 100083, China
H I G H L I G H T S G R A P H I C A L A B S T R A C T
• Current pesticide modeling status and

concerns reviewed at basin level using
SWAT
• Algorithm improvement & advanced ap-
plication recently attract more attention.
• Algorithm improvement: fate simulation,
new pathways, transport & pollutant
control
• Model integration helps assess ecological
risks & conduct cost-effective analysis.
• Future improvements recommended for
addressing research gaps and concerns
a r t i c l e i n f o a b s t r a c t
Article history: The application of pesticides in agriculture is a widely-used way to alleviate pest stresses. However, it also intro-
Received 6 February 2019 duces various environmental concerns due to the offsite movement of pesticide residues towards receiving water
Received in revised form 7 March 2019 bodies. While the application of process-based modeling approaches can provide quantitative information on
Accepted 9 March 2019
pesticide exposure, there are nonetheless growing requirements for model development and improvement to
Available online 12 March 2019
better represent various hydrological and physico-chemical conditions at watershed scale, and for better
Editor: Jay Gan model integration to address environmental, ecological and economic concerns. The Soil and Water Assessment
Tool (SWAT) is an ecohydrological model used in over 3000 published studies, including about 50 for simulating
Keywords: pesticide fate and transport at the watershed scale. To better understand its strengths and limitations, we con-
Pesticide ducted a rigorous review of published studies that have used SWAT for pesticide modeling. This review provides
SWAT recommendations for improving the interior algorithms (fate simulation, pathway representation, transport/pol-
Fate and transport lution control, and other hydrological related improvement) to better represent natural conditions, and for fur-
Watershed modeling ther extension of pesticide exposure modeling using SWAT by linking it with other models or management
Review
tools to effectively address the various concerns of environmental researchers and local decision makers.
⁎ Corresponding authors.
E-mail addresses: [email protected] (C. Wang), [email protected] (M. Zhang).
https://doi.org/10.1016/j.scitotenv.2019.03.141
0048-9697/© 2019 Elsevier B.V. All rights reserved.
R. Wang et al. / Science of the Total Environment 669 (2019) 512–526 513
Going beyond past studies, we also recommend future improvement to fill research gaps in developing modular-
ized field level simulation, improved BMPs, new in-pond and in-stream modules, and the incorporation of soft
data. Our review pointed out a new insight of pesticide fate and transport modeling at watershed level, which
should be seen as steps leading to the direction for model development, as well as better addressing management
concerns of local stakeholders for model implementation.
© 2019 Elsevier B.V. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
2. Fundamental mechanisms of SWAT pesticide modeling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
2.1. Field scale modeling (pesticide fate) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
2.2. Basin level modeling (pesticide transport) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
2.3. In-stream process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516
3. Worldwide applications of SWAT and study types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516
4. Algorithm improvement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 518
4.1. Fate simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
4.1.1. Three phases distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
4.1.2. Multiple degradation rate at different layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
4.1.3. Kd computation during in-stream processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
4.2. Pathway representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
4.2.1. Direct losses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
4.2.2. Drifting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
4.3. Pollution control and transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
4.4. Other improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
4.4.1. Rice paddy modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
4.4.2. Flood diversion simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
5. Advanced applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
5.1. Model integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
5.1.1. Coupling SWAT with drift or atmospheric models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
5.1.2. Coupling SWAT with eco-toxicology models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
5.1.3. Coupling SWAT with bio-economic models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
5.2. Optimal BMP selection and placement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
6. Research gaps and future improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
6.1. Incorporating modularized field simulation within a basin scale framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
6.2. Improving best management practices (BMPs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
6.3. Including in-pond process simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
6.4. Improving in-stream transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
6.5. Prioritizing monitoring and incorporating soft data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
Appendix A. Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
1. Introduction and food may result in acute/chronic toxic effects on human health
(Pingali and Roger, 2012). Fantke et al. (2012) quantified the health im-
Pesticides have been used worldwide in agricultural systems since pacts from pesticide exposure in 24 European countries, and concluded
the 1950s to provide essential protection against pests or weeds, main- that 2.6 h of lifetime is lost per person due to pesticide exposure.
tain high crop yields and ensure food security (Zhang et al., 2015). Pes- The offsite movements of pesticide residues, which finally enter the
ticide application is one of the vital approaches to match the increasing surface or groundwater systems, also lead to the deterioration of water
food demand under the rapid growth of the human population and the quality, as well as the ensuing ecological risks to aquatic species living in
alteration of dietary habits towards high quality food (Popp et al., 2013), receiving water bodies (Bach et al., 2001; Brown et al., 2004; Wang et al.,
especially in developing countries. Food and Agriculture Organization of 2019b). In the United States, the ecological risk assessment for individ-
the United Nations (FAO) indicated that top five high pesticide use in- ual pesticides is evaluated by Aquatic life Benchmarks (https://www.
tensity countries for year 2016 are Maldives, Trinidad and Tobago, epa.gov/pesticide-science-and-assessing-pesticide-risks/aquatic-life-
Costa Rica, Bahamas and Barbados (http://www.fao.org/faostat/en/ benchmarks-and-ecological-risk), which are assessed by U.S. Environ-
#data/EP), which are all above 20 kg/ha yr. The developed country, for mental Protection Agency, Office of Pesticide Programs (USEPA, OPP).
example, United States, accounted for one-fifth of the global pesticide The benchmarks are based on toxicity values found in scientific studies
usage, which are 372 and 389 million kg (as active ingredient) in 2006 that are reviewed by USEPA, and can be used for evaluating acute and
and 2007 (Grube et al., 2011). chronic risks for fish, invertebrates, vascular and nonvascular plants. In
A continuous increase in global usage of pesticides since 1950 Europe, the predicted-no-effect concentration (PNEC) is based on the
(Rathore and Nollet, 2012) introduces unintended negative environ- EU Pesticide Directive 91/414/EEC. If pesticide concentrations are over
mental impacts. For example, direct contact with pesticides during ap- 1/100 of the acute LC50 of fish or Daphnia, then water quality requires
plication, or indirect contact/intake from pesticide residues in water further evaluation (Brock et al., 2006; Maund and Mackay, 1998).
514 R. Wang et al. / Science of the Total Environment 669 (2019) 512–526
Many monitoring studies across the world have reported the ex- representing atmosphere, plant, soil, aquifer, river and landscape
ceedance of ambient pesticide concentrations over water quality rec- mitigation. Compared to other watershed level models (MIKE SHE
ommendations. For example, Stone et al. (2014) examined national ADM, LEACHM-runoff, GR5-pesticides, SACADEAU, STREAM-pesticide,
pesticide monitoring data in 39 major U.S. rivers from 1992 to 2011 FLOWT, VESPP, I-Phy-Bvci and PHYLOU), the Soil and Water Assessment
and found that a large proportion (45%–69%) of assessed rivers in agri- Tool (SWAT) model (Arnold et al., 1993) covered the widest range of
cultural areas had one or more pesticides exceeding the aquatic-life processes and was deemed to be the most comprehensive model for
benchmark. In the U.S. Corn Belt, frequently detected pesticides in wa- representing the relevant hydrological and pesticide cycles. Quilbé
terways included atrazine, acetochlor, metolachlor, alachlor, and sima- et al. (2006) proposed a pragmatic approach for selecting models to ex-
zine (Sullivan et al., 2009). For rivers in the State of California, 2,4-D, plore pesticide fate and transport at the watershed scale based on
diuron, prometon, chlorpyrifos and diazinon were frequently identified 1) modeling characteristics; 2) output variables; 3) model applicability;
(Anderson et al., 2018; Zhang et al., 2012). Many other countries also re- 4) BMPs (Best Management Practices) capacity; and, 5) simplicity of
ported pesticide residues due to offsite movement. Silva et al. (2019) use. Quilbé et al. (2006) reviewed 36 widely used pesticide models
analyzed 76 pesticide residues in 317 European topsoils. They detected based on their pragmatic approach, and SWAT was deemed one of the
one or more residues in 83% of the soils and 58% of samples contained best for modeling pesticide contamination under a wide variety of
two or more mixtures of residues in eleven EU Member states. Glypho- conditions. Many model surveys have acknowledged SWAT's great
sate and its metabolites are the most frequently detected residues at usefulness in advancing scientific understandings in the fate and trans-
higher concentrations. Monitoring programs across Europe also demon- port of agrochemicals, as well as the potential environmental impacts
strated the presence of various pesticides in rivers, lakes and subsurface on agricultural and domestic activities (e.g., Wang et al., 2018) and
waters, which may pose potential ecological risks to aquatic species one indication of its unparalleled utility is that the number of peer-
(Brown and van Beinum, 2009; Loos et al., 2009; Proia et al., 2013). reviewed journal articles on pesticide modeling at the watershed level
Quantitative assessment of pesticide occurrence and exposure which used SWAT is almost as many as all other models combined
benefits researchers and local stakeholders for understanding pesticide (Table S1).
contamination status and potential ecological risks, in addition to eval- As eloquently stated by Borah and Bera (2003) “A clear understand-
uating the effectiveness of mitigation strategies in pollution abatement. ing of a model is important for its appropriate use. The compilation of
The development of monitoring networks is important for collecting the mathematical bases of these models would be useful to determine
pesticide contamination information at the watershed scale. However, the problems, situations, or conditions for which the models are most
data monitoring is always expensive and labor intensive, so more lim- suitable, the accuracies and uncertainties expected, their full potential
ited pesticide concentration data is available over a broad temporal uses and limitations, and directions for their enhancements or new de-
and spatial scope. Therefore, ecohydrologic modeling has become in- velopments.” Therefore, understanding SWAT's representation of phys-
creasingly popular in investigations of pesticide fate and transport pro- ical/chemical process is critical for end users, since it is the prerequisite
cesses, and has been greatly facilitated by the growth of computational for model improvement when interior watershed characteristics are not
power and geographic information system (GIS) technologies. accurately captured by the model. This may occur if key-processes are
Ecohydrologic models are usually built upon a series of mathematical either over-simplified or not represented at all. The importance of
representations of hydrological, physiochemical, and biological processes, algorithm improvement to better reflect interior processes related to
which are more flexible for generating pollution information across a pesticide fate and transport is gradually becoming better recognized
wide range of scenarios and scales (Chen et al., 2017b; Gitau et al., by researchers, but is known to vary case by case (Boithias et al.,
2016; Guo et al., 2018a; Li et al., 2019; Wang, 2010; Yuan et al., 2018), 2014a; Boulange et al., 2014; Ligaray et al., 2017).
and differentiating the effects of individual factors (e.g., management As an ecohydrologic model, there are also many requirements
practices, input or structure uncertainties) among all other possible fac- beyond the applicability of the model, but of great concerns for
tors (Chen et al., 2019; Guo et al., 2018b; Wang and Kalin, 2018; Wang other researchers and decision makers. When applying mitigation
et al., 2014; Yen et al., 2017; Yen et al., 2018). strategies to alleviate pesticide exposure, stakeholders often need
Pesticide runoff problems at the basin scale have attracted more at- to consider both economic and environmental benefits, requiring
tention in recent years, providing opportunities for model development an integration of both exposure and bio-economic models. Thus,
and application. However, it is difficult to completely capture the mech- studies which extend pesticide model usage via integration have
anisms involved in trace pollutant transport and the resulting environ- become increasingly popular (Bannwarth et al., 2016; Maringanti
mental consequences (Ouyang et al., 2017b). Therefore, all models are et al., 2011; Zhang et al., 2018), because they are able to disentangle
mathematical simplifications of natural processes. Many important re- various problems which are beyond the scope of simply modeling
quirements of all models are from the perspective of environmental pesticide fate and transport.
management (Hoogeweg et al., 2012), such as: 1) examine the necessity Building upon previous published studies, we collected information
for mitigation strategies (e.g., BMPs installation) based on current from watershed scale SWAT pesticide modeling case studies around the
pollutant estimations; 2) identify the most important areas and critical world. By systematically studying this large, diverse corpus we are
periods for pollution abatement; 3) evaluate the effect of different BMP uniquely able to address algorithm improvement and advanced applica-
strategies for reducing pollutants; and 4) determine the most appropri- tions which extend the scope of exposure modeling for pesticide fate
ate BMP or a certain setting (e.g., size) of a specific BMP for fields gener- and transport information at the watershed scale. We use SWAT as an
ating agrochemical contaminants. example to first illustrate the mathematical basis of pesticide modeling,
Many models have been developed for simulating the fate and trans- demonstrate recent pesticide modeling hotspot areas, and identify po-
port of pesticides at the watershed scale over the last few decades tential development opportunities.
(Fohrer et al., 2014). These models differ in modules which represent This review goes further in specifically emphasizing how previous
hydrological/water quality processes, spatial and temporal scales, studies overcame model limitations via algorithm improvement to better
management tools, and graphic user interfaces. Each model has its ad- represent key natural conditions; and how modelers have extended
vantages and weaknesses, making model selection quite critical for potential uses of pesticide exposure simulation by integrating with
end users. Some models may either lack specific mechanisms to reflect other models and management tools to better address the concerns of
important natural processes, or be too costly in terms of computational local decision makers. Important research gaps and suggestions for future
resources. development are provided, which are not only beneficial for SWAT
Payraudeau and Gregoire (2012) compared the ability of model modelers, but also very helpful for non-SWAT users and developers,
components to capture 42 critical hydrological and pesticide processes who are interested in pesticide fate and transport at watershed scale.
2. Fundamental mechanisms of SWAT pesticide modeling Similarly, the amount of pesticide attached to soil particles, and
ready to be transported by surface runoff during the soil erosion process
2.1. Field scale modeling (pesticide fate) is calculated by Eq. (5)
A portion of pesticides applied to agricultural fields is lost mainly 100∙K p ∙pst s;ly;t sed
pst 0sed ¼ 0:001∙ ∙ ∙εpst;sed ð5Þ
because of wind. The rest of the pesticides are partially intercepted SAT ly þ K p ∙ρb ∙depthly areahru
by crop foliage, and some portion also reaches the soil surface. The
amount of pesticide that reaches the soil surface is represented by where pst′sed is the amount of sorbed pesticide ready to be transported
Eq. (1). by surface runoff via soil erosion (kg pst/ha); sed is the sediment yield
1:99532− erfc½1:333 LAI−2
on a given day (tons); areahru is the hydrologic response unit (HRU)
pest surf ¼ pest apef 1− ð1Þ area (ha); and, εpst,sed is the pesticide enrichment ratio.
2:1
2.2. Basin level modeling (pesticide transport)
where pestsurf is the amount of pesticide applied to the soil surface
(kg/ha); pest is the actual amount of pesticide applied (kg/ha); apef After pesticide fate is simulated at the field scale, moveable pesticide
is the pesticide application efficiency; LAI is the leaf area index, (pstflow and pstsed) generated at the edge of the field could be
which affects ground cover; and erfc is the complementary error transported to the main channel. Equations used to simulate pesticide
function. Based on Eq. (1), as LAI increases, the amount of pesticide movement in the land phase of the hydrologic cycle were adopted
reaching the soil surface decreases due to the interception effect of from GLEAMS (Leonard et al., 1987). Soluble pesticide (pstflow) could
growing foliage. be transported by surface runoff, lateral flow or percolation in the
Pesticides in the soil environment are further distributed between amounts calculated by Eqs. (6)–(8). Algorithms used by SWAT to com-
interstitial water and soil particles. SWAT employs a chromatographic pute pesticide percolation simultaneously solve for loss of pesticide in
module (King and McCarty, 1968) to represent the phase distribution surface runoff and also lateral flow.
of pesticides, and the partition ratio is determined by the soil adsorption
pst flow
coefficient (Kp). Kp is affected by both soil organic matter and pesticide pst 0surf ¼ βpst ∙ ∙Q ð6Þ
wmobile surf
properties (Eq. (2)). Higher Kp values indicate that pesticides are more
prone to association with soil particles than soil water. 8
> pst flow
>
< βpst ∙ ∙Q for top 10 mm
pst 0lat ¼ wmobile lat;surf ð7Þ
C solidphase orgC > pst flow
Kp ¼ ¼ K oc ∙ ð2Þ >
: ∙Q for lower layers
C solution 100 wmobile lat;ly
pst flow
where Kp is the soil adsorption coefficient (m3/ton); Csolidphase (mg/kg) pst 0per;ly ¼ ∙w ð8Þ
and Csolution (mg/L) are the concentrations of pesticide within soil solid wmobile perc;ly
and liquid phases; Koc is the soil adsorption coefficient normalized for
where pst′surf, pst′lat and pst′per,ly are the pesticide amounts transported
the soil organic carbon content (m3/ton); and, orgC is the percentage
by surface runoff, lateral flow and percolation to the underlying layer in
of organic carbon in the soil.
each day (kg/ha); Qsurf is the surface runoff generated on a given day
Pesticides in the soil environment also experience degradation due
(mm); Qlat,suf and Qlat,ly are the amounts of water discharged from the
to various physico-chemical processes (volatilization, photolysis, hy-
soil layer by lateral flow at the top 10 mm layer and underlying layers,
drolysis, and biolysis), which are controlled by first-order kinetics.
respectively (mm); βpst is the pesticide percolation coefficient, and,
SWAT employs a lumped half-life (hlife_s) to represent the overall
wperc,ly is the amount of water percolating to the underlying layer
degradation of pesticide in the soil environment (both in soil water
(mm). Pesticide leaching out of the soil profile will pass through the
and particles), as shown in Eq. (3).
vadose zone and finally reach the shallow aquifer. Pesticides reaching
the shallow aquifer are not further tracked by SWAT.
0:693
pst s;ly;t ¼ pst s;ly;o ∙ exp ∙t ð3Þ For a large watershed with a time of concentration more than 1 day,
hlife s
only part of the generated flow/sediment is able to be delivered to
the main channel. This adjustment represents a lag between field-
where psts,ly,t and psts,ly,o are the amounts of pesticide in the soil layer at
generated pesticide and channel-delivered pesticide. SWAT employs
times t and 0 (kg/ha); and, hlife_s is the half-life of the pesticide in the
Eqs. (9)–(11) to describe this “lag” process.
soil (days). SWAT only tracks the fate of the parent pesticide. Although
the degradation process will generate metabolites from the parent
−surlag

pesticides, those metabolites are not simulated by SWAT. pst surf ¼ pst 0surf þ pst surstor;i−1 ∙ 1− exp ð9Þ
t conc
Pesticide that is distributed to the liquid phase is able to be

transported by soil water fluxes (surface runoff, lateral flow and −1
percolation). The amount of pesticide in the soil soluble phase that is pst lat ¼ pst 0lat þ pst latstor;i−1 ∙ 1− exp ð10Þ
TT lat
ready for transport is calculated by Eq. (4)

−surlag
" #! pst sed ¼ pst 0sed þ pst sedstor;i−1 ∙ 1− exp ð11Þ
−wmobile t conc
pst flow ¼ pst s;ly;t ∙ 1− exp ð4Þ
SAT ly þ K p ∙ρb ∙depthly
where pstsurf, and pstlat are the amounts of dissolved pesticide
discharged to the main channel on a given day (tons) via surface runoff
where pstflow is the amount of pesticide contained in the soil water and lateral flow, respectively; pstsed is the amount of sediment-sorbed
flux (kg pst/ha); wmobile is the amount of mobile water (mm); SATly is pesticide discharged to the main channel via runoff; pstsurstor,i−1,
the amount of water in the soil layer at saturation (mm); ρb is the pstlatstor,i−1, and pstsedstor,i−1 are the amounts of pesticide either stored
bulk density of the soil layer (Mg/m3); and, depthly is the depth of the or lagged from the previous day in surface runoff, lateral flow and sed-
soil layer (mm). iment; surlag is the surface runoff lag coefficient, TTlag is the lateral
flow travel time (days); and, tconc is the time of concentration for the pesticide removed via outflow (mg); pstdeg,wtr, pstvol,wtr and pstrsp,wtr
HRU (h). are the amounts of pesticide removed from the water layer via degrada-
tion, volatilization and settling, respectively (mg); pstdeg,sed is the
2.3. In-stream process amount of pesticide removed from the sediment layer via degradation
(mg); pstdif is the amount of pesticide transferred between liquid and
Upon delivery to the main channel, the pesticide may be transported solid phases via diffusion (mg); and, pstbur is the amount of pesticide re-
by channel flow from upstream to downstream receiving water bodies. moved via burial (mg). Most of the variables in Eqs. (14) and (15) are
Similar to the distribution between soil and water at the field scale, the functions of Fd and Fp from Eqs. (12) and (13). It was noted that sedi-
pesticide in the water layer in each reach segment will first partition ment and pesticide in-stream processes are computed separately in
into particulate and dissolved forms, and SWAT assumes a uniformly SWAT. Therefore, the model does not consider the interaction between
mixed water layer over the reach sediment layer, which is described those two processes.
by Eqs. (12) and (13).
3. Worldwide applications of SWAT and study types
1
Fd ¼ ð12Þ
1 þ K d ∙concsed The worldwide applications of the SWAT model for addressing pes-
ticide fate and transport indicate that it is a versatile tool for investigat-
K d ∙concsed ing agrochemical exposure and its ecological impacts in aquatic
F p ¼ 1− F d ¼ ð13Þ
1 þ K d ∙concsed systems. Based on the SWAT literature database (https://www.card.
iastate.edu/swat_articles/), around 50 peer-reviewed articles have
where Fd is fraction of total water layer pesticide in dissolved phase; Fp been published on pesticide modeling. Most published SWAT pesticide
is the total water layer pesticide in particulate phase; Kd is the pesticide studies were conducted in the United States, and they are mainly con-
partition coefficient in the reach segment (m3/g); and, concsed is the centrated in the Midwest and California. Other studies have been con-
concentration of suspended solids in the water (g/m3). ducted in Europe and East Asia (Fig. 1).
SWAT employs a simple mass balance equation to simulate the mass These studies can be roughly categorized into five groups based on
change in both reach water and channel bed sediment due to other the primary purpose: 1) sensitivity analysis; 2 exposure modeling for
physiochemical processes, as described in Eqs. (14) and (15). fate and transport; 3) mitigation strategy by using Best Management
Practices (BMPs); 4) Algorithm improvement; and, 5) advanced
Δpst rchwtr ¼ pst in − pst sol;o þ pst sorb;o −pst deg;wtr −pst vol;wtr −pst stl;wtr implementations. Fig. 1 displays the types of SWAT modeling studies
þ pst rsp;wtr pst diff by continent (Fig. 1a–c) and worldwide (Fig. 1d). Counts of the five
ð14Þ study types by year showed that studies on exposure modeling and
BMP mitigation continue to outnumber those in the other categories,
Δpst rchsed ¼ pst deg;sed þ pst stl;wtr −pst rsp;wtr −pst bur pst diff ð15Þ though studies on algorithm improvement and advanced applications
have increased in recent years (Fig. 2).
where Δpstrchwtr is the mass change in the water layer of the reach seg- Sensitivity analysis is often employed to determine which “sensi-
ments (mg); Δpstrchsed is mass change in the sediment layer of the reach tive” parameters most strongly drive modeled behaviors, which is
segments (mg); pstin is the pesticide added to the reach via inflow very useful for further model calibration and validation. Ecohydrological
(mg); pstsol,o and pstsorb,o are the amounts of dissolved and particulate models utilize many physically-based or empirical equations to reflect
Fig. 1. Global applications of SWAT for pesticide modeling.

Fig. 2. Categories of recent published SWAT pesticide modeling studies.
and simplify hydrological and water quality processes. Those equations as part of the molecules are sorbed to the suspended solids. The
are characterized by parameters, and in turn affect modeled behavior. streamflow, sediment and particulate pesticide should be calibrated se-
Some parameters used in empirical equations may not have physical quentially, because of the increased complexity in modeled processes
meanings, introducing difficulties to a priori estimations. Other param- and decreased accuracy of modeled variables.
eters have physical meanings, and could be measured or calculated Exposure prediction and mitigation strategies (BMPs) are popular
from lab experiments or field studies. However, some of those parame- topics for SWAT pesticide modeling studies, accounting around half of
ters have extreme heterogeneity, which limits the accuracy of their the published studies (Fig. 1). BMP effectiveness in controlling sediment
measurement, especially for an entire watershed. Understanding key and nutrients has been intensively studied by previous studies
parameters controlling water quality processes is very critical for (Liu et al., 2015; Liu et al., 2017; Rittenburg et al., 2015). Commonly
modeling pesticide fate and transport. The most commonly used applied structural BMPs include strip cropping, grassed waterways,
sensitivity analysis approach for SWAT pesticide modeling is the Latin vegetated filter strips (VFS) and sediment ponds, and each method
Hypercube—One-factor-At-a-Time (LH-OAT) method (Boithias et al., may be used for the management of either pesticides, sediment, and/or
2011; Holvoet et al., 2005; Van Griensven et al., 2006b). Although nutrient. Strip cropping and VFS has similar functions. VFS applied a
SWAT parameters exhibit different sensitivities for different pesticides, dense vegetation strip (e.g., bush) between crop fields and streams to re-
the most sensitive parameters are Koc, hlife_s, and Kd (Boithias et al., tain sediment and remove sediment attached pesticide. Strip cropping
2011; Holvoet et al., 2005; Kannan et al., 2006; Luo et al., 2008; increases surface roughness by applying bands of other crops within an
Quansah et al., 2008), which control pesticide sorption and degradation agricultural plot to retain pesticide transported by flow and sediment.
rates. A grassed waterway is a shallow vegetated channel between fields,
Model calibration/validation is the inverse way to determine the which prolongs flow retention time and sediment trapping efficiency.
optimum set of parameters by minimizing differences between mea- Sediment ponds or wetlands are impoundments receiving pesticide/
surements and simulation outputs. Most SWAT pesticide modeling sediment runoffs. It helps removing suspended solids and associated
studies conducted parameter calibration to ensure a satisfying model pesticide from water phase. Dissolved pesticides in ponds also
performance. Due to the scarcity of observed data, most studies con- experience more time for dissipation and degradation processes.
ducted pesticide calibration at the outlet of the study watershed Non-structural BMPs indicate any management approach, but not via
(Bannwarth et al., 2014; Ficklin et al., 2013a; Larose et al., 2007). If “physical structures”, capable of abating agrochemical loadings.
more data is available, calibration could be conducted at multiple sta- Examples would include using cover crops or residue management to
tions inside the target watershed (Quansah et al., 2008). Calibration cover bare soils and prevent erosion (Holvoet et al., 2007a), or alternat-
could be conducted at annual (Luo et al., 2008), monthly (Heathman ing pesticide application dates to avoid major precipitation events
et al., 2008) or daily scales (Fohrer et al., 2014), primarily depending (Farrand and Heidenreich, 2000).
on the sampling or measurement frequency. Most SWAT pesticide While exposure modeling and BMP mitigation are important appli-
studies conducted calibration/validation on dissolved pesticide, cations of SWAT, the more important revelations concern developments
while only a few studies have paid attention to both dissolved and in improving SWAT's algorithms and more advanced applications. Con-
particulate pesticide (Boithias et al., 2014b; Boithias et al., 2011). sequently, they will receive a more thorough and nuanced discussions
Model performance for particulate pesticide is also strongly dependent in Sections 4 and 5. Table 1 shows more detailed information on the
on the model's capacity to capture soil erosion and sediment transport, pesticides, study areas, and specific purposes of those studies.
Table 1
Summary of SWAT simulated pesticides, areas and purposes.
Case study Pesticide Type Area Size Purpose
King and Balogh (2001) 2,4-D Herbicide Dallas, TX; Columbia, MO; Minneapolis, MN 1.74 km2 Exposure modeling
Kim et al. (2004) Atrazine Herbicide Auglaize Watershed, OH 770 km2 Exposure modeling
Ramanarayanan et al. (2005) Isoxaflutole, RPA 202248 Herbicide Missouri, Nebraska, Iowa, USA 0.5–1433 km2 Exposure modeling
Holvoet et al. (2005) Atrazine Herbicide Nil catchment, Belgium 32 km2 Sensitivity analysis
Vazquez-Amabile et al. Atrazine Herbicide Cedar Creek Watershed, IN 2800 km2 Advanced application
(2006) (risk assessment)
Du et al. (2006) Atrazine Herbicide Walnut Creek watershed, IA 51.3 km2 Algorithm improvement
Kannan et al. (2006) Bentazone Herbicide Bedfordshire, UK 1.42 km2 Exposure modeling
Coupe (2007) Fluometuron Herbicide Bogue Phalia Basin, MS 1270 km2 Exposure modeling
Parker et al. (2007) Atrazine, metolachlor, trifluralin Herbicide White River Basin, IN 242 km2 Exposure modeling
Larose et al. (2007) Atrazine Herbicide Cedar Creek Watershed, IN 707 km2 Exposure modeling
Holvoet et al. (2007c) Atrazine Herbicide Nil catchment, Belgium 32 km2 Exposure modeling
Kannan et al. (2007) Linear alkylbenzene sulphonate Anionic surfactant Colworth, UK 1.42 km2 Exposure modeling
Holvoet et al. (2007a) Atrazine Herbicide Nil catchment, Belgium 32 km2 BMP
Heathman et al. (2008) Atrazine Herbicide Cedar Creek Watershed, IN 707 km2 Exposure modeling
Quansah et al. (2008) Atrazine Herbicide Cedar Creek Watershed, IN 707 km2 BMP
Prato and Woo (2008) Atrazine Herbicide Route J watershed, MO 20.87 km2 Advanced application
(bioeconomic)
Lerch et al. (2008) Atrazine Herbicide Salt River Basin, MO 6417 km2 BMP
Luo et al. (2008) Diazinon Insecticide San Joaquin Valley watershed, CA 14,983 km2 Exposure modeling
Zhang et al. (2008) Organophosphate, pyrethroid Insecticide San Joaquin River Watershed, CA 14,983 km2 BMP
Holvoet et al. (2008) Atrazine Herbicide Nil catchment, Belgium 32 km2 Algorithm improvement
Luo and Zhang (2009a) Chlorpyrifos Insecticide Orestimba Creek, CA 563 km2 Sensitivity analysis
Luo and Zhang (2009b) Organophosphate Insecticide San Joaquin River Watershed, CA 14,983 km2 Advanced application
(risk assessment)
Ficklin et al. (2010) Diazinon and chlorpyrifos Insecticide San Joaquin River watershed, CA 14,983 km2 Sensitivity analysis
Love et al. (2011) Atrazine, bromoxynil, glyphosate, Herbicide Michigan's lower Peninsular, MI 53,358 km2 BMP
2,4-D, metolachlor, pendimethalin,
sethoxydim, triflualin
Zhang and Zhang (2011) Organophosphates Insecticide Orestimba Creek, CA 563 km2 BMP
Maringanti et al. (2011) Atrazine Herbicide Wildcat Creek Watershed, IN 1956 km2 BMP optimization
Boithias et al. (2011) Metolachlor, trifluralin Herbicide Save river catchment, France 1110 km2 Exposure modeling
Ahmadi et al. (2013) Atrazine Herbicide Eagle Creek Watershed, IN 248.1 km2 BMP optimization
(Ficklin et al., 2013a) Chlorpyrifos Insecticide Sacramento river watershed, CA 23,300 km2 Algorithm improvement
Ahmadi et al. (2014) Atrazine Herbicide Eagle Creek Watershed, IN 248.1 km2 Advanced application
(GCM)
Boithias et al. (2014b) Metolachlor, aclonifen Herbicide Save river catchment, France 1110 km2 BMP
Fohrer et al. (2014) Flufenacet, metazachlor Herbicide Kielstau watershed, Germany 50 km2 Exposure modeling
Boithias et al. (2014a) Alachlor, atrazine, DEA, Herbicide Save river, France 1110 km2 Algorithm improvement
isoproturon, metolachlor,
tebuconazole, trifluralin
Bannwarth et al. (2014) Atrazine, endosulfan, Herbicide/insecticide Mae Sa catchment in northern Thailand 77 km2 Exposure modeling
chlorothalonil /fungicide
Boulange et al. (2014) Mefenacet Herbicide Sakura River basin, Japan 345 km2 Algorithm improvement
Ahmadi et al. (2015) Atrazine Herbicide Eagle Creek Watershed, IN 248.1 km2 BMP optimization
Baffaut et al. (2015) Atrazine Herbicide Goodwater Creek Experimental 73 km2 Exposure modeling
Watershed, MO
Chiu et al. (2016) Chlorpyrifos, diazinon Insecticide Sacramento river Watershed, CA 23,300 km2 Advanced application
(risk assessment)
2
Gali et al. (2016) Chlorpyrifos Insecticide Seven Mile Creek Watershed, MN 90.9 km BMP
Bannwarth et al. (2016) Chlorothalonil, cypermethrin Fungicide, insecticide Mae Sa catchment in northern Thailand 77 km2 Advanced application
Whitfield Aslund et al. Dimethoate Insecticide Delta region, CA 38,283 km2 Advanced application
(2017) (risk assessment)
Chiu et al. (2017) Chlorpyrifos, diazinon Insecticide Sacramento river Watershed, CA 23,300 km2 Advanced application
(GCM)
Chen et al. (2017a) Diuron Herbicide San Joaquin River watershed, CA 14, 983 km2 Exposure modeling
Ligaray et al. (2017) Malathion Insecticide Pagsanjan-Lumban Basin, Philippines 454.45 km2 Algorithm improvement
Ouyang et al. (2017a) Atrazine, oxadiazon, isoprothiolane Middle high latitude study watershed, 141.5 km2 Exposure modeling
Northeast China
Winchell et al. (2018b) Atrazine Herbicide Midwest US and Texas 20–170 km2 Exposure modeling
Winchell et al. (2018a) Malathion Insecticide Mill Creek and Three mile Creek, OR 164.6 and Algorithm improvement
53.7 km2
Zhang et al. (2018) Chlorpyrifos Insecticide Orestimba Creek, CA 563 km2 Advanced application
(Drift)
4. Algorithm improvement watershed level, so errors and uncertainties in flow and sediment sim-
ulation can negatively impact pesticide performance. It is very difficult
Improvements in SWAT's internal algorithms allow it to predict pes- to capture pesticide loadings with the same prediction accuracy as
ticide levels that are more realistic or accurate. Compared to other water flow or sediment. Cases where some key natural processes are over-
quality variables (e.g., sediment, nutrient), concentrations of pesticides simplified or not reflected in the modeling system at all may result in
are often at ppm or even ppb levels, which increases modeling difficul- poor model performance (Wang et al., 2019a). Under these circum-
ties to accurately capture pesticide dynamics. Furthermore, pesticide stances, calibration may not help much in improving model perfor-
residue transport is closely associated with flow and sediment at the mance. Alternatively, modifying the original SWAT algorithm for
pesticide modeling becomes crucial. Efforts have been made to improve pesticide specific, and can be obtained either from published work, or
the original code in 1) fate simulation; 2) pathway representation; estimated by solubility.
3) transport and pollution control; and 4) other improvement related
to hydrologic processes. Kd ¼ 3:085 10−8 ∙K ow ð18Þ
4.1. Fate simulation SWAT recommended Chapra's method to calculate the pesticide
partition coefficient in the river segment, which assumes that Kd is a
4.1.1. Three phases distribution constant value in the whole drainage network, without variation
The original SWAT model employs a traditional two-phase in time and space. However, many studies have indicated that the
partitioning approach to consider pesticide distribution between soil distribution of pesticides are not only decided by their hydrophobicity
and water. This method is deemed as inadequate if dissolved organic (represented by Kow), but are also related to many environmental fac-
carbon (DOC) is abundant in the water phase (Burkhard, 2000; tors, such as pH, total suspended matter, and dissolved and particulate
Ligaray et al., 2016). DOC mentioned here is organic carbon particles organic carbon (Karickhoff et al., 1979; Toul et al., 2003; Xu et al.,
with sizes between 0.22 and 0.45 μm (Ogura, 1970). Consequently, 1999). Based on 2-year water sampling for seven pesticides in Save
the original SWAT model was modified (Bergknut et al., 2010; Ligaray River, France, Boithias et al. (2014a) found that total suspended matter
et al., 2016; Ligaray et al., 2017) to consider three-phase distribution, (TSM) and particulate organic carbon (POC) are the most relevant envi-
treating dissolved carbon as an additional phase to absorb pesticide, ronmental factors affecting pesticide adsorption processes. Chapra's
so applied pesticides are partitioned into three phases in soils, which method was improved by including TSM when computing Kd in river
are freely dissolved pesticide in soil water, DOC associated part, and segments (Eq. (19)). Therefore, it is possible that Kd could vary tempo-
particle-bound pesticide (Eq. (16)). rally and spatially.

C tbs ¼ C pbs þ C fd þ C DoC ð16Þ 0:094
bs bs
Kd ¼ þ 0:021 ∙ 7:55 10−3 ∙K 0:36
ow ð19Þ
TSM−5
The consideration of additional pesticide distribution to DOC
includes a more detailed formalism (Ligaray et al., 2017), which The improved Kd computation method introduced by Boithias et al.
broadens the scope of the default SWAT pesticide fate module. (2014a) can be further extended to other catchments with sufficient
Ligaray et al. (2017) applied the default SWAT model and the data measurements, or to other non-charged organic pollutants
modified three-phase partitioning model in Pagsanjan-Lumban (e.g., pharmaceuticals).
Basin, investigating the fate and transport of malathion, which is a
commonly used pesticide for tomatoes in the Philippines. The 4.2. Pathway representation
modified model performed better than the default SWAT, especially
in estimating the peak concentration after pesticide application. The 4.2.1. Direct losses
consideration of DOC is also helpful for improving the prediction of Direct losses are a type of point losses which indicate the portion of
bioavailable agro-chemical concentrations for better ecosystem pesticide directly discharged to rivers or water bodies, by potential
exposure assessment in shallow water environments. Similar ideas spills, leakage or cleaning of the spray equipment (Gerecke et al.,
have been used in other models, such as the ChimERA fate model 2002). Unlike diffuse losses, adsorption or transport processes have
(Morselli et al., 2015) and ChimERA + soilPlus coupling system limited effects on point losses, which may cause an abrupt peak of
(Morselli et al., 2018a). pesticide concentration in the receiving water systems. A monitoring
study in Belgium indicated that direct losses could accounted for 60%–
4.1.2. Multiple degradation rate at different layers 90% of atrazine losses during 1998–2002 for a small river system
SWAT employs a lumped factor, pesticide soil half-life (Hlife_s), to (Holvoet et al., 2007b), so ignoring direct losses will yield unreasonable
reflect the degradation rate of pesticides in soils. Pesticide degradation modeling results.
depends on soil types, organic carbon amount, soil temperature and To solve this problem, a point losses algorithm is included in SWAT
moisture, and pH (Davie-Martin et al., 2015; Getzin, 1981; Goring and tested in Nil Catchment (Holvoet et al., 2008), where point losses
et al., 1975). Therefore, the half-life varies within the soil environment, are represented as the function of application efficiency (APEF), shown
and applying a single value of Hlife_s may not be sufficient to reflect the in Eq. (20)
degradation of the pesticide within soils. To address this concern, Du
et al. (2006) introduced “a second pesticide degradation half-life in direct losspoint ¼ aprate ð1−AP EF Þ areahru 1e8 ð20Þ
soil” (hlife_s2) which allows users to apply another half-life value for
layers below the surface soil. Similar to Hlife_s, Hilife_s2 still follows
the first-order kinetics: where, direct_losspoint is the amount of pesticide diverted directly to the
river (mg); aprate is the initial applied pesticide amount (kg/ha); APEF is

0:693 the application efficiency; and areahru is the HRU area on which the
pst s;ly2;t ¼ pst s;ly2;o ∙ exp ∙t ð17Þ
hlife s2 pesticide is applied (km2).
After the implementation of the point losses algorithm in areas
where psts,ly2,t and psts,ly2,o are the amounts of pesticide in the second where pesticides were applied to the field via spray equipment (not
soil layer at time t and 0 (kg/ha); and hlife_s2 is the half-life of the pes- airplane application), SWAT performance at the mouth of the river Nil
ticide in the second soil layer (days). on atrazine concentration prediction in middle spring were found to
be substantially improved (Holvoet et al., 2008).
4.1.3. Kd computation during in-stream processes
When pesticide is transported in the reach segment, it is also distrib- 4.2.2. Drifting
uted between particle and liquid phases. Similar to the partitioning in Pesticide drift is known as the transport of pesticide droplets from
soil phase at the field level, Kd is the key parameter which determines crop fields by wind, and the droplets are deposited to nearby areas
the amount of dissolved and particulate forms of pesticides in river (e.g., rivers), when pesticide is applied using spray nozzles. The original
courses. Chapra (2008) proposed a simple method to estimate Kd by SWAT code has been modified to consider droplet drift via a simple re-
linking Kd with Kow (octanol-water partition coefficient), where Kow is gression equation. The percentage of drift is described as a function of
travel distance, crop types and growth stages (Linders et al., 2003), as this case, many studies have found better results in pesticide modeling
shown in Eq. (21) when the hydrology module was improved.
A
Drift% ¼ zBþ1 Bþ1
2 −z1 ð21Þ
ðz2 −z1 Þ ðB þ 1Þ 4.4.1. Rice paddy modeling
The rice growing environment is different from other field crops due
where Drift% is the drift fraction (%); z1 is the distance from the edge of to its unique hydrologic conditions. Rice paddies are usually flooded
the treated HRU to the closest edge of the water body (m); z2 is the dis- during the rice growth period, and water in the paddies is renewed in
tance from the edge of the treated field to the farthest edge of the water order to maintain acceptable levels of oxygen, temperature and salinity.
body (m); and, A and B are the constant and exponential regression fac- Ponded water is finally released at the end of the growing season for
tors, respectively, which are extracted from an existing database harvesting purposes. The challenge for rice field modeling mainly lies
(Ganzelmeier et al., 1995). in the correct characterization of surface-water hydrology and water
Recently, Winchell et al. (2018a) has further improved the drift management associated with rice production. Flooding and draining
module by considering wind direction and speed when simulating mal- practices also alternate the soil anaerobic/aerobic conditions, which
athion in Mill Creek and Three Mile Creek, Oregon, United States. Wind affect the fate of pesticides in the water and sediment layer system
direction is used to decide if the specific application will cause drift to (Wang et al., 2019b).
the location of interest or not, while wind speed is used to refine the The default SWAT model lacks the capacity to correctly represent the
drift fraction. It was found that the inclusion of actual wind direction unique condition of paddy fields. Therefore, it is not appropriate to
and speed data in the drift code resulted in a more accurate estimate apply it to areas with a high proportion of rice fields for pesticide model-
of mean daily malathion concentration than the original, unmodified ing (Sakaguchi et al., 2014a; Sakaguchi et al., 2014b; Xie and Cui, 2011).
SWAT algorithm. Since SWAT's HRU is spatially scattered, which bring Efforts have been made to improve the water balance computation
problems in spray drift estimation when considering wind speed and module of SWAT (Sakaguchi et al., 2014a; Sakaguchi et al., 2014b;
wind directions, Winchell et al. (2018a) also developed a spatially ex- Tsuchiya et al., 2016). For example, Boulange et al. (2014) introduced
plicit approach to estimate receiving water body area and drift fraction PCPF-1@SWAT model for mefenacet simulation in Sakura River Water-
using actual HRU boundaries. shed, Japan, which is dominated by rice paddies. For this purpose, SWAT
had been modified in several ways: 1) rice pothole is modified from
4.3. Pollution control and transport “conical shape” to “shallow box” type, which is more realistic; 2) the de-
fault overflow calculation method (SCS-CN method) is modified, which
Vegetative filter strip (VFS) is a commonly used structure to trap becomes a function of precipitation, water levels in rice fields and bund
sediment or sediment attached agrochemicals (e.g., pesticides) from depths; 3) the auto-irrigation amount is computed based on precipita-
crop fields to receiving water bodies. Original SWAT uses an empirical tion, water level, and minimum ponding depth to trigger irrigation;
equation to compute the trap efficiency (Barlund et al., 2007), which 4) PCPF-1 model (Watanabe and Takagi, 2000) is employed to calculate
is only related to the width of filter strip. Deletic (2001) included the pesticide concentration in paddy water/sediment (top 10 mm) sys-
more physically based parameters to describe the sedimentation pro- tem for each SWAT rice HRU; 5) pesticide leaching through the 10 mm
cess in VFS, and updated the trap efficiency as Eq. (19). Holvoet et al. layer is passed to SWAT as input, following the chemical transport algo-
(2007a) adopted Deletic's trap efficiency equation in SWAT, and rithm of SWAT model in the soil environment; and, 6) pesticide runoff
asserted that the updated equation is able to reflect hydrologic varia- from rice paddies follows the SWAT surface movement and channel
tions for different runoff events. routine algorithms.
N 0:69
f ;s
T r;s ¼ ð22Þ
N 0:69
f ;s þ 4:95 4.4.2. Flood diversion simulation
Many watersheds contain trans-basin flow diversion (Guo et al.,
where Tr,s is the trapping efficiency, while Nf,s is the particle fall number, 2012; Li et al., 2013) through bypass or canals between the main chan-
which is a function of the width of buffer strip, Stokes settling velocity nel and tributaries. The flow conveyance structures change the natural
for particles, depth of the surface runoff, and average mean overland water routing paths, which increases the complexity of flow modeling
flow velocity. in the study region. For example, Ficklin (2010) reported six main
The SWAT default routine module has also been modified to accom- flood weirs in Sacramento River basin to control water depth in the
modate the computation of point losses during low flow periods. For main channel by diverting flow to wetlands or other receiving water
original SWAT code, the threshold of flow that triggers the in-stream bodies, and therefore alleviate the potential winter flood risk for the
water quality computation is set to 0.01 m3/s. Van Griensven et al. major rivers.
(2006a) argued that this threshold is too high, resulting in inaccuracies Directly applying the default SWAT in watersheds with trans-basin
and numerical errors. This is especially true for small catchments flow diversion cannot correctly capture the flow pattern, and therefore
which receive large amounts of pesticide point losses during the low introduces more uncertainties in the subsequent water quality estima-
flow condition (Gevaert et al., 2008). Therefore, modified SWAT code tions. Ficklin et al. (2013a) modified the default SWAT by adding a
reduces the water quality calculation threshold to 10−6 m3/s, and the flood conveyance algorithm to represent weir management. If the
updated routing module computes flow velocity and depth of flow via river stage is above certain threshold, then excess water is routed to
solving the Manning equations in an iterative process (Van Griensven the diversion destination, which could be wetland, other streams or
et al., 2006a). completely out of the watershed under investigation. Therefore,
any discharge above the specific flood discharge is removed from the
4.4. Other improvements reaches, which are not simulated via the SWAT channel transport mod-
ule. The implementation of the new algorithm was found to improve
All of the improvements discussed above are focused on the pesti- SWAT flow performance significantly, especially the peak flow in winter
cide module only. However, incorrect water simulation may also cause months, and therefore it decreased the uncertainty in chlorpyrifos esti-
problems in capturing pesticide loadings since water quality is closely mation due to the flood diversion effect. The flood conveyance modifi-
associated with hydrology. For sediment-adsorbed pesticide, the cor- cation could be used by other SWAT studies where inter or intra basin
rectness of loadings is affected by both suspended solid and flow. In water transfer has substantial impacts on local water management.
5. Advanced applications Yen et al. (2016a) that conservation practices in case studies cannot
be practically informative to decision makers if the given scenarios do
5.1. Model integration not align with reasonably estimated budgets. Some recent pesticide
modeling studies have begun to properly reflect the associated limita-
5.1.1. Coupling SWAT with drift or atmospheric models tions, and try to address the economic concerns via coupling SWAT
Although a modified SWAT model can add a pathway to represent with bio-economic models (Lescot et al., 2013; Vernier et al., 2017).
the drift loss during pesticide application, it is still unable to spatially Prato and Woo (2008) coupled SWAT model with a bioeconomic-
reflect the fate of drifted part explicitly due to over simplification. An al- weed management model, WeedSOFT (Neeser et al., 2004), to explore
ternate option is to couple SWAT with a drift or atmospheric model, the effects on water quality (atrazine exposure) and profitability of dif-
which provides SWAT with spatially explicit drift deposition after the ferent weed treatment methods for corn production. Based on the eval-
air dynamic process. uation of 46 pre-emergence, 67 post-emergence and 260 two-pass
Zhang et al. (2018) coupled AgDRIFT (Teske et al., 2002) with SWAT, treatment approaches, it turned out that post-emergence and two-
and developed a GIS modeling system to better represent drift simula- pass atrazine-free herbicide treatments not only protect water quality,
tion. This modeling system targets pesticide application areas and but also control weeds effectively and generate positive net returns.
spray events with drift potential via detailed usage information and Bannwarth et al. (2016) proposed an integrated approach to consider
wind speed and direction measurements. The amounts of pesticides both environmental benefits and economic costs in Mae Sa Watershed,
moving offsite for each drift event are estimated by the AgDRIFT Thailand. SWAT was initiated to decide pesticide application thresholds
model, considering the application method and downwind distance. to achieve the non-hazardous concentrations of chlorothalonil and
Finally, pesticide depositions to the receiving waters for each drift cypermethrin in Mae Sa River. The change in farm household incomes
event are treated as point sources, which are routed to the watershed in response to pesticide application reduction was quantified by an eco-
outlet, following the SWAT channel routing and in-stream process algo- nomic model, MPMAS (Mathematical Programming-based Multi-Agent
rithms. Zhang et al. (2018) applied the integrated modeling system in systems). The results indicated that current pesticide application rates
Orestimba Creek Watershed, California, to investigate the drift and would have to be reduced by 80% to 99% to achieve the predefined
runoff-related inputs of pesticides to receiving water. Runoff contrib- eco-toxicological thresholds. Additionally, it also suggested a decreased
uted 81% of total chlorpyrifos loading in the study area, while spray farmer household income (17.3% to 38.3%) in conjunction with the rate
drift accounted for another 19%. The AgDRIFT coupling improved the reduction goal.
overall performance in pesticide simulation at the outlet of the entire
Orestimba Creek Watershed. 5.2. Optimal BMP selection and placement
5.1.2. Coupling SWAT with eco-toxicology models The most effective way to eliminate pesticide pollution in receiving
Many ecological risk assessment studies require modeling or mea- water bodies is to eliminate the associated application source(s).
sured pesticide data to represent exposure levels, and then incorporate However, if the decreased pesticide usage results in crop yield reduc-
the organized data to individual or community responses for ecological tion, and the loss of farm household incomes, then it will not be
risk characterization (Solomon et al., 2000; Wang et al., 2018). Under accepted as an economically feasible approach. Alternatively, other mit-
this circumstance, simulated exposures are provided by SWAT as igation strategies, structural or non-structural BMPs for example, should
evidence to evaluate whether polluted water poses toxicities to any eco- be considered to alleviate the transport from applied fields to receiving
logical entity under investigation. For example, Chiu et al. (2016) water bodies.
employed SWAT model to simulate the spatiotemporal dynamics of Many SWAT pesticide modeling studies have indicated the useful-
pesticide mixtures in streams of the Sacramento River watershed. ness of BMPs in controlling pesticide loadings (Lerch et al., 2008;
High temporal resolution exposures were then used to compute aver- Quansah et al., 2008; Zhang et al., 2008), but only a few of those really
age and maximum pesticide toxic unit (TU), and linked with macroin- considered the balance between environmental effectiveness and
vertebrate communities' responses via a regression model. In another socio-economic efficiency when applying BMPs in pesticide application
example, Whitfield Aslund et al. (2017) used SWAT to model concentra- regions. A major reason is that with a great deal of heterogeneity in to-
tions of dimethoate in 150 reaches which overlapped with Pacific pography, soil, and hydrological connectivity, the effectiveness in reduc-
salmon critical habitat in the California central valley. Reach-specific ing agrochemical loads by certain BMP are not always the same, and the
risk curves were then generated based on cumulative probability of ex- cost for BMP implementation and maintenance can vary. This means
ceedance for exposure (maximum 96 h average estimated pesticide that the suitable BMP for one field may not always be suitable for
concentration) and magnitude of effect (concentration-response rela- other fields. For a watershed with multiple fields or farms and various
tionship for inhibition of brain AChE activity in salmon juveniles). BMP options, the economic efficiency and environmental effectiveness
Based on the risk curve (joint probability curve), the probability of a for the entire basin may require a variety of different BMP-field combi-
value being exceeded for any given exposure level, and the percentage nations. Therefore, it becomes a daunting task to identify the appropri-
of affected species under such exposure level can be determined. ate combination of BMPs to strike the balance between environmental
and economic benefits, which requires an optimization approach to se-
5.1.3. Coupling SWAT with bio-economic models lect the best placement of suitable BMPs.
Most SWAT modeling is used to predict pesticide concentrations Based on the pollution reduction and cost information from BMP tool
for decisions on reducing pesticide inputs or applying conservation database (Maringanti et al., 2009), Maringanti et al. (2011) combined
practices to limit pollution to receiving systems. In general, pesticide different Nonpoint Source pollutants into a single objective function,
exposure in the receiving water bodies under scrutiny is provided and optimized the selection and placement of BMPs along with the net
through model simulations under current applications or management costs. Optimization was performed in Wildcat Creek, Indiana, a corn-
practices. Suggestions of conservation practices made by those studies soybean dominated watershed, and the suitable location for BMPs
(Lerch et al., 2008; Love et al., 2011; Quansah et al., 2008) to local stake- implementations with minimum cost were geographically mapped. Sim-
holders are either to reduce pesticide usage, and/or to implement BMPs ilarly, Ahmadi et al. (2013) improved the computation efficiency of the
to eliminate the potential risks due to high residue exposures. The major Genetic Algorithm (GA), and developed “a mixed discrete-continuous
drawback of those studies is that the environmental “effect” is the only variable multi-objective genetic algorithm” to identify the optimal loca-
consideration, but socio-economic “cost” of adopting mitigation strate- tions and conservation types to control agrochemical loads at the water-
gies to eliminate contaminations are not considered. It was stated by shed level. Their optimization algorithm aimed to minimize the atrazine
load and the cost of implementation of each BMP simultaneously in Eagle et al., 2008; Bracmort et al., 2006). However, in reality, the size of the
Creek Watershed, Indiana. The proposed optimization approach pro- ditch (length, width, and depth), cover grass types, and coverage
vided more realistic watershed management options for local stake- percentages and densities are different from field to field, resulting in
holders to minimize pesticide pollution in an affordable way. discrepancies in removal efficiency (Moore et al., 2011). Therefore, in-
corporating detailed information of vegetated agricultural drainage
6. Research gaps and future improvements ditches in SWAT BMP modeling is another suggestion. As another
example, empirical equations are used to reflect the efficiency of two-
Based on our review work, there are still many deficits in current stage ditches in reducing pollutant loadings at headwater subbasins
studies of pesticide fate and transport modeling. Those deficits and lim- (Christopher et al., 2017). Therefore, further improvements in the
itations are not specific for SWAT model only, but are also common representation of BMPs in SWAT will provide better results to support
problems for other pesticide models at watershed scale. effective management decisions.
6.1. Incorporating modularized field simulation within a basin scale 6.3. Including in-pond process simulation
framework
Many studies have indicated the effectiveness of sediment ponds
Although the SWAT model itself is able to conduct both field level and wetland in abating suspended solids and pollutants associated
fate and basin level transport simulation for pesticides, end users still with particles from both urban and agricultural lands (Schulz and
should have the freedom to employ a more specialized module to com- Peall, 2001). A pond or wetland could be placed at either field or subba-
pute field level pesticide loadings (Boulange et al., 2014; Chinen et al., sin level for pesticide mitigation before runoff is routed into the main
2016; Luo et al., 2013; Tsuchiya et al., 2018; Tu et al., 2018). Default channel. For example, the European Forum for the Coordination of
SWAT may not be able to accurately capture the pesticide fate under Pesticide Fate Models and their Use (FOCUS) workgroup proposed a hy-
very special field conditions (e.g., rice paddy module or residential mod- pothetical pond (FOCUS pond) for pesticide risk assessment in aquatic
ule). However, the hydrology simulation in SWAT is more complete ecosystems (FOCUS, 2001). However, the default SWAT in-pond pro-
when compared to many field-scale models, especially at the basin cesses are only for water, sediment and nutrients, but do not include
level. Therefore, modelers should be able to develop a new, customized pesticides. Thus, similar SWAT algorithms for pesticide simulation in
framework, which couples their own preferred field level model with lakes and reservoirs could be adapted for modeling pesticide abatement
the SWAT routing module. For example, PRZM (Pesticide Root-Zone via in-pond process (Luo and Zhang, 2009a).
Model) is a commonly used field level model for estimating pesticide
offsite movement from treated surfaces (Suárez, 2005), which has un- 6.4. Improving in-stream transport
dergone an extensive validation effort against numerous field-scale run-
off and leaching studies conducted for pesticides in the United States When transported in the main channel, both the particulate and dis-
(Jones and Russell, 2001). The model has been integrated into several solved forms of pesticides are subject to degradation. Current SWAT
watershed assessments in the U.S., including the Sacramento River wa- model employs first order kinetics with a fixed rate to represent the
tershed (Snyder and Williams, 2004). With the same set of limited input degradation or removal of pesticides in water. However, the rate itself
data, PRZM has comparable modeling performance to other more com- is a function of water temperature (Nicholls et al., 1984). A modification
plicated models (Zhang and Goh, 2015), offering modeling options for could be incorporated by introducing water temperature as a factor to
pesticide regulation and management. Incorporating pesticide-focused realistically adjust the degradation rate. For example, the temperature
PRZM results from the field level into basin level SWAT is possible, but factor used by USEPA is represented as 2(t − t0) / 10, where t0 is the ref-
not convenient without a customized programmable interface. A recent erence temperature (usually 20 or 25 °C), and t is the real water temper-
study, Gali et al. (2016) designed Pesticide Runoff Management Tool ature in the main channel. This factor indicates that a 10 °C increase in
(PRMT) which automatically links PRZM-predicted pesticide fluxes at water temperature doubles the degradation rate. Therefore, for regions
edge-of-field to SWAT for further fate and transport simulations at the with water temperatures much higher or lower than the reference tem-
basin level. Similar modeling efforts have been conducted for a revised perature, pesticide degradation rates during the in-stream transport
and object based SWAT+ model (Bieger et al., 2017), as well as non- would be either under or over estimated. For example, pesticide man-
SWAT basin level ecohydrological models (Morselli et al., 2018a; agement regulation is affected by winter rain seasons in Mediterranean
Morselli et al., 2018b; Moser et al., 2018). climate regions (e.g., California, USA). A reduced dormant spray and res-
idential insecticide usage are recommended for this season due to the
6.2. Improving best management practices (BMPs) slowed degradation in winter waters (Brown et al., 2008; Budd et al.,
2015). Another modification is to link the in-stream processes of sedi-
Default SWAT model has two main limitations for evaluating the ef- ment and particulate pesticide together. The mass balance of particulate
fects of BMP implementation on pesticide management. First, its Vege- pesticide in the channel for current SWAT is completely decoupled from
tation Filter Strips (VFS) employs empirical regression equations to the model's in-stream sedimentation process. Sediment deposited and
determine the reduction of flow or pollutant during each runoff event, re-entrained in the reach segment during channel routing should be
while the pesticide residues and fate processes in a VFS are not fully considered for future model development when computing pesticide
evaluated. The effectiveness of VFS can be highly variable, being affected loadings during in-stream transport.
by vegetation types, soil, land use and properties of pollutants
(Reichenberger et al., 2007). Continuous simulation of pesticide fate is 6.5. Prioritizing monitoring and incorporating soft data
necessary for more realistic representation of the mitigation effects on
applying VFS (Carluer et al., 2017; Cibin et al., 2018; Muñoz-Carpena Compared to spatial and temporal data for flow and nutrients, ob-
et al., 2018), especially when VFS installed sites experience a seasonal served data on pesticide concentrations is more limited (Petersen et al.,
shallow water table (Carpena et al., 2017; Lauvernet and Munoz- 2012; Stehle et al., 2013), mainly due to the sampling and detection ex-
Carpena, 2018), which reduces buffer zone efficiency. Second, several penses and a large number of potential pesticide types of interests.
BMPs actually installed at the field level are represented at the subbasin Therefore, modeling work could be vital for evaluating the exposure
level in SWAT, which may not be suitable for BMPs evaluation. For ex- risks for different pesticides as a preliminary step, and then designing ei-
ample, vegetated ditches are simulated in SWAT by considering the ther an efficient monitoring plan for the pesticides of greatest concern
roughness, cover, and erodibility to all channels in a subbasin (Arabi (Morselli et al., 2018a), and/or the most appropriate sampling period
and approaches. For example, grab sampling, which is often performed optimal BMP selection and placement. Algorithm improvements and
in non-flood periods, is not sufficient to provide exposure information advanced applications summarized in this paper broaden the view of
under large events, especially the peak information of chemicals. Passive current pesticide exposure modeling, and could serve as good refer-
sampling devices, however, are able to provide a time-weighted average ences for both SWAT and non-SWAT modelers. Research gaps and
during an exposure period, which has better temporal resolution than future development needs, including modularized field simulation
grab sampling. When funding and labor availability permit, a high- within basin framework, more physical based BMPs, in-pond or in-
resolution automatic sampling approach is always recommended to stream modules, prioritizing monitoring and soft data incorporation
cover all possible situations (O'Brien et al., 2016). should be seen as steps leading the direction for better accounting for
In return, monitoring also provides scientific proof for evaluating real natural processes for model development, as well as in better ad-
model performance in capturing pesticide loadings in a time series dressing the management concerns of local stakeholders for model
manner. Pesticide calibration should follow the flow-sediment- implementation.
pesticide sequence. Model performance in dissolved pesticide is mainly
related to flow. However, model performance in particulate pesticide Acknowledgements
also depends on model capacity for representing sediment fate and
transport. The interdependencies among constituents are due to shared This research project was funded by the California Department of
transport processes. When monitoring data is sufficient, it is also Pesticide Regulation, #18-C0018, and U.S. Environmental Protection
recommended to calibrate model at multiple stations, not only at the Agency, #58-6060-6-019.
watershed outlet.
Pesticide monitoring data usually refer to temporal observation re- Appendix A. Supplementary data
cords, obtained at the watershed or subbasin outlet (point observations,
or “hard data”). Information provided by hard data is always aggregated Supplementary data to this article can be found online at https://doi.
(Yen et al., 2016b), without adequate delineation of the spatial distribu- org/10.1016/j.scitotenv.2019.03.141.
tion of individual sources and sinks for various constituents which con-
tribute to pesticide fate and transport. On the other hand, “soft data” References
(Seibert and McDonnell, 2002) refers to anything not covered by hard
Ahmadi, M., Arabi, M., Hoag, D.L., Engel, B.A., 2013. A mixed discrete-continuous variable
data, which reflects the interior behavior of a watershed. Soft data
multiobjective genetic algorithm for targeted implementation of nonpoint source
includes information from soil profile, field plots, watershed and other pollution control practices. Water Resour. Res. 49, 8344–8356.
research efforts among other processes, for example, soil water content Ahmadi, M., Records, R., Arabi, M., 2014. Impact of climate change on diffuse pollutant
fluxes at the watershed scale. Hydrol. Process. 28, 1962–1972.
of different layers (Wang et al., 2016a), pesticide application informa-
Ahmadi, M., Arabi, M., Fontane, D.G., Engel, B.A., 2015. Application of multicriteria deci-
tion (rate, date, application methods) at the plot scale (Orlando and sion analysis with a priori knowledge to identify optimal nonpoint source pollution
Kuivila, 2004), regional crop leaf development (Wang et al., 2016b), or control plans. J. Water Resour. Plan. Manag. 141, 04014054.
physical/spatial similarity between two watersheds (Wang and Kalin, Anderson, B.S., Phillips, B.M., Voorhees, J.P., Deng, X., Geraci, J., Worcester, K., Tjeerdema,
R.S., 2018. Changing patterns in water toxicity associated with current use pesticides
2011). Minimizing the differences between simulation and observed in three California agriculture regions. Integr. Environ. Assess. Manag. 14, 270–281.
(or hard) data via parameter calibration is a common way to improve Arabi, M., Frankenberger, J.R., Engel, B.A., Arnold, J.G., 2008. Representation of agricultural
the accuracy of model performance (Ficklin et al., 2013b). However, it conservation practices with SWAT. Hydrol. Process. 22, 3042–3055.
Arnold, J.G., Allen, P.M., Bernhardt, G., 1993. A comprehensive surface-groundwater flow
is also possible to force a model to capture hard data successfully via model. J. Hydrol. 142, 47–69.
over-parameterization, while failing to follow the actual interior water- Bach, M., Huber, A., Frede, H.-G., 2001. Input pathways and river load of pesticides in
shed behaviors. Under this circumstance, available soft data could serve Germany—a national scale modeling assessment. Water Sci. Technol. 43, 261–268.
Baffaut, C., John Sadler, E., Ghidey, F., Anderson, S.H., 2015. Long-term agroecosystem re-
as constraining criteria during the calibration process, and avoid a search in the central Mississippi River basin: SWAT simulation of flow and water
“precisely wrong” model due to equifinality. Currently published pesti- quality in the Goodwater Creek experimental watershed. J. Environ. Qual. 44, 84–96.
cide modeling studies rarely employ soft data as calibration constraints, Bannwarth, M.A., Sangchan, W., Hugenschmidt, C., Lamers, M., Ingwersen, J., Ziegler, A.D.,
Streck, T., 2014. Pesticide transport simulation in a tropical catchment by SWAT. En-
mainly because of inadequate measurements at both outlet or field
viron. Pollut. 191, 70–79.
levels. However, incorporating soft data (e.g., flow/sediment/pesticide Bannwarth, M.A., Grovermann, C., Schreinemachers, P., Ingwersen, J., Lamers, M., Berger,
loadings at the edge of field; pesticide concentration in field soils; soil T., Streck, T., 2016. Non-hazardous pesticide concentrations in surface waters: an in-
tegrated approach simulating application thresholds and resulting farm income ef-
moisture; crop coverage) in the calibration process should be seen as
fects. J. Environ. Manag. 165, 298–312.
a crucial step for improving future pesticide fate and transport modeling Barlund, I., Kirkkala, T., Malve, O., Kamari, J., 2007. Assessing SWAT model performance in
studies. the evaluation of management actions for the implementation of the Water Frame-
work Directive in a Finnish catchment. Environ. Model Softw. 22, 719–724.
Bergknut, M., Meijer, S., Halsall, C., Ågren, A., Laudon, H., Köhler, S., Jones, K.C., Tysklind,
7. Conclusions M., Wiberg, K., 2010. Modelling the fate of hydrophobic organic contaminants in a bo-
real forest catchment: a cross disciplinary approach to assessing diffuse pollution to
This review has pointed out the current status and future research surface waters. Environ. Pollut. 158, 2964–2969.
Bieger, K., Arnold, J.G., Rathjens, H., White, M.J., Bosch, D.D., Allen, P.M., Volk, M.,
needs for pesticide fate and transport simulation at watershed level. Srinivasan, R., 2017. Introduction to SWAT+, a completely restructured version of
The SWAT model is used as an example to first illustrate the fundamen- the Soil and Water Assessment Tool. J. Am. Water Res. Assoc. 53, 115–130.
tal mechanisms of pesticide exposure simulation. Based on current Boithias, L., Sauvage, S., Taghavi, L., Merlina, G., Probst, J.-L., Sánchez Pérez, J.M., 2011. Oc-
currence of metolachlor and trifluralin losses in the Save river agricultural catchment
published SWAT pesticide studies, we found that interior algorithm im- during floods. J. Hazard. Mater. 196, 210–219.
provement and external model integration have become important for Boithias, L., Sauvage, S., Merlina, G., Jean, S., Probst, J.-L., Sánchez Pérez, J.M., 2014a. New
watershed level pesticide modeling studies. Many studies involve insight into pesticide partition coefficient Kd for modelling pesticide fluvial transport:
application to an agricultural catchment in south-western France. Chemosphere 99,
algorithm improvement according to various watershed behaviors, 134–142.
management strategies, and pesticide properties. When research or Boithias, L., Sauvage, S., Srinivasan, R., Leccia, O., Sánchez-Pérez, J.-M., 2014b. Application
management interests are beyond the scope of exposure in receiving date as a controlling factor of pesticide transfers to surface water during runoff
events. Catena 119, 97–103.
water bodies, extended model usage, or model integration is highly en-
Borah, D., Bera, M., 2003. Watershed-scale hydrologic and nonpoint-source pollution
couraged, and has been actually widely utilized in recent years. SWAT models: review of mathematical bases. Trans. ASAE 46, 1553–1566.
can either take results from other models for better exposure estimation Boulange, J., Watanabe, H., Inao, K., Iwafune, T., Zhang, M., Luo, Y., Arnold, J., 2014. Devel-
(e.g., AgDRIFT-SWAT coupling), or provide agrochemical loadings or opment and validation of a basin scale model PCPF-1@SWAT for simulating fate and
transport of rice pesticides. J. Hydrol. 517, 146–156.
concentrations as a “starting point” for ecological risk assessment Bracmort, K.S., Arabi, M., Frankenberger, J., Engel, B.A., Arnold, J.G., 2006. Modeling long-
(e.g., SWAT-TU), cost-effectiveness analysis (e.g., SWAT-MPMAS), or term water quality impact of structural BMPs. Trans. ASABE 49, 367–374.
Brock, T.C., Arts, G.H., Maltby, L., Van den Brink, P.J., 2006. Aquatic risks of pesticides, eco- Switzerland: pesticide load through waste water treatment plants-current situation
logical protection goals, and common aims in European Union legislation. Integr. En- and reduction potential. Chemosphere 48, 307–315.
viron. Assess. Manag. 2, e20–e46. Getzin, L., 1981. Degradation of chlorpyrifos in soil: influence of autoclaving, soil mois-
Brown, C.D., van Beinum, W., 2009. Pesticide transport via sub-surface drains in Europe. ture, and temperature. J. Econ. Entomol. 74, 158–162.
Environ. Pollut. 157, 3314–3324. Gevaert, V., Van Griensven, A.N.N., Holvoet, K., Seuntjens, P., Vanrolleghem, P.A., 2008.
Brown, C.D., Dubus, I.G., Fogg, P., Spirlet, M., Gustin, C., 2004. Exposure to sulfosulfuron in SWAT developments and recommendations for modelling agricultural pesticide mit-
agricultural drainage ditches: field monitoring and scenario-based modelling. Pest igation measures in river basins. Hydrol. Sci. J. 53, 1075–1089.
Manag. Sci. 60, 765–776. Gitau, M.W., Chen, J.Q., Ma, Z., 2016. Water quality indices as tools for decision making
Brown, D.L., Giles, D.K., Oliver, M.N., Klassen, P., 2008. Targeted spray technology to re- and management. Water Resour. Manag. 30, 2591–2610.
duce pesticide in runoff from dormant orchards. Crop Prot. 27, 545–552. Goring, C., Laskowski, D., Hamaker, J., Meikle, R., 1975. Principles of Pesticide Degradation
Budd, R., Ensminger, M., Wang, D., Goh, K.S., 2015. Monitoring fipronil and degradates in in Soil. Environmental Dynamics of Pesticides. Springer, pp. 135–172.
California surface waters, 2008–2013. J. Environ. Qual. 44, 1233–1240. Grube, A., Donaldson, D., Kiely, T., Wu, L., 2011. Pesticides Industry Sales and Usage. US
Burkhard, L.P., 2000. Estimating dissolved organic carbon partition coefficients for non- EPA, Washington, DC.
ionic organic chemicals. Environ. Sci. Technol. 34, 4663–4668. Guo, X., Hu, T., Zhang, T., Lv, Y., 2012. Bilevel model for multi-reservoir operating policy in
Carluer, N., Lauvernet, C., Noll, D., Munoz-Carpena, R., 2017. Defining context-specific sce- inter-basin water transfer-supply project. J. Hydrol. 424–425, 252–263.
narios to design vegetated buffer zones that limit pesticide transfer via surface runoff. Guo, T., Gitau, M., Merwade, V., Arnold, J., Srinivasan, R., Hirschi, M., Engel, B., 2018a. Com-
Sci. Total Environ. 575, 701–712. parison of performance of tile drainage routines in SWAT 2009 and 2012 in an exten-
Carpena, R.M., Lauvernet, C., Carluer, N., 2017. Shallow water table effects on water, sed- sively tile-drained watershed in the Midwest. Hydrol. Earth Syst. Sc. 22, 89–110.
iment, and pesticide transport in vegetative filter strips—part 1: nonuniform infiltra- Guo, T., Engel, B.A., Shao, G., Arnold, J.G., Srinivasan, R., Kiniry, J.R., 2018b. Development
tion and soil water redistribution. Hydrol. Earth Syst. Sci. 22, 53–70. and improvement of the simulation of woody bioenergy crops in the Soil and
Chapra, S.C., 2008. Surface Water-Quality Modeling. Waveland Press. Water Assessment Tool (SWAT). Environ. Modell. Softw. https://doi.org/10.1016/j.
Chen, H., Luo, Y., Potter, C., Moran, P.J., Grieneisen, M.L., Zhang, M., 2017a. Modeling pes- envsoft.2018.08.030 (In Press).
ticide diuron loading from the San Joaquin watershed into the Sacramento-San Heathman, G., Flanagan, D., Larose, M., Zuercher, B., 2008. Application of the soil and
Joaquin Delta using SWAT. Water Res. 121, 374–385. water assessment tool and annualized agricultural non-point source models in the
Chen, J., Liu, Y., Gitau, M.W., Engel, B.A., Flanagan, D.C., Harbor, J.M., 2019. Evaluation of St. Joseph River watershed. J. Soil Water Conserv. 63, 552–568.
the effectiveness of green infrastructure on hydrology and water quality in a com- Holvoet, K., van Griensven, A., Seuntjens, P., Vanrolleghem, P.A., 2005. Sensitivity analysis
bined sewer overflow community. Sci. Total Environ. 665, 69–79. for hydrology and pesticide supply towards the river in SWAT. Phys. Chem. Earth A/
Chen, J.Q., Theller, L., Gitau, M.W., Engel, B.A., Harbor, J.M., 2017b. Urbanization impacts on B/C 30, 518–526.
surface runoff of states the contiguous United States. J. Environ. Manag. 187, 470–481. Holvoet, K., Gevaert, V., van Griensven, A., Seuntjens, P., Vanrolleghem, P.A., 2007a.
Chinen, K., Lau, S.L., Nonezyan, M., McElroy, E., Wolfe, B., Suffet, I.H., Stenstrom, M.K., Modelling the effectiveness of agricultural measures to reduce the amount of pesti-
2016. Predicting runoff induced mass loads in urban watersheds: linking land use cides entering surface waters. Water Resour. Manag. 21, 2027–2035.
and pyrethroid contamination. Water Res. 102, 607–618. Holvoet, K., Seuntjens, P., Mannaerts, R., De Schepper, V., Vanrolleghem, P., 2007b. The dy-
Chiu, M.-C., Hunt, L., Resh, V.H., 2016. Response of macroinvertebrate communities to namic water–sediment system: results from an intensive pesticide monitoring cam-
temporal dynamics of pesticide mixtures: a case study from the Sacramento River paign. Water Sci. Technol. 55, 177–182.
watershed, California. Environ. Pollut. 219, 89–98. Holvoet, K., van Griensven, A., Gevaert, V., Seuntjens, P., Vanrolleghem, P.A., 2008. Modi-
Chiu, M.-C., Hunt, L., Resh, V.H., 2017. Climate-change influences on the response of mac- fications to the SWAT code for modelling direct pesticide losses. Environ. Model
roinvertebrate communities to pesticide contamination in the Sacramento River, Cal- Softw. 23, 72–81.
ifornia watershed. Sci. Total Environ. 581–582, 741–749. Holvoet, K.M.A., Seuntjens, P., Vanrolleghem, P.A., 2007c. Monitoring and modeling pesti-
Christopher, S.F., Tank, J.L., Mahl, U.H., Yen, H., Arnold, J.G., Trentman, M.T., Sowa, S.P., cide fate in surface waters at the catchment scale. Ecol. Model. 209, 53–64.
Herbert, M.E., Ross, J.A., White, M.J., Royer, T.V., 2017. Modeling nutrient removal Hoogeweg, C.G., Denton, D.L., Breuer, R., Williams, W.M., TenBrook, P., Racke, K., 2012. De-
using watershed-scale implementation of the two-stage ditch. Ecol. Eng. 108, velopment of a spatial-temporal co-occurrence index to evaluate relative pesticide
358–369. risks to threatened and endangered species. Pesticide Regulation and the Endangered
Cibin, R., Chaubey, I., Helmers, M.J., Sudheer, K.P., White, M.J., Arnold, J.G., 2018. An im- Species Act. 1111, pp. 303–323.
proved representation of vegetative filter strips in SWAT. Trans. ASABE 61, Jones, R., Russell, M., 2001. Final Report of the FIFRA Environmental Model Validation Task
1017–1024. Force.
Coupe, R.H., 2007. Use of a watershed model to characterize the fate and transport of Kannan, N., White, S.M., Worrall, F., Whelan, M.J., 2006. Pesticide modelling for a small
fluometuron, a soil-applied cotton herbicide, in surface water. Int. J. Environ. Anal. catchment using SWAT-2000. J. Environ. Sci. Health B 41, 1049–1070.
Chem. 87, 883–896. Kannan, N., White, S.M., Whelan, M.J., 2007. Predicting diffuse-source transfers of surfac-
Davie-Martin, C.L., Hageman, K.J., Chin, Y.-P., Rougé, V., Fujita, Y., 2015. Influence of tem- tants to surface waters using SWAT. Chemosphere 66, 1336–1345.
perature, relative humidity, and soil properties on the soil–air partitioning of Karickhoff, S.W., Brown, D.S., Scott, T.A., 1979. Sorption of hydrophobic pollutants on nat-
semivolatile pesticides: laboratory measurements and predictive models. Environ. ural sediments. Water Res. 13, 241–248.
Sci. Technol. 49, 10431–10439. Kim, D.-S., Kumar, A., Parab, R., Palmer, M., 2004. Simulation of atrazine discharge in the
Deletic, A., 2001. Modelling of water and sediment transport over grassed areas. J. Hydrol. Auglaize watershed using satellite-generated images. Bull. Environ. Contam. Toxicol.
248, 168–182. 73, 319–325.
Du, B., Saleh, A., Jaynes, B.D., Arnold, G.J., 2006. Evaluation of swat in simulating nitrate ni- King, K., Balogh, J., 2001. Water quality impacts associated with converting farmland and
trogen and atrazine fates in a watershed with tiles and potholes. Trans. ASABE 49, forests to turfgrass. Trans. ASAE 44, 569.
949. King, P.H., McCarty, P.L., 1968. A chromatographic model for predicting pesticide migra-
Fantke, P., Friedrich, R., Jolliet, O., 2012. Health impact and damage cost assessment of tion in soils. Soil Sci. 106, 248–261.
pesticides in Europe. Environ. Int. 49, 9–17. Larose, M., Heathman, G.C., Norton, L.D., Engel, B., 2007. Hydrologic and atrazine simulation
Farrand, D.T., Heidenreich, L.K., 2000. Monroe City Route J Watershed: Environmental of the Cedar Creek watershed using the SWAT model. J. Environ. Qual. 36, 521–531.
Analysis. Lauvernet, C., Munoz-Carpena, R., 2018. Shallow water table effects on water, sediment,
Ficklin, D., Luo, Y., Zhang, M., 2013a. Watershed modelling of hydrology and water quality and pesticide transport in vegetative filter strips — part 2: model coupling, applica-
in the Sacramento River watershed, California. Hydrol. Process. 27, 236–250. tion, factor importance, and uncertainty. Hydrol. Earth Syst. Sci. 22, 71–87.
Ficklin, D.L., 2010. Modeling the Impacts of Climate Change on Hydrology and Agricultural Leonard, R.A., Knisel, W.G., Still, D.A., 1987. Gleams — groundwater loading effects of ag-
Pollutant Runoff in California's Central Valley. Ph.D dissertation. University of Califor- ricultural management-systems. Trans. ASAE 30, 1403–1418.
nia, Davis. Lerch, R., Sadler, E., Kitchen, N., Sudduth, K., Kremer, R., Myers, D., Baffaut, C., Anderson, S.,
Ficklin, D.L., Luo, Y., Luedeling, E., Gatzke, S.E., Zhang, M., 2010. Sensitivity of agricultural Lin, C.-H., 2008. Overview of the Mark Twain Lake/Salt River basin conservation ef-
runoff loads to rising levels of CO2 and climate change in the San Joaquin Valley wa- fects assessment project. J. Soil Water Conserv. 63, 345–359.
tershed of California. Environ. Pollut. 158, 223–234. Lescot, J.M., Bordenave, P., Petit, K., Leccia, O., 2013. A spatially-distributed cost-
Ficklin, D.L., Luo, Y., Zhang, M., 2013b. Climate change sensitivity assessment of effectiveness analysis framework for controlling water pollution. Environ. Model
streamflow and agricultural pollutant transport in California's Central Valley using Softw. 41, 107–122.
Latin hypercube sampling. Hydrol. Process. 27, 2666–2675. Li, F., Liu, Y., Engel, B.A., Chen, J., Sun, H., 2019. Green infrastructure practices simulation of
FOCUS, E., 2001. FOCUS surface water scenarios in the EU evaluation process under 91/ the impacts of land use on surface runoff: case study in Ecorse River watershed,
414/EEC. Technical Report SANCO/4802/2001-rev. 2. European Commission. Michigan. J. Environ. Manag. 233, 603–611.
Fohrer, N., Dietrich, A., Kolychalow, O., Ulrich, U., 2014. Assessment of the environmental Li, Y., Tang, C., Wang, C., Tian, W., Pan, B., Hua, L., Lau, J., Yu, Z., Acharya, K., 2013. Assessing
fate of the herbicides flufenacet and metazachlor with the SWAT model. J. Environ. and modeling impacts of different inter-basin water transfer routes on Lake Taihu
Qual. 43, 75–85. and the Yangtze River, China. Ecol. Eng. 60, 399–413.
Gali, R.K., Cryer, S.A., Poletika, N.N., Dande, P.K., 2016. Modeling pesticide runoff from Ligaray, M., Baek, S.S., Kwon, H.O., Choi, S.D., Cho, K.H., 2016. Watershed-scale modeling
small watersheds through field-scale management practices: Minnesota watershed on the fate and transport of polycyclic aromatic hydrocarbons (PAHs). J. Hazard.
case study with chlorpyrifos. Air Soil Water Res. 9, ASWR.S32777. Mater. 320, 442–457.
Ganzelmeier, H., Rautmann, D., Spangenberg, R., Streloke, M., Herrmann, M., Ligaray, M., Kim, M., Baek, S., Ra, J.-S., Chun, J., Park, Y., Boithias, L., Ribolzi, O., Chon, K.,
Wenzelburger, H., 1995. Untersuchungen zur Abtrift von Pflanzenschutzmitteln: Cho, K., 2017. Modeling the fate and transport of malathion in the Pagsanjan-
Ergebnisse eines bundesweiten Versuchsprogrammes. Mitteilungen aus der Lumban Basin, Philippines. Water 9, 451.
Biologischen Bundesanstalt für Land-und Forstwirtschaft Berlin-Dahlem. Linders, J., Adriaanse, P., Allen, R., Capri, E., Gouy, V., Hollis, J., Jarvis, N., Klein, M., Lolos, P.,
Gerecke, A.C., Scharer, M., Singer, H.P., Muller, S.R., Schwarzenbach, R.P., Sagesser, M., Maier, W., 2003. FOCUS surface water scenarios in the EU evaluation process under
Ochsenbein, U., Popow, G., 2002. Sources of pesticides in surface waters in 91/414/EEC. Sanco/4802/2001-rev. 2.
Liu, Y., Bralts, V.F., Engel, B.A., 2015. Evaluating the effectiveness of management practices Proia, L., Osorio, V., Soley, S., Köck-Schulmeyer, M., Pérez, S., Barceló, D., Romaní, A.M.,
on hydrology and water quality at watershed scale with a rainfall-runoff model. Sci. Sabater, S., 2013. Effects of pesticides and pharmaceuticals on biofilms in a highly im-
Total Environ. 511, 298–308. pacted river. Environ. Pollut. 178, 220–228.
Liu, Y., Engel, B.A., Flanagan, D.C., Gitau, M.W., McMillan, S.K., Chaubey, I., 2017. A review Quansah, J., Engel, A., Chaubey, I., 2008. Tillage practices usage in early warning prediction
on effectiveness of best management practices in improving hydrology and water of atrazine pollution. Trans. ASABE 51, 1311.
quality: needs and opportunities. Sci. Total Environ. 601–602, 580–593. Quilbé, R., Rousseau, A.N., Lafrance, P., Leclerc, J., Amrani, M., 2006. Selecting a pesticide
Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S., Bidoglio, G., 2009. EU-wide fate model at the watershed scale using a multi-criteria analysis. Water Qual. Res.
survey of polar organic persistent pollutants in European river waters. Environ. J. Can. 41, 283–295.
Pollut. 157, 561–568. Ramanarayanan, T., Narasimhan, B., Srinivasan, R., 2005. Characterization of fate and
Love, B.J., Einheuser, M.D., Nejadhashemi, A.P., 2011. Effects on aquatic and human transport of isoxaflutole, a soil-applied corn herbicide, in surface water using a water-
health due to large scale bioenergy crop expansion. Sci. Total Environ. 409, shed model. J. Agric. Food Chem. 53, 8848–8858.
3215–3229. Rathore, H.S., Nollet, L.M., 2012. Pesticides: Evaluation of Environmental Pollution. CRC
Luo, Y., Zhang, M., 2009a. Management-oriented sensitivity analysis for pesticide trans- Press.
port in watershed-scale water quality modeling using SWAT. Environ. Pollut. 157, Reichenberger, S., Bach, M., Skitschak, A., Frede, H.-G., 2007. Mitigation strategies to re-
3370–3378. duce pesticide inputs into ground-and surface water and their effectiveness; a re-
Luo, Y., Zhang, M., 2009b. Multimedia transport and risk assessment of organophosphate view. Sci. Total Environ. 384, 1–35.
pesticides and a case study in the northern San Joaquin Valley of California. Rittenburg, R.A., Squires, A.L., Boll, J., Brooks, E.S., Easton, Z.M., Steenhuis, T.S., 2015. Agri-
Chemosphere 75, 969–978. cultural BMP effectiveness and dominant hydrological flow paths: concepts and a re-
Luo, Y., Zhang, X., Liu, X., Ficklin, D., Zhang, M., 2008. Dynamic modeling of organophos- view. J. Am. Water Res. Assoc. 51, 305–329.
phate pesticide load in surface water in the northern San Joaquin Valley watershed Sakaguchi, A., Eguchi, S., Kasuya, M., 2014a. Examination of the water balance of irrigated
of California. Environ. Pollut. 156, 1171–1181. paddy fields in SWAT 2009 using the curve number procedure and the pothole mod-
Luo, Y., Spurlock, F., Jiang, W., Jorgenson, B.C., Young, T.M., Gan, J., Gill, S., Goh, K.S., 2013. ule. Soil Sci. Plant Nutr. 60, 551–564.
Pesticide washoff from concrete surfaces: literature review and a new modeling ap- Sakaguchi, A., Eguchi, S., Kato, T., Kasuya, M., Ono, K., Miyata, A., Tase, N., 2014b. Develop-
proach. Water Res. 47, 3163–3172. ment and evaluation of a paddy module for improving hydrological simulation in
Maringanti, C., Chaubey, I., Popp, J., 2009. Development of a multiobjective optimization SWAT. Agric. Water Manag. 137, 116–122.
tool for the selection and placement of best management practices for nonpoint Schulz, R., Peall, S.K., 2001. Effectiveness of a constructed wetland for retention of
source pollution control. Water Resour. Res. 45. nonpoint-source pesticide pollution in the Lourens River catchment, South Africa. En-
Maringanti, C., Chaubey, I., Arabi, M., Engel, B., 2011. Application of a multi-objective op- viron. Sci. Technol. 35, 422–426.
timization method to provide least cost alternatives for NPS pollution control. Envi- Seibert, J., McDonnell, J.J., 2002. On the dialog between experimentalist and modeler in
ron. Manag. 48, 448–461. catchment hydrology: use of soft data for multicriteria model calibration. Water
Maund, S., Mackay, N., 1998. Aquatic risk assessment and risk management for pesticides. Resour. Res. 38.
Issues in Environmental Science and Technology (United Kingdom). Silva, V., Mol, H.G.J., Zomer, P., Tienstra, M., Ritsema, C.J., Geissen, V., 2019. Pesticide resi-
Moore, M.T., Denton, D.L., Cooper, C.M., Wrysinski, J., Miller, J.L., Werner, I., Horner, G., dues in European agricultural soils — a hidden reality unfolded. Sci. Total Environ.
Crane, D., Holcomb, D.B., Huddleston III, G.M., 2011. Use of vegetated agricultural 653, 1532–1545.
drainage ditches to decrease pesticide transport from tomato and alfalfa fields in Cal- Snyder, N., Williams, W., 2004. Exposure assessment model for diazinon sources in the
ifornia, USA. Environ. Toxicol. Chem. 30, 1044–1049. Sacramento River Basin's main drainage canal. Sacramento River Watershed Pro-
Morselli, M., Semplice, M., De Laender, F., Van den Brink, P.J., Di Guardo, A., 2015. Impor- gram, Sacramento, CA.
tance of environmental and biomass dynamics in predicting chemical exposure in Solomon, K., Giesy, J., Jones, P., 2000. Probabilistic risk assessment of agrochemicals in the
ecological risk assessment. Sci. Total Environ. 526, 338–345. environment. Crop Prot. 19, 649–655.
Morselli, M., Terzaghi, E., Galimberti, F., Di Guardo, A., 2018a. Pesticide fate in cultivated Stehle, S., Knäbel, A., Schulz, R., 2013. Probabilistic risk assessment of insecticide concen-
mountain basins: the improved DynAPlus model for predicting peak exposure and trations in agricultural surface waters: a critical appraisal. Environ. Monit. Assess. 185,
directing sustainable monitoring campaigns to protect aquatic ecosystems. 6295–6310.
Chemosphere 210, 204–214. Stone, W.W., Gilliom, R.J., Ryberg, K.R., 2014. Pesticides in U.S. streams and rivers: occur-
Morselli, M., Vitale, C.M., Ippolito, A., Villa, S., Giacchini, R., Vighi, M., Di Guardo, A., 2018b. rence and trends during 1992–2011. Environ. Sci. Technol. 48, 11025–11030.
Predicting pesticide fate in small cultivated mountain watersheds using the DynAPlus Suárez, L., 2005. PRZM-3, a Model for Predicting Pesticide and Nitrogen Fate in the Crop
model: toward improved assessment of peak exposure. Sci. Total Environ. 615, Root and Unsaturated Soil Zones: Users' Manual for Release 3.12.2. Ecosystems Re-
307–318. search Division, National Exposure Research Laboratory, Washington, DC.
Moser, A., Wemyss, D., Scheidegger, R., Fenicia, F., Honti, M., Stamm, C., 2018. Modelling Sullivan, D.J., Vecchia, A.V., Lorenz, D.L., Gilliom, R.J., Martin, J.D., 2009. Trends in pesticide
biocide and herbicide concentrations in catchments of the Rhine basin. Hydrol. concentrations in corn-belt streams, 1996–2006. U.S. Geological Survey.
Earth Syst. Sci. 22, 4229–4249. Teske, M.E., Bird, S.L., Esterly, D.M., Curbishley, T.B., Ray, S.L., Perry, S.G., 2002. AgDRIFT®:
Muñoz-Carpena, R., Fox, G.A., Ritter, A., Perez-Ovilla, O., Rodea-Palomares, I., 2018. Effect a model for estimating near-field spray drift from aerial applications. Environ.
of vegetative filter strip pesticide residue degradation assumptions for environmental Toxicol. Chem. 21, 659–671.
exposure assessments. Sci. Total Environ. 619–620, 977–987. Toul, J., Bezděk, J., Kovářová, M., Boháček, Z., Hanák, J., Milička, J., Müller, P., 2003. Sorption
Neeser, C., Dille, J.A., Krishnan, G., Mortensen, D.A., Rawlinson, J.T., Martin, A.R., Bills, L.B., of hydrophobic organic pollutants on soils and sediments. Bull. Geosci. 78, 205–223.
2004. WeedSOFT®: a weed management decision support system. Weed Sci. 52, Tsuchiya, R., Kato, T., Jeong, J., 2016. Development of SWAT-PADDY for Simulating Low-
115–122. land Paddy Fields.
Nicholls, P.H., Briggs, G.G., Evans, A.A., 1984. The influence of water solubility on the Tsuchiya, R., Kato, T., Jeong, J., Arnold, J.G., 2018. Development of SWAT-Paddy for simu-
movement and degradation of simazine in a fallow soil. Weed Res. 24, 37–49. lating lowland Paddy fields. Sustainability 10, 3246.
O'Brien, D., Lewis, S., Davis, A., Gallen, C., Smith, R., Turner, R., Warne, M., Turner, S., Tu, L.H., Boulange, J., Iwafune, T., Yadav, I.C., Watanabe, H., 2018. Improvement and appli-
Caswell, S., Mueller, J.F., Brodie, J., 2016. Spatial and temporal variability in pesticide cation of the PCPF-1@SWAT2012 model for predicting pesticide transport: a case
exposure downstream of a heavily irrigated cropping area: application of different study of the Sakura River watershed. Pest Manag. Sci. 74, 2520–2529.
monitoring techniques. J. Agric. Food Chem. 64, 3975–3989. Van Griensven, A., Holvoet, K., Yang, J., 2006a. Open sources model codes: critical review
Ogura, N., 1970. On the presence of 0.1–0.5-μ dissolved organic matter in seawater. on the routing modules of the Soil and Water Assessment Tool (SWAT). 7th Interna-
Limnol. Oceanogr. 15, 476–479. tional Conference on Hydroinformatics HIC.
Orlando, J.L., Kuivila, K.M., 2004. Changes in rice pesticide use and surface water concen- Van Griensven, A., Meixner, T., Grunwald, S., Bishop, T., Diluzio, M., Srinivasan, R., 2006b. A
trations in the Sacramento River watershed, California. US Geological Survey. US De- global sensitivity analysis tool for the parameters of multi-variable catchment
partment of the Interior. models. J. Hydrol. 324, 10–23.
Ouyang, W., Cai, G., Tysklind, M., Yang, W., Hao, F., Liu, H., 2017a. Temporal-spatial pat- Vazquez-Amabile, G., Engel, B.A., Flanagan, D.C., 2006. Modeling and risk analysis of
terns of three types of pesticide loadings in a middle-high latitude agricultural water- nonpoint-source pollution caused by atrazine using SWAT. Trans. ASABE 49, 667.
shed. Water Res. 122, 377–386. Vernier, F., Leccia-Phelpin, O., Lescot, J.-M., Minette, S., Miralles, A., Barberis, D., Scordia, C.,
Ouyang, W., Gao, X., Wei, P., Gao, B., Lin, C.Y., Hao, F.H., 2017b. A review of diffuse pollu- Kuentz-Simonet, V., Tonneau, J.-P., 2017. Integrated modeling of agricultural scenar-
tion modeling and associated implications for watershed management in China. ios (IMAS) to support pesticide action plans: the case of the Coulonge drinking water
J. Soils Sediments 17, 1527–1536. catchment area (SW France). Environ. Sci. Pollut. Res. 24, 6923–6950.
Parker, R., Arnold, J.G., Barrett, M., Burns, L., Carrubba, L., Neitsch, S.L., Snyder, N.J., Wang, R., 2010. Modeling Hydrologic and Water Quality Responses to Changing Climate
Srinivasan, R., 2007. Evaluation of three watershed-scale pesticide environmental and Land Use/Cover in the Wolf Bay Watershed, South Alabama. Master Thesis. Au-
transport and fate models. J. Am. Water Resour. Assoc. 43, 1424–1443. burn University, Auburn, Alabama.
Payraudeau, S., Gregoire, C., 2012. Modelling pesticides transfer to surface water at the Wang, R., Kalin, L., 2018. Combined and synergistic effects of climate change and urbani-
catchment scale: a multi-criteria analysis. Agron. Sustain. Dev. 32, 479–500. zation on water quality in the Wolf Bay watershed, southern Alabama. J. Environ. Sci.
Petersen, J., Grant, R., Larsen, S.E., Blicher-Mathiesen, G., 2012. Sampling of herbicides in 64, 107–121.
streams during flood events. J. Environ. Monit. 14, 3284–3294. Wang, R., Yuan, Y., Luo, Y., Pitchford, A., Bingner, R.L., Denton, D., Yen, H., Zhang, M., 2018.
Pingali, P.L., Roger, P.A., 2012. Impact of Pesticides on Farmer Health and the Rice Environ- Tiered approaches in analyzing rice field pesticide fate and transport for ecological
ment. 7. Springer Science & Business Media. risk assessment. Managing and Analyzing Pesticide Use Data for Pest Management,
Popp, J., Peto, K., Nagy, J., 2013. Pesticide productivity and food security. A review. Agron. Environmental Monitoring, Public Health, and Public Policy. 1283. American Chem-
Sustain. Dev. 33, 243–255. ical Society, pp. 347–377.
Prato, T., Woo, B.-J., 2008. Integrated analysis of weed control practices for reducing atra- Wang, R., Chen, H., Luo, Y., Yen, H., Arnold, J., Zhang, M., 2019a. Modeling pesticide fate
zine contamination in an agricultural watershed. J. Soil Water Conserv. 63, 99–104. and transport at watershed scale via SWAT — general applications and mitigation
strategies (accepted). Pesticides in Surface Water: Monitoring, Modeling, Risk Assess- Yen, H., White, M.J., Arnold, J.G., Keitzer, S.C., M-VV, Johnson, Atwood, J.D., Daggupati, P.,
ment, and Management. American Chemical Society. Herbert, M.E., Sowa, S.P., Ludsin, S.A., 2016a. Western Lake Erie Basin: soft-data-
Wang, R., Luo, Y., Chen, H., Yuan, Y., Bingner, R.L., Denton, D., Locke, M., Zhang, M., 2019b. constrained, NHDPlus resolution watershed modeling and exploration of applicable
Environmental fate and impact assessment of thiobencarb application in California conservation scenarios. Sci. Total Environ. 569, 1265–1281.
rice fields using RICEWQ. Sci. Total Environ. 664, 669–682. Yen, H., White, M.J., Ascough, J.C., Smith, D.R., Arnold, J.G., 2016b. Augmenting watershed
Wang, R.Y., Kalin, L., 2011. Modelling effects of land use/cover changes under limited data. model calibration with incorporation of ancillary data sources and qualitative soft
Ecohydrology 4, 265–276. data sources. J. Am. Water Resour. Assoc. 52, 788–798.
Wang, R.Y., Kalin, L., Kuang, W.H., Tian, H.Q., 2014. Individual and combined effects of Yen, H., Lu, S., Feng, Q., Wang, R., Gao, J., Brady, D.M., Sharifi, A., Ahn, J., Chen, S.-T., Jeong, J.,
land use/cover and climate change on Wolf Bay watershed streamflow in southern 2017. Assessment of optional sediment transport functions via the complex water-
Alabama. Hydrol. Process. 28, 5530–5546. shed simulation model SWAT. Water 9, 76.
Wang, R.Y., Bowling, L.C., Cherkauer, K.A., 2016a. Estimation of the effects of climate var- Yen, H., Wang, R., Feng, Q., Young, C.-C., Chen, S.-T., Tseng, W.-H., Wolfe III, J.E., White, M.J.,
iability on crop yield in the Midwest USA. Agric. For. Meteorol. 216, 141–156. Arnold, J.G., 2018. Input uncertainty on watershed modeling: evaluation of precipita-
Wang, R.Y., Cherkauer, K., Bowling, L., 2016b. Corn response to climate stress detected tion and air temperature data by latent variables using SWAT. Ecol. Eng. 122, 16–26.
with satellite-based NDVI time series. Remote Sens. 8. Yuan, Y., Wang, R., Cooter, E., Ran, L., Daggupati, P., Yang, D., Srinivasan, R., Jalowska, A.,
Watanabe, H., Takagi, K., 2000. A simulation model for predicting pesticide concentrations 2018. Integrating multimedia models to assess nitrogen losses from the Mississippi
in paddy water and surface soil. I. Model development. Environ. Technol. 21, River basin to the Gulf of Mexico. Biogeosciences 15, 7059–7076.
1379–1391. Zhang, M.H., Zeiss, M.R., Geng, S., 2015. Agricultural pesticide use and food safety:
Whitfield Aslund, M., Breton, R.L., Padilla, L., Winchell, M., Wooding, K.L., Moore, D.R.J., California's model. J. Integr. Agric. 14, 2340–2357.
Teed, R.S., Reiss, R., Whatling, P., 2017. Ecological risk assessment for Pacific salmon Zhang, X., Goh, K.S., 2015. Evaluation of three models for simulating pesticide runoff from
exposed to dimethoate in California. Environ. Toxicol. Chem. 36, 532–543. irrigated agricultural fields. J. Environ. Qual. 44, 1809–1820.
Winchell, M.F., Pai, N., Brayden, B.H., Stone, C., Whatling, P., Hanzas, J.P., Stryker, J.J., 2018a. Zhang, X., Zhang, M., 2011. Modeling effectiveness of agricultural BMPs to reduce sedi-
Evaluation of watershed-scale simulations of in-stream pesticide concentrations from ment load and organophosphate pesticides in surface runoff. Sci. Total Environ.
off-target spray drift. J. Environ. Qual. 47, 79–87. 409, 1949–1958.
Winchell, M.F., Peranginangin, N., Srinivasan, R., Chen, W., 2018b. Soil and Water Assess- Zhang, X., Liu, X., Luo, Y., Zhang, M., 2008. Evaluation of water quality in an agricultural
ment Tool model predictions of annual maximum pesticide concentrations in high watershed as affected by almond pest management practices. Water Res. 42,
vulnerability watersheds. Integr. Environ. Assess. Manag. 14, 358–368. 3685–3696.
Xie, X.H., Cui, Y.L., 2011. Development and test of SWAT for modeling hydrological pro- Zhang, X., Luo, Y., Goh, K.S., 2018. Modeling spray drift and runoff-related inputs of pesti-
cesses in irrigation districts with paddy rice. J. Hydrol. 396, 61–71. cides to receiving water. Environ. Pollut. 234, 48–58.
Xu, F., Liang, X.M., Su, F., Zhang, Q., Lin, B.C., Wu, W.Z., Yediler, A., Kettrup, A., 1999. A col- Zhang, X.Y., Starner, K., Goh, K.S., Gill, S., 2012. Analysis of diazinon agricultural use in re-
umn method for determination of soil organic partition coefficients of eight pesti- gions of frequent surface water detections in California, USA. Bull. Environ. Contam.
cides. Chemosphere 39, 787–794. Toxicol. 88, 333–337.

Science of The Total Environment

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Science of The Total Environment

Uploaded by

Copyright:

Available Formats

Science of the Total Environment 669 (2019) 512–526

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

A review of pesticide fate and transport simulation at watershed level

• Current pesticide modeling status and

Fig. 1. Global applications of SWAT for pesticide modeling.

Fig. 2. Categories of recent published SWAT pesticide modeling studies.

Case study Pesticide Type Area Size Purpose

You might also like