Environmental Modelling & Software 96 (2017) 335e346

Contents lists available at ScienceDirect

Environmental Modelling & Software

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

Integrating water supply constraints into irrigated agricultural

simulations of California
Jonathan M. Winter a, *, Charles A. Young b, Vishal K. Mehta b, Alex C. Ruane c,
Marzieh Azarderakhsh d, Aaron Davitt e, Kyle McDonald f, Van R. Haden g,
Cynthia Rosenzweig c
a
Department of Geography, Dartmouth College, Hanover, NH 03755, USA
b
Stockholm Environment Institute, Davis, CA 95616, USA
c
NASA Goddard Institute for Space Studies, New York, NY 10025, USA
d
Fairleigh Dickinson University, Teaneck, NJ 07666, USA
e
The Graduate Center, City University of New York, New York, NY, 10016, USA
f
The City College of New York, New York, NY, 10031, USA
g
The Ohio State University, Wooster, OH, 44691, USA

a r t i c l e i n f o a b s t r a c t

Article history: Simulations of irrigated croplands generally lack key interactions between water demand from plants
Received 5 January 2017 and water supply from irrigation systems. We coupled the Water Evaluation and Planning system
Received in revised form (WEAP) and Decision Support System for Agrotechnology Transfer (DSSAT) to link regional water sup-
19 June 2017
plies and management with field-level water demand and crop growth. WEAP-DSSAT was deployed and
Accepted 29 June 2017
Available online 18 July 2017
evaluated over Yolo County in California for corn, rice, and wheat. WEAP-DSSAT is able to reproduce the
results of DSSAT under well-watered conditions and reasonably simulate observed mean yields, but has
difficulty capturing yield interannual variability. Constraining irrigation supply to surface water alone
Keywords:
Agriculture
reduces yields for all three crops during the 1987e1992 drought. Corn yields are reduced proportionally
Irrigation with water allocation, rice yield reductions are more binary based on sufficient water for flooding, and
Water resources management wheat yields are least sensitive to irrigation constraints as winter wheat is grown during the wet season.
Crop model © 2017 Elsevier Ltd. All rights reserved.
Evapotranspiration
United States

Software availability Program size: 500 MB

Name of software: WEAP-DSSAT 1. Introduction

Developers: Charles Young, Jonathan Winter, Vishal Mehta, Alex
Ruane Irrigated farms account for 80%e90% of consumptive water use
Contact address: Dartmouth College, 6017 Fairchild Hall, Hanover, in the United States and $118.5 billion of US agricultural production
NH 03755 (Solley et al., 1998; Schaible and Aillery, 2012). Despite the high
Telephone: þ1 603-646-6456 productivity of irrigated croplands, agriculture is typically the
Email: jwinter@dartmouth.edu lowest value sector in a water resources system, and, subject to
Year first available: 2016 water regulations and rights, vulnerable to reductions during
Hardware required: PC, Intel with 4 GB RAM recommended drought. A major challenge for the hydrologic and agricultural
Software required: Microsoft Windows communities is assessing the effects of climate change on the
Availability and cost: Licensed software sustainability of regional water resources and irrigated agricultural
Program language: Fortran, Python, and Delphi land (Walthall et al., 2013). A key component of this challenge is the
fact that most agricultural models that have sophisticated repre-
sentations of crop physiology, management, and yield, and are
* Corresponding author. 6017 Fairchild Hall, Hanover, NH 03755, USA. thoroughly evaluated at the field scale, lack constraints on irriga-
E-mail address: jwinter@dartmouth.edu (J.M. Winter). tion supply (Winter et al., 2017). Many crop models are run with

http://dx.doi.org/10.1016/j.envsoft.2017.06.048
1364-8152/© 2017 Elsevier Ltd. All rights reserved.
336 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346
scheduled irrigation or unlimited automatic irrigation (e.g.,
Bondeau et al., 2007; Jones et al., 2003; Raes et al., 2009), each of
which has important disadvantages. With scheduled irrigation the
dates and amounts of irrigation are prescribed in advance. This is
problematic in locations and years where this information is not
readily available, or in predictive applications (e.g., seasonal fore-
casts, climate change projections). Automatic irrigation doesn't
require prior knowledge of water applications, instead relying on a
rule-based approach (often linked to soil moisture) to irrigate from
an unlimited water supply. This leads to inconsistent, and occa-
sionally implausible, biases and errors that muddle the picture for
stakeholders and policymakers seeking to understand agricultural
sustainability issues.
Multiple studies have addressed the impacts of climate on irri-
gated agriculture at scales ranging from regional to global; how-
ever, few have explicitly coupled a crop model to a calibrated
hydrologic and water resources allocation model. Elliott et al.
(2014a) compared water supply projections from ten global hy-
drologic models and water demand projections from six global
gridded crop models. Results suggested that the effects of reduced
irrigation were comparable to the direct impacts of climate change
on global production of maize, soybean, wheat, and rice, with some
acute regional impacts. However, Deryng et al. (2016) noted dis-
crepancies in projections of agricultural water supply and demand
due to varying responses of crop water use to increased CO2 by
global gridded crop models, and the lack of response of crop water
use to increased CO2 by some global hydrologic models. Piontek
et al. (2014) explored climate change impacts on multiple sectors,
including water availability and agriculture. Several areas were
found to have overlapping risk for severe change both in water
availability and climate, including the southern Amazon Basin and
regions in South Asia. Wada et al. (2013) used a set of seven global
hydrologic models to explore the change in irrigation water de-
mand by the end of the century, finding a considerable increase
during summer months in the northern hemisphere. Huntington
and Niswonger (2012) focused on the seasonal timing of stream-
flow and surface and groundwater interactions over the Western
United States. Specifically, they used an integrated model of surface
water and groundwater inclusive of snowpack and snowmelt
forced with twelve general circulation model (GCM) projections.
Future climate was shown to decrease summertime flows by more
than 30% averaged across the ensemble, with reductions found
even in GCM simulations that projected increased annual precipi-
tation. Groundwater is a critical source of water for irrigated agri-
culture, and groundwater management remains a salient issue for
sustainable irrigated agriculture (Do € ll et al., 2012). Taylor et al.
(2013) outline the complexity of groundwater response to climate
change and human impacts, including annual precipitation and
streamflow; timing, intensity, and duration of precipitation and
streamflow; land use; snowpack; groundwater pumping; and sur-
face water irrigation. Groundwater pumping has been shown to be
unsustainable in the Central Valley of California; however, to date
the use of groundwater has been largely unrestricted in California
(Famiglietti, 2014).
The water-agriculture nexus has been identified as a high pri-
ority area within the Agricultural Model Intercomparison and
Improvement Project (AgMIP; Rosenzweig et al., 2015, 2013). The
importance of irrigated agriculture to global food production, as
well as the response of crop water supply and demand to climate,
necessitate explicitly simulating effects of water availability on
irrigated yields. In the following sections, we describe the devel-
opment, application, and evaluation of a coupled hydrologic, water
resources allocation, and crop model. The objective of creating this
coupled model is to more realistically simulate irrigated agricul-
tural yields, and specifically to develop a modeling system in which
water shortages (e.g., decreased precipitation, enhanced evapo-
transpiration, changes in allocation) directly impact irrigated crop
yields, with applications for identifying and testing policy and
management approaches.
2. Model description and development
To simulate water demand and supply for irrigated agriculture, a
model must link information about regional water supplies and
management with field-level water demand and crop response as it
develops throughout the season. To accomplish this, we coupled
the Decision Support System for Agrotechnology Transfer (DSSAT;
Hoogenboom et al., 2012; Jones et al., 2003) to the Water Evaluation
and Planning system (WEAP; Yates et al., 2005a, 2005b). Below we
describe the development of the coupled model, WEAP-DSSAT, as
well as our iterative simulation approach, which together add
water supply constraints to automatic irrigation in DSSAT and
detailed crop water use and yields to WEAP.
2.1. Water Evaluation and Planning system
The Water Evaluation and Planning system (WEAP) is used to
model water supply and management within WEAP-DSSAT. WEAP
is an object-orientated model that solves a time evolving mass
balance based on a water allocation objective function. At each time
step, hydrologic fluxes from the surface and near surface are passed
to appropriate river and groundwater objects, where they are
balanced with an objective function that maximizes satisfaction of
demand and instream flow requirements, subject to supply prior-
ities based on water rights and regulations, demand site prefer-
ences, mass balances, and other constraints (Yates et al., 2005a,
2005b). WEAP has a flexible time step that can range from daily
to annual, which also determines the time scale over which water
allocation is calculated.
WEAP divides study regions into user-defined sub-catchments;
groundwater basins; irrigated areas; urban/export uses; environ-
mental requirements; and water system elements such as canals,
diversions, and reservoirs. Water supply in WEAP is provided by an
embedded hydrologic model forced by an external climate dataset
that simulates runoff, groundwater-surface water interactions, and
snow processes. Before the addition of DSSAT, agricultural water
demand in WEAP could be simulated using a variety of approaches
that range in complexity from a simple crop coefficient method to
more complex hydrology-based algorithms that incorporate runoff,
infiltration, soil moisture storage, deep percolation, and evapo-
transpiration as a function of soil moisture status.
The WEAP framework readily accommodates user specified
models, or modules, that can be plugged into and controlled by
WEAP's water budget and allocation logic. WEAP has been used for
range of applications, including large river basins with substantial
irrigation such as California's Central Valley (Mulligan et al., 2011;
Sandoval-Solis et al., 2010), village scale modeling of community
livelihoods (Varela-Ortega et al., 2011), and the exploration of
climate change impacts on hydropower generation (Mehta et al.,
2011). WEAP deployed over the Central Valley has been shown to
adequately represent both local and regional water balances, and
the allocation of groundwater and surface water supplies (Purkey
et al., 2008; Yates et al., 2009, 2008).
A feature critical to assessing the impacts of climate change on
agriculture, and which is notably lacking in the California imple-
mentation of WEAP, is a representation of plant physiology in plant
water use and yields. WEAP simulates crop water use by assigning a
seasonal cycle of agricultural vegetation to every user-defined sub-
catchment. For each time step, potential evapotranspiration is
scaled by a crop coefficient. This approach, while reasonable for
 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346 337

hydrologic applications, does not include climate impacts on agri-

cultural productivity or capture the effects of a changing environ-
ment on plant physiology, such as increased carbon assimilation
and water use efficiency under enhanced carbon dioxide concen-
trations or increased transpiration under warmer temperatures
(Kimball, 2011; Tubiello et al., 2007). These effects are critical for
projections of irrigation demand, crop yields, and water manage-
ment. Coupling DSSAT, a process-based crop model, to WEAP ex-
pands WEAP's capabilities beyond hydrological applications, and
makes it ideally suited for exploring water-agriculture interactions.

2.2. Decision Support System for Agrotechnology Transfer

Farm-level crop demands and the growth processes leading to

yield are handled within WEAP-DSSAT by the Crop Estimation
through Resource and Environment Synthesis (CERES) component
of the Decision Support System for Agrotechnology Transfer
(DSSAT). DSSAT is a set of applications designed to understand,
predict, and manage cropping systems at the field scale
(Hoogenboom et al., 2012; Jones et al., 2003). DSSAT is essentially a
point-based (single hectare) model that runs on a daily time step,
with yields dependent on surface meteorology, a detailed soil
profile, cultivar genetics, and crop management. DSSAT simulates
biological processes that track carbon, nitrogen, water, and energy
budgets (as well as associated stresses) through a plant's devel-
opmental stages as it interacts with the environment and is
managed by a grower. DSSAT simulates a closed water balance that
has compared well with soil moisture measurements in the field
(Casanova et al., 2006; Sau et al., 2004). DSSAT supports models for Fig. 1. WEAP-DSSAT irrigation flowchart. Components are color-coded based on the
42 unique crop species, including all the major staple crops, and has length of time step considered for action e blue is yearly, green is user-defined (in this
been used in agricultural systems around the world for a variety of case monthly), and red is daily.
applications, including climate change impacts (Mearns et al., 1996;
Parry et al., 2005; Rosenzweig et al., 2014; Ruane et al., 2013; White
and deep percolation as a function of percentage irrigation demand
et al., 2011). In our simulations we run the CERES-Maize, CERES-
fulfilled, or the portion of the required irrigation water that is
Wheat, and CERES-Rice applications of DSSAT, which have been
allocated based on supply, hereafter referred to as allocation. To
widely used in crop modeling and studies of California's Central
develop these equations, WEAP runs DSSAT once with unlimited or
Valley (e.g., Brumbelow and Georgakakos, 2001; Greenwald et al.,
full irrigation and once with zero irrigation for the entire water
2006; Lobell and Ortiz-Monasterio, 2007).
year. Linear equations are then generated that relate allocation to
surface runoff and deep percolation so that WEAP can calculate
2.3. Coupling DSSAT to WEAP
surface runoff and deep percolation from irrigation dates and
amounts passed by DSSAT. In addition, the full irrigation run
The link between WEAP and DSSAT was implemented by editing
transcribes the irrigation dates and amounts from the automatic
the source code of WEAP, written in Delphi, to write DSSAT input
irrigation algorithm assuming unlimited water to a DSSAT input file
files, call the DSSAT executable, and read DSSAT output files. Within
for the catchment (blue boxes in Fig. 1).
WEAP, catchment objects are used to represent hydrological pro-
Once the linear relationships are generated, the water allocation
cesses, including crop water use. Within each catchment climate is
problem is solved and the allocation is computed at a user-defined
assumed uniform. Sub-catchments represent unique combinations
time step (in this case monthly, green boxes in Fig. 1) for every
of hydrologic parameters (e.g., land cover, soils). It is at the sub-
catchment. For each allocation time step and catchment, WEAP-
catchment scale that hydrologic calculations are performed and
DSSAT checks the demand from the DSSAT full irrigation run
DSSAT was integrated into the WEAP framework. For the sub-
against the water supply for agriculture (water for other uses is
catchments in which DSSAT is used, DSSAT simulates the hydro-
considered unavailable) passed from WEAP. If the allocation is 100%
logic cycle including irrigation surface runoff, infiltration, evapo-
then the full irrigation results are stored for each day (red boxes in
transpiration, and deep percolation, with surface runoff and deep
Fig. 1), no additional DSSAT runs are conducted, and WEAP-DSSAT
percolation being passed back to WEAP. WEAP is used to calculate
moves to the next allocation time step. If the allocation is less
the flow of water in the stream network, storage of water in res-
than 100%, then DSSAT is rerun from the beginning of the water
ervoirs and groundwater, and allocation of water to demands for all
year with reduced irrigation. This reduction in irrigation is
sub-catchments, as well as hydrologic processes in the sub-
accomplished through the modification of DSSAT input files. For
catchments DSSAT is not used in. WEAP water allocation is solved
each instance of DSSAT run by WEAP, WEAP creates an input file
at a user-defined time step that can range from daily to yearly, in
that contains settings for the DSSAT simulation, including irrigation
contrast to the fixed daily time step of DSSAT. If the WEAP time step
dates and amounts. As described above, this input file is first
is longer than a day, DSSAT output is aggregated for input to WEAP.
created assuming unlimited water. WEAP-DSSAT is then run iter-
The process for conducting a WEAP-DSSAT simulation is
atively, stopping the simulation for any month that the allocation is
described in Fig. 1. Water allocation in WEAP is solved by a linear
less than 100%, editing the input file to irrigate on the same days as
program, so for each allocation time step (in this case, monthly) it is
the water unlimited run but with the irrigation amount scaled by
necessary to develop linear equations that represent surface runoff
338 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346
allocation, and then rerunning from the beginning of the water year
using surface runoff and deep percolation from the reduced allo-
cation simulation.
This process is repeated each month until the simulation ends or
the harvest date is reached. Because DSSAT cannot run a single
simulation with both prescribed and automatic irrigation, the
DSSAT algorithm for automatically triggering irrigation was
implemented in the coupling of WEAP to DSSAT, allowing WEAP-
DSSAT simulations to use prescribed irrigation amounts for past
monthly time steps and automatic irrigation amounts for the cur-
rent monthly time step. The result of this iterative approach is a
time series of irrigation dates determined using the same automatic
irrigation algorithm in DSSAT with irrigation amounts adjusted
based on allocation. Once the simulation finishes, some key DSSAT
output variables are aggregated by WEAP-DSSAT into a single file
for analysis, including soil moisture from the DSSAT soil water file;
planting date, harvest date, and season length from the DSSAT plant
growth file; evapotranspiration from the DSSAT soil-plant-
atmosphere file; and agronomic results (yield, biomass, etc.) from
the DSSAT summary file. We note that both WEAP and DSSAT are
relatively computationally efficient, so our iterative approach over
regional spatial scales such as the Central Valley is not limiting.
The final component of the WEAP-DSSAT coupling is the spec-
ification of soil moisture initial conditions for the next model year.
The use of continuously simulated soil moisture and nutrients, as
opposed to annually reinitialized simulations, has been shown by
Basso et al. (2015) to be extremely important for accurate crop
modeling. At the end of each yearly simulation, soil moisture values
for each soil layer are read from the soil water file and then used in
the next year's input file as the initial soil moisture conditions.
3. WEAP-DSSAT deployment over Yolo County and evaluation
datasets
We apply WEAP-DSSAT to a county in California's Central Valley
as a test case to explore the impacts of climate and water supply on
a productive and drought-prone irrigated agricultural system.
3.1. Study region
California's $42.5-billion agricultural sector is supported by a
complex water management system prone to acute shortages and
susceptible to climate change (Garfin et al., 2014; NASS, 2014). The
scale of water development in California is among the most sub-
stantial in the world, including massive shifts in water from one
basin to another over distances of hundreds of kilometers in order
to satisfy water demands (National Research Council, 2011). We
focus our test case deployment of WEAP-DSSAT on Yolo County, and
our agricultural analyses on Yolo County Flood Control and Water
Conservation District (YCFCWCD), shown in Fig. 2.
Yolo County is a well-studied, intensively managed, and water-
limited agricultural area in the Sacramento River basin in northern
Central Valley, with crops covering 57% of the land and agriculture
accounting for almost 95% of the County's total water withdrawal of
approximately 1.23 billion m3 yr1 (Mehta et al., 2013). YCFCWCD
covers 41% of Yolo County's irrigated area and is located in the
western-central portion of the county (Mehta et al., 2013).
YCFCWCD supplies surface water to its agricultural customers from
Cache Creek, primarily from Clear Lake and the Indian Valley
Reservoir. Despite this infrastructure, total irrigation demand ex-
ceeds what YCFCWCD can supply, and there have been three times
in the past 40 years with absolutely no water supplied by the
District (1977, 1990, 2014). Additional details on Yolo County and
YCFCWCD agriculture can be found in Mehta et al. (2013).
3.2. Model setup
WEAP-DSSAT is based on a WEAP model developed for Yolo
County (Mehta et al., 2013). In that model the entire area of Yolo
County as well as the portions of Lake County that provide water to
Yolo County were simulated using catchment objects. The bound-
aries of the areas represented by each catchment object are shown
in Fig. 2. In the upper Cache Creek watershed the model contains
reservoir objects for Clear Lake and Indian Valley Reservoir, the
main sources of surface water in Yolo County during the summer
months. Reservoir operations are controlled by the Solano Decree
which has been programmed into WEAP (Mehta et al., 2013).
Downstream, on the valley floor, the irrigated area is subdivided
into catchments that represent the Capay Valley, Yolo County Flood
Control and Water Conservation District, and all other irrigated area
in Yolo County. Water preferences are set in the model so that
surface water is used first if available, and groundwater is used
second. The model was calibrated using crop water use information
from the Department of Water Resources, reservoir storage infor-
mation, and historical diversions. Additional details on WEAP setup
and calibration can be found in Mehta et al. (2013).
WEAP-DSSAT was initialized October 1, 1980, and run through
September 30, 2008, capturing the 1981e2008 water years, with
WEAP run at a monthly time step to leverage the calibration and
evaluation of WEAP over Yolo County by Mehta et al. (2013) and for
computational efficiency, and DSSAT run at the default daily time
step. WEAP and DSSAT were forced using surface (2-m) air tem-
perature and precipitation from the Maurer et al. (2002) 1/8
gridded observational dataset, and surface wind speed, surface
relative humidity, and insolation from AgMERRA (Ruane et al.,
2015). Specifically, WEAP was forced using monthly temperature
and precipitation, and DSSAT was forced using daily temperature,
precipitation, wind speed, relative humidity, and insolation data,
allowing both models to respond consistently to changes in climate.
Input climate time series were extracted from gridded datasets
using the nearest neighbor of the WEAP catchment centroid. DSSAT
has primarily been used for staple crops, so our initial assessment of
climate impacts on irrigated agriculture is focused on corn (maize;
Zea mays), winter wheat (Triticum aestivum), and rice (Oryza sativa).
However, we note that specialty crops (e.g., tomatoes, almonds,
grapes) dominate agricultural production by value in Yolo County.
Crop management information is described in Table 1.
Both the planting and harvesting end date are consistent with
the crop phenology and typical planting dates in the region (Pen ~ a-
Barraga n et al., 2011; USDA NASS, 2010), and are expected not to be
met only in exceptional circumstances. Soil properties, including
soil moisture initialization, were defined for all crops using a
generic deep sandy loam profile. Details on the construction and
attributes of the soil profile can be found in the source code
(Hoogenboom et al., 2012). We focus on the role of water in limiting
crop production, so shut off all nutrient stresses in DSSAT under an
assumption of well-fertilized fields. Automatic irrigation for corn
and wheat is triggered when the fraction of soil moisture in the top
30 cm of the soil profile falls below 50% of soil saturation. Water is
then applied as furrow irrigation until the soil moisture level rea-
ches 100% saturated. Rice is flooded on the first day of the planting
window through harvest to a depth of 50 mm. Bund height is set to
100 mm, and percolation rate is assumed to be 4 mm day1. Sim-
ulations for each crop were run using the observed atmospheric
CO2 concentration time series of Keeling et al. (2001). We note that
while planting date, harvesting date, and irrigation applications
will vary by year due to weather, all other crop management is
static.
Crop varieties were selected for each of the three crops based on
the field and management characteristics of Yolo County. Field corn
 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346 339

Fig. 2. Modeled agricultural area (YCFCWCD) and Yolo and Lake County catchments, with inset of California's Central Valley and Yolo County.

Table 1
Crop management for WEAP-DSSAT simulations. Note that the planting for wheat occurs in the year previous to harvest year, and that crops are harvested automatically at
maturity. The harvesting end date is included to ensure crops that are planted but do not mature are cleared before the start of the next water year.

Crop Density [m2] Row Spacing [cm] Depth [cm] Planting Window Harvesting End Date

Corn 6 61 7 3/1e6/1 9/30

Rice 200 25 2.5 5/1e5/25 9/30
Wheat 270 16 3 10/20e12/31 9/30

in the Sacramento Valley is grown both for grain and silage, with
the harvested acreage planted for grain accounting for approxi-
mately 70% of total field corn harvested acreage. However, the split
in grain and silage production is variable, and often dual-purpose
varieties are planted that can be used for either grain or silage
based on market conditions (Frate and Schwanki, n.d.). Farmers in
this area generally use slow maturing corn varieties to take
advantage of the long growing season in the region (Frate and
Schwanki, n.d.; Pen ~ a-Barraga
n et al., 2011). We selected a corn
cultivar in DSSAT that contains the generic attributes of California
corn, but does not match a specific variety grown in Yolo County to
maintain flexibility for model applications beyond Yolo County but
within the Central Valley. The majority of California's wheat pro-
duction is hard red spring wheat planted in the fall (University of
California Division of Agriculture and Natural Resources, 2006).
Important characteristics for California wheat production include
short stature, lodging resistance, soil saturation tolerance, shatter
resistance, and nitrogen responsiveness (University of California
Division of Agriculture and Natural Resources, 2006). A medium
grain hard red spring wheat cultivar with the desired properties
was not available in DSSAT, so a modified hard red winter wheat
available in DSSAT was determined to be the best choice. Rice
production in California is dominated by semi-dwarf, medium-
grain varieties, with the majority classified as early maturity
(Espino et al., 2013; Geisseler and Horwath, 2013). These varieties
provide a high yield and quality of rice, as well as flexibility in
planting dates due to the relatively short time to maturity. As with
corn, a rice cultivar in DSSAT was selected that contains the key
properties of rice currently grown in California, but is not a specific
rice variety from Yolo County to ensure that the model is valid for
340 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346
applications throughout California.
Due to the unsustainable use of groundwater in California
(Famiglietti, 2014), we ran two sets of simulations to capture the
two extremes of groundwater supply. In the first, we automatically
irrigated with no limitations on water amount (WEAP-DSSAT Full).
In the second set of simulations (WEAP-DSSAT Surface), we high-
light the potential impacts of eliminating groundwater pumping by
irrigating exclusively with surface water. In these simulations,
crops are automatically irrigated until irrigation water demand
exceeds surface water supply, and then irrigation applications are
proportionally reduced after that based on the ratio of surface
water supply to irrigation water demand. We note that the areas
simulated are all within the surface irrigation canal service area of
YCFCWCD, so are therefore less sensitive to the elimination of
groundwater pumping than Yolo County overall. We further note
that the WEAP-DSSAT Surface simulations are hypothetical and
proportionally reducing irrigation applications is a simplistic
approach. A more practical modeling or scenario-based treatment
of groundwater, as well as representations of crop prioritization
(e.g., irrigating higher value fields or perennials first), fallowing,
and deficit irrigation, would be necessary for more realistic simu-
lations of Yolo County irrigated agriculture.
3.3. Observed yield and evapotranspiration datasets
To assess the accuracy of WEAP-DSSAT predictions of grain yield,
we compared simulated yields of corn, wheat, and rice to average
yields reported for Yolo County by the USDA's National Agricultural
Statistical Service (1980e2014) and observed yields from agro-
nomic field trials conducted in Yolo County. Field trial data for
irrigated corn (1994e2007) and irrigated wheat (1994e2013)
yields were obtained from a long-term experiment conducted at
the University of California's Russel Ranch Sustainable Agriculture
Facility (Scow, 2009). The corn varieties planted in this experiment
were long duration hybrids: Pioneer 3162 (1994e2000) and
ST7570RR (2001e2007). Over the period of this field trial three
hard red spring wheat varieties were used; Yolo (1994e2002),
Summit (2003e2005), and Cal Rojo (2006e2013). Observed rice
yields were obtained from a database of annual variety trials con-
ducted by the University of California at three field sites in Yolo
County (University of California Cooperative Extension, 2015):
Greer Farm (1981e1997, 1999e2003), Erdman Farm (2004e2005),
and Webster Farm (2006e2010, 2012e2014). Yield data from the
M-202 cultivar was used for all years. We note that while cultivars
are known for the field trials, we selected generic cultivars in
WEAP-DSSAT to keep our test case consistent with a regional
application of WEAP-DSSAT in which cultivars would not be
known. Generic, uncalibrated cultivars have been used extensively
in regional- to global-scale crop model simulations (e.g., Elliott
et al., 2014b; Kucharik, 2003; Rosenzweig et al., 2014); however,
we acknowledge that this simplification has consequences for
interpretating results. We did conduct an analysis to explore the
sensitivity of simulated yields to cultivar parameters. Specifically,
for corn and rice we increased P1 and P5 by 10% and decreased P1
and P5 by 10%, and for wheat we increased P5 by 10% and decreased
P5 by 10%. Corn yields were most sensitive to changes in P1 and P5,
approximately 15% on average, followed by rice, ~4% on average,
and with wheat being relatively insensitive to changes in P5, less
than 2% (not shown).
Some observed yield units needed to be converted for consis-
tency with WEAP-DSSAT, which simulates dry grain in kg ha1.
NASS corn was converted from bu ac1 to kg ha1 assuming 56 lb
bu1. NASS wheat was converted from bu ac1 to kg ha1 assuming
60 lb bu1. Rice field trial yield data were converted from units of lb
ac1 at 14% moisture to kg ha1 dry grain yield. All other
observations either needed no or simple unit conversions.
WEAP-DSSAT evapotranspiration in Yolo County was compared
to remotely sensed evapotranspiration estimates from the Terres-
trial Observation and Prediction System (TOPS) Satellite Irrigation
Management Support (SIMS; Melton et al., 2012). While both WEAP
and DSSAT simulate ET, we only consider ET output from DSSAT as it
should best represent plant water loss over agricultural areas. SIMS
integrates satellite, ground observations, and model results to
produce near real-time simulations and forecasts of environmental
conditions, including potential evapotranspiration (Nemani et al.,
2009). Crop evapotranspiration in SIMS is calculated using the
California Department of Water Resources Spatial California Irri-
gation Management Information System (Hart et al., 2009) and a
remotely sensed crop coefficient (Kc), described in more detail
below. This presents a marked improvement over the use of
weather station potential ET with a fixed Kc, which is spatially
coarse and does not account for crop rotation, phenology, and land
use changes between years.
SIMS combines TOPS with Landsat 5 Thematic Mapper (TM),
Landsat 7 Enhanced Thematic Mapper (ETMþ) and MODerate
resolution Imaging Spectroradiometer (MODIS) satellite imagery to
create a map product that covers approximately 4 million ha of
agricultural land in California at a 30-m resolution (Melton et al.,
2012). Landsat imagery provides the spatial resolution for field
scale measurements, while MODIS provides temporal resolution
with gap-filling due to the Landsat 7 Scan Line Corrector (SLC)
failure, thus providing timely data accessibility and applicability
(Melton et al., 2012). SIMS provides information on crop conditions
and water use through the TOPS SIMS website (http://ecocast.arc.
nasa.gov/dgw/sims/), including basal crop evapotranspiration
(ETcb; Fig. 3), basal crop coefficient (Kcb), fractional cover (FC),
normalized difference vegetation index (NDVI) from a combination
of Landsat and MODIS, and NDVI from Landsat only. We note that
ETcb doesn't account for soil evaporation, free water surface evap-
oration, or reduced transpiration from crop water stress unless the
stress affects crop development.
We obtained monthly SIMS ETcb for 2011e2014 for Yolo County.
Data for 2012 relied more heavily on MODIS as Landsat 5 data were
not available, and Landsat 8 was not yet collecting data. Crop
boundary information downloaded from Yolo County (http://www.
yolocounty.org) was used to identify and extract ETcb for each field
of corn, rice, and wheat, and then generate a county-scale mean of
ETcb for each crop. We caveat that due to temporal coverage of
climate data used to force WEAP-DSSAT, SIMS data does not overlap
with WEAP-DSSAT simulations, so seasonal cycles of ET for WEAP-
DSSAT (1981e2008) and SIMS (2011e2014) were calculated by crop
to produce an approximate comparison between remotely sensed
and modeled crop ET for corn, rice, and wheat in Yolo County.
4. Results and discussion
WEAP-DSSAT produces near-identical (within 1%) yields relative
to the original DSSAT model under well-watered conditions for
corn, rice, and wheat, with one exception (not shown). DSSAT ir-
rigates on the day of harvest in the wheat simulation for 2001,
while WEAP-DSSAT does not because it is unrealistic. This alters the
initial soil moisture conditions enough for 2002 to produce a 24%
increase in DSSAT wheat yield relative to WEAP-DSSAT. Below we
compare yields and evapotranspiration simulated by WEAP-DSSAT
to observations for Yolo County.
4.1. Yield evaluation
Fig. 4 shows WEAP-DSSAT simulated, NASS, and field trial yields
for corn, rice, and wheat, and Table 2 contains the means of and
 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346
Fig. 3. SIMS mean annual ETcb (mm/month) over Yolo County for 2011e2014.
correlations across simulated and observed yields. As described
above, we ran WEAP-DSSAT with both a full allocation of water,
representing surface water augmented by unlimited groundwater,
and then a surface water only allocation, which falls short of irri-
gation water demand in some years. We primarily analyze WEAP-
DSSAT simulations with a full allocation (WEAP-DSSAT Full)
because we are using irrigated field trial data, and believe full
irrigation is more relevant to NASS, and WEAP-DSSAT Full allows us
to confirm the consistency of WEAP-DSSAT to the offline version of
DSSAT when run with unlimited water. However, we also include
the results from WEAP-DSSAT irrigated exclusively with surface
water (WEAP-DSSAT Surface) in Fig. 4 and Table 2 for reference.
On average WEAP-DSSAT Full corn yields agree with field trials,
and are approximately 10% higher than NASS yields. Differences
between field trial and NASS yields could be the result of a wide
range of management and environmental factors. In addition, field
trials can be managed in a way that is not practical at larger scales,
and NASS averages yields over the entire county. WEAP-DSSAT Full
overestimates rice yields by 13% and 26% relative to field trials and
NASS, respectively. Unlike corn and wheat, water stress is probably
not a cause of differences across yields as all fields are flooded.
WEAP-DSSAT Surface better simulates mean yields of corn when
compared to NASS and rice when compared to either observational
dataset, but low correlations suggest this accuracy is false. WEAP-
DSSAT Full simulates wheat well, with yields only slightly lower
than both NASS (7%) and field trials (4%). When rounded, wheat
yields produced by WEAP-DSSAT Full and WEAP-DSSAT Surface are
identical.
WEAP-DSSAT Full simulations of corn produce the highest cor-
relation coefficients, both with field trials (0.43) and NASS (0.36).
The correlation of WEAP-DSSAT Surface and NASS is poor while the
correlation of WEAP-DSSAT Full and WEAP-DSSAT Surface to field
trials is identical for corn. The unexpected identical correlations of
341
WEAP-DSSAT Full and WEAP-DSSAT Surface to field trials is a result
of field trial yield data availability coinciding with years in which
surface water provided a full allocation. For both rice and wheat,
WEAP-DSSAT Full yield correlations to NASS and field trial data are
lower, averaging approximately 0.30. For comparison, correlations
between NASS and field trial yields are about 0.50 across all crops.
WEAP-DSSAT Surface correlations for rice and wheat across both
observational datasets are much lower, and often close to zero or
even negative, underscoring the importance of groundwater
extraction for the region. We note that more accurate simulations
of yields could likely be achieved by calibrating the model using
field trial data, as well as more specific cultivars, planting dates, and
water and nutrient management.
4.2. Evapotranspiration analysis
Overall WEAP-DSSAT reasonably simulates average growing
season evapotranspiration (ET) for corn and wheat, and captures
some features of the ET seasonal cycle when compared to SIMS ET;
however, there are differences between WEAP-DSSAT and SIMS
within season. We again caveat that the seasonal cycles and mean
values for WEAP-DSSAT and SIMS are for 1981e2008 and
2011e2014, respectively, and note that California was in a persis-
tent drought during the growing seasons SIMS is available for,
which could have led to discrepancies between WEAP-DSSAT and
SIMS planting, crop phenology, and harvesting, and consequently
ET. While caution should be exercised in interpreting results, we
nonetheless feel that comparing WEAP-DSSAT and SIMS ET is useful
given the scarcity of comparisons between crop model simulated
and observed ET.
WEAP-DSSAT Full and Surface capture some features of the SIMS
ET corn seasonal cycle, but greatly underestimate ET in August
(Fig. 5). WEAP-DSSAT Full is more accurate on average (7%
342 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346

overestimates ET early in the growing season (May and June) and

underestimates ET late in the growing season (September). While
there are a variety of potential reasons for the differences in ET
between WEAP-DSSAT and SIMS, of particular note are planting
date and the timing and length of flooding, which are likely not
uniform across WEAP-DSSAT simulations and SIMS observations.
WEAP-DSSAT Full and Surface ET over wheat, which is largely
rainfed and grown in the winter, are similar to (7% underestima-
tion) SIMS ET on average (Table 3). WEAP-DSSAT overestimates ET
from November through February, and then underestimates ET in
April. This could be due to differences in the timing of crop coef-
ficient values in SIMS and crop phenology in WEAP-DSSAT, which
tends to harvest in late March or early April and thus reduces ET
late in the growing season, or a misclassification of mature wheat in
SIMS as a crop with higher ET.

4.3. Groundwater supply sensitivity assessment

Assessing exactly how irrigation patterns change under drought

Fig. 4. WEAP-DSSAT with unlimited water and surface water only simulated yields,
and NASS and field trial observed yields for: a) corn, b) rice, and c) wheat.
underestimation); however, this is partially a result of compen-
sating errors in the seasonal cycle. The SIMS ET seasonal cycle is
more consistent with WEAP-DSSAT Surface, particularly from May
through July. The underestimation of WEAP-DSSAT ET in August
could be due to discrepancies in maturity date either from the
cultivar used in WEAP-DSSAT simulations or corn taking longer to
mature due to water stress in 2011e2014. WEAP-DSSAT Full and
Surface overestimate ET for rice by 29% and 23%, respectively
(Table 3). WEAP-DSSAT accurately simulates rice ET in July and
August, especially when run with surface water only, but
in Yolo County is very challenging. While there is some fallowing
and deficit irrigation, many growers access groundwater to enable
high yields despite reduced surface water allocations. However,
California's Sustainable Groundwater Management Act (SGMA),
passed in the fall of 2014, mandates the development of Ground-
water Sustainability Plans, which are likely to place limits on
groundwater pumping by the early 2020s.
Fig. 6 shows the water allocation and yield ratios of WEAP-
DSSAT Surface, in which irrigation is supplied exclusively by sur-
face water mimicking a simplified application of the SGMA, to
WEAP-DSSAT Full. Overall, both corn and rice fare relatively well
without groundwater, as in most years irrigation demand is met by
surface water, with the exception of the middle of the 1987e1992
drought (California Department of Water Resources, 1993). In the
early part of the drought (1987 and 1988), water allocations for corn
and rice were close to full, and yields were unchanged from well-
watered. In 1989, 1991, both corn and rice received no irrigation,
resulting in failed crops. In 1990, fields were allocated roughly 35%
of irrigation demand. This resulted in an approximately 75%
reduction of corn yield relative to a full water allocation yield and a
complete loss for rice. The following year (1991) water allocations
were ~70% of irrigation demand, which reduced corn productivity
by about 25% but had no impact on rice production relative to well-
watered. The remainder of the time series to 2008 contains a
number of small reductions in water allocation. Most are less than
5% and none impact yield with the exception of corn in 2008, where
a water allocation of 86% of irrigation demand resulted in corn
yields 2% lower than well-watered. We again note that the areas
simulated are all within the surface irrigation canal service area of
YCFCWCD, so are therefore expected to be less sensitive to the

Table 2
Means of and correlations across simulated and observed yields. Both metrics are calculated over the length of record, shown in Fig. 4.

Source and Crop Mean [tons ha1] Correlation Field Trial [ ] Correlation NASS [ ]

WEAP-DSSAT Full Corn 11.8 0.43 0.36

WEAP-DSSAT Surface Corn 10.6 0.43 0.03
NASS Corn 10.7 0.51 e
Field Trial Corn 11.5 e 0.51
WEAP-DSSAT Full Rice 10.8 0.33 0.30
WEAP-DSSAT Surface Rice 9.6 0.05 0.19
NASS Rice 8.6 0.52 e
Field Trial Rice 9.6 e 0.52
WEAP-DSSAT Full Wheat 5.1 0.27 0.29
WEAP-DSSAT Surface Wheat 5.1 0.01 0.18
NASS Wheat 5.5 0.50 e
Field Trial Wheat 5.3 e 0.50
 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346 343

a) Simulated and Observed Corn Evapotranspiration a) Simulated Water Allocation and Corn Yield Ratios
300 1.8
WEAP-DSSAT Full WEAP-DSSAT Surface SIMS 1.6 Water Yield
250
1.4
200 1.2
1.0
150
0.8
100 0.6
0.4
50
0.2
0 0
4 5 6 7 8 1980 1985 1990 1995 2000 2005 2010

b) Simulated and Observed Rice Evapotranspiration b) Simulated Water Allocation and Rice Yield Ratios
]
-1

300 1.8
1.6
Evapotranspitation [mm month

250
1.4
200 1.2

Ratio [ ]
1.0
150
0.8
100 0.6
0.4
50
0.2
0 0
5 6 7 8 9 1980 1985 1990 1995 2000 2005 2010

c) Simulated and Observed Wheat Evapotranspiration c) Simulated Water Allocation and Wheat Yield Ratios
150 1.8
1.6
1.4
100 1.2
1.0
0.8
50 0.6
0.4
0.2
0 0
10 11 12 1 2 3 4 1980 1985 1990 1995 2000 2005 2010
Month Year

Fig. 5. WEAP-DSSAT simulated (1981e2008) and SIMS (2011e2014) evapotranspira- Fig. 6. Water allocation and yields ratios (WEAP-DSSAT Surface divided by WEAP-
tion growing seasonal cycles for: a) corn, b) rice, and c) wheat. Blue and red lines show DSSAT Full) for: a) corn, b) rice, and c) wheat. Water allocation ratios less than one
irrigation supply derived from unlimited water and surface water only, respectively, denote irrigation water demand exceeding surface water supply. Yield ratios less and
while crosses denote overlapping 95% confidence intervals. more than one describe reduced and enhanced yields, respectively.

Table 3 First, wheat in California is largely grown in the winter when

Mean of simulated and observed evapotranspiration. Both metrics are rainfall is more plentiful. When combined with a relatively
calculated over the length of record for the growing season, as described
in Fig. 5.
drought-tolerant crop like wheat, these fields can usually be
managed as rainfed. We note that Fig. 6 does not describe the
Source and Crop Mean [mm] amount of precipitation during the growing season (larger for
WEAP-DSSAT Full Corn 123 wheat) or amount of irrigation applied (less for wheat). In 1989,
WEAP-DSSAT Surface Corn 114 1991, wheat received no surface water for irrigation and yields were
SIMS Corn 132
reduced by 18% on average relative to well-watered. As California
WEAP-DSSAT Full Rice 171
WEAP-DSSAT Surface Rice 164 was emerging from the drought in 1992, wheat received 67% of its
SIMS Rice 133 full water allocation, and yields were 12% lower than well-watered.
WEAP-DSSAT Full Wheat 38 In addition to the intuitive differences in wheat yields between
WEAP-DSSAT Surface Wheat 38 WEAP-DSSAT Full and WEAP-DSSAT Surface, we also found unex-
SIMS Wheat 41
pected yield increases of 35% and 62% in 1998 and 2003, respec-
tively, despite a full or slight decrease in growing season water
elimination of groundwater pumping than Yolo County and the
allocation. These spikes are a result of irrigation applications in the
Central Valley overall.
previous or current water year, which caused relatively small re-
While the drought of 1987e1992 clearly impacts water alloca-
ductions in soil moisture that substantially altered planting. The
tions for wheat, overall wheat yield reductions are small when
DSSAT automatic planting algorithm requires soil moisture greater
compared to corn and rice (Fig. 6). There are several reasons for this.
344 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346
than 40% of saturation, and in 1998 and 2003 the reduction in soil
moisture delayed planting by a few weeks, shifting the entire
growing season and thus changing the temperatures and precipi-
tation experienced by the plants. While there is some precedent for
this sensitivity in the literature (Basso et al., 2015), in the Central
Valley practical considerations would likely cause growers to plant
wheat earlier regardless of soil moisture and then irrigate to ensure
the ability to double-crop their fields.
5. Conclusions
The utility of crop models that can only be run with scheduled
irrigation or unlimited automatic irrigation is diminished for pre-
dictive applications over irrigated areas. Growers will adapt irri-
gation schedules to future weather conditions, which is
incompatible with a priori scheduling of timing and amount of
irrigation, and the vast majority of irrigated croplands are irrigated
because water is scarce locally, making irrigation from an unlimited
water supply a flawed assumption. To address the deficiencies in
each of these approaches, we coupled a crop model to a calibrated
hydrologic and water resources allocation model. DSSAT adds the
ability to simulate yields and a more sophisticated representation
of crop water use to WEAP, WEAP provides water supply for irri-
gated agriculture, and an iterative approach to running WEAP-
DSSAT applies the water supply limitation from WEAP to auto-
matic irrigation in DSSAT. This functionality allows WEAP-DSSAT to
explore a variety of issues at the water-agriculture interface, such as
future yields under limited water availability, irrigation optimiza-
tion, and strategies for increasing resiliency to drought. We deploy
and test our coupled model in Yolo County, an agriculturally
important and water-limited area of California. WEAP-DSSAT is
largely able to reproduce DSSAT yields under well-watered condi-
tions. WEAP-DSSAT reasonably simulates mean yields of corn, rice,
and wheat, and demonstrates mixed skill in capturing observed
yield time series. The means and seasonal cycles of evapotranspi-
ration modeled by WEAP-DSSAT are generally consistent with SIMS
remotely sensed ET estimates, especially for corn and rice mid-
growing season; however, some significant differences exist. Pre-
dicting future water supplies for irrigation in Yolo County is
complicated by the unsustainable extraction of groundwater, which
is used to support agricultural production even in extreme drought
years. However, groundwater regulations recently enacted by the
State of California have the potential to substantially reduce current
pumping rates. We test the sensitivity of corn, rice, and wheat
production in Yolo County Flood Control and Water Conservation
District to an elimination of groundwater use for irrigation. We find
yield reductions and failures in all crops examined during the
1987e1992 drought. Corn yields are reduced proportionally with
water allocation, rice yields resemble more of a step function based
on sufficient water for flooding, and wheat yields are least sensitive
to irrigation constraints as winter wheat is grown during the wet
season so does receive some rainfall.
While we believe that WEAP-DSSAT is an important advance-
ment in the simulation of agriculture-water interactions, we note
that our treatment of groundwater is simplistic and our analyses
are limited to field crops. True groundwater use for agriculture in
Yolo County is subject to a complicated decision criteria including
crop water demand, cost of pumping, depth to water table, and
investment in infrastructure, which are outside of the scope of our
work. Yolo County is agriculturally diverse, so while corn, wheat,
and rice are important crops, we did not assess other key crops for
the region such as tomatoes, almonds, alfalfa, and walnuts, because
modules for these specialty crops are largely unavailable in DSSAT.
Future work includes expanding both the number of crops and area
simulated to enable statewide assessments for California
agriculture from seasonal to climatological timescales. Refinement
of both groundwater and how growers practically respond to
reduced irrigation allocations (e.g., fallowing, deficit irrigation,
groundwater pumping) would also create more realistic assess-
ments of climate impacts on agriculture. In addition, an evaluation
of DSSAT to simulate water-limited yields in California is needed to
assess the accuracy and constraints of the model. Finally, better
management information would provide an opportunity to
improve automatic planting, harvesting, and irrigation algorithms
in WEAP-DSSAT.
Acknowledgments
This work was supported by the NASA Applied Sciences Program
(NNH11ZDA001N-WATER). We thank Lyndon Estes and Cheryl
Porter for sharing methods for compiling and running DSSAT on
Linux platforms. We thank Forrest Melton and Alberto Guzman for
providing access to SIMS evapotranspiration data for our study
area.
References
Basso, B., Hyndman, D.W., Kendall, A.D., Grace, P.R., Robertson, G.P., 2015. Can im-
pacts of climate change and agricultural adaptation strategies Be accurately
quantified if crop models are annually Re-Initialized? PLoS ONE 10, e0127333.
http://dx.doi.org/10.1371/journal.pone.0127333.
Bondeau, A., Smith, P.C., Zaehle, S., Schaphoff, S., Lucht, W., Cramer, W., Gerten, D.,
Lotze-Campen, H., Müller, C., Reichstein, M., Smith, B., 2007. Modelling the role
of agriculture for the 20th century global terrestrial carbon balance. Glob.
Change Biol. 13, 679e706. http://dx.doi.org/10.1111/j.1365-2486.2006.01305.x.
Brumbelow, K., Georgakakos, A., 2001. An assessment of irrigation needs and crop
yield for the United States under potential climate changes. J. Geophys. Res.
Atmos. 106, 27383e27405. http://dx.doi.org/10.1029/2001JD900034.
California Department of Water Resources, 1993. California's 1987-92 Drought: a
Summary of Six Years of Drought.
Casanova, J.J., Judge, J., Jones, J.W., 2006. Calibration of the CERES-Maize model for
linkage with a microwave remote sensing model. Trans. ASAE 49, 783e792.
Deryng, D., Elliott, J., Folberth, C., Muller, C., Pugh, T.A.M., Boote, K.J., Conway, D.,
Ruane, A.C., Gerten, D., Jones, J.W., Khabarov, N., Olin, S., Schaphoff, S.,
Schmid, E., Yang, H., Rosenzweig, C., 2016. Regional disparities in the beneficial
effects of rising CO2 concentrations on crop water productivity. Nat. Clim.
Change 6, 786e790.
Do€ll, P., Hoffmann-Dobrev, H., Portmann, F.T., Siebert, S., Eicker, A., Rodell, M.,
Strassberg, G., Scanlon, B.R., 2012. Impact of water withdrawals from ground-
water and surface water on continental water storage variations. J. Geodyn.
59e60, 143e156. http://dx.doi.org/10.1016/j.jog.2011.05.001.
Elliott, J., Deryng, D., Müller, C., Frieler, K., Konzmann, M., Gerten, D., Glotter, M.,
Flo€rke, M., Wada, Y., Best, N., Eisner, S., Fekete, B.M., Folberth, C., Foster, I.,
Gosling, S.N., Haddeland, I., Khabarov, N., Ludwig, F., Masaki, Y., Olin, S.,
Rosenzweig, C., Ruane, A.C., Satoh, Y., Schmid, E., Stacke, T., Tang, Q., Wisser, D.,
2014a. Constraints and potentials of future irrigation water availability on
agricultural production under climate change. Proc. Natl. Acad. Sci. 111,
3239e3244.
Elliott, J., Kelly, D., Chryssanthacopoulos, J., Glotter, M., Jhunjhnuwala, K., Best, N.,
Wilde, M., Foster, I., 2014b. The parallel system for integrating impact models
and sectors (pSIMS). Environ. Model. Softw. 62, 509e516. http://dx.doi.org/
10.1016/j.envsoft.2014.04.008.
Espino, L., Fischer, A., Godfrey, L., Greer, C., Hill, J., Horwath, W., Klonsky, K.,
Linquist, B., Livingston, P., McKenzie, K., Mutters, C., Ruark, M., Salmon, T., Six, J.,
Whisson, D., Williams, J., van Kessel, C., 2013. Rice Production Workshop
Manual.
Famiglietti, J.S., 2014. The global groundwater crisis. Nat. Clim. Change 4, 945e948.
Frate, C., Schwanki, L., n.d. Crop Irrigation Strategies: Corn [WWW Document]. URL
http://ucmanagedrought.ucdavis.edu/Agriculture/Crop_Irrigation_Strategies/
Corn/ (Accessed 24 November 14)
Garfin, G., Franco, G., Blanco, H., Comrie, A., Gonzalez, P., Piechota, T., Smyth, R.,
Waskom, R., 2014. Ch. 20: southwest. In: Melillo, J.M., Richmond, T.C.,
Yohe, G.W. (Eds.), Climate Change Impacts in the United States. The Third Na-
tional Climate Assessment. U.S. Global Change Research Program.
Geisseler, D., Horwath, W.R., 2013. Rice production in California.
Greenwald, R., Bergin, M.H., Xu, J., Cohan, D., Hoogenboom, G., Chameides, W.L.,
2006. The influence of aerosols on crop production: a study using the CERES
crop model. Agric. Syst. 89, 390e413. http://dx.doi.org/10.1016/
j.agsy.2005.10.004.
Hart, Q.J., Brugnach, M., Temesgen, B., Rueda, C., Ustin, S.L., Frame, K., 2009. Daily
reference evapotranspiration for California using satellite imagery and weather
station measurement interpolation. Civ. Eng. Environ. Syst. 26, 19e33. http://
dx.doi.org/10.1080/10286600802003500.
 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346
Hoogenboom, G., Jones, J.W., Wilkens, P.W., Porter, C.H., Boote, K.J., Hunt, L.A.,
Singh, U., Lizaso, J.L., White, J.W., Uryasev, O., 2012. Decision Support System for
Agrotechnology Transfer (DSSAT) Ver. 4.5 [CD-ROM]. University of Hawaii,
Honolulu, Hawaii.
Huntington, J.L., Niswonger, R.G., 2012. Role of surface-water and groundwater
interactions on projected summertime streamflow in snow dominated regions:
an integrated modeling approach. Water Resour. Res. 48, W11524. http://
dx.doi.org/10.1029/2012WR012319.
Jones, J., Hoogenboom, G., Porter, C., Boote, K., Batchelor, W., Hunt, L., Wilkens, P.,
Singh, U., Gijsman, A., Ritchie, J., 2003. The DSSAT cropping system model. Eur. J.
Agron. 18, 235e265. http://dx.doi.org/10.1016/S1161-0301(02)00107-7.
Keeling, C.D., Piper, S.C., Bacastow, R.B., Wahlen, M., Whorf, T.P., Heimann, M.,
Meijer, H.A., 2001. Exchanges of Atmospheric CO2 and 13CO2 with the
Terrestrial Biosphere and Oceans from 1978 to 2000. I. Global Aspects.
Kimball, B.A., 2011. Lessons from FACE: CO2 Effects and Interactions with Water,
Nitrogen and Temperature. Handbook of Climate Change and Agroecosystems:
Impacts, Adaptation, and Mitigation. Imperial College Press, London,
pp. 87e107.
Kucharik, C.J., 2003. Evaluation of a process-based agro-ecosystem model (Agro-
IBIS) across the U.S. Corn belt: simulations of the interannual variability in
maize yield. Earth Interact. 7, 1e33. http://dx.doi.org/10.1175/1087-3562(2003)
007<0001:EOAPAM>2.0.CO;2.
Lobell, D.B., Ortiz-Monasterio, J.I., 2007. Impacts of day versus night temperatures
on spring wheat yields. Agron. J. 99, 469e477. http://dx.doi.org/10.2134/
agronj2006.0209.
Maurer, E.P., Wood, A.W., Adam, J.C., Lettenmaier, D.P., Nijssen, B., 2002. A long-term
hydrologically based dataset of land surface fluxes and States for the conter-
minous United States. J. Clim. 15, 3237e3251.
Mearns, L., Rosenzweig, C., Goldberg, R., 1996. The effect of changes in daily and
interannual climatic variability on CERES-Wheat: a sensitivity study. Clim.
Change 32, 257e292. http://dx.doi.org/10.1007/BF00142465.
Mehta, V.K., Haden, V.R., Joyce, B.A., Purkey, D.R., Jackson, L.E., 2013. Irrigation de-
mand and supply, given projections of climate and land-use change, in Yolo
County, California. Agric. Water Manag. 117, 70e82. http://dx.doi.org/10.1016/
j.agwat.2012.10.021.
Mehta, V., Rheinheimer, D., Yates, D., Purkey, D., Viers, J., Young, C., Mount, J., 2011.
Potential impacts on hydrology and hydropower production under climate
warming of the Sierra Nevada. J. Water Clim. Change 2, 29e43.
Melton, F.S., Johnson, L.F., Lund, C.P., Pierce, L.L., Michaelis, A.R., Hiatt, S.H.,
Guzman, A., Adhikari, D., Purdy, A.J., Rosevelt, C., Votava, P., Trout, T.J.,
Temesgen, B., Frame, K., Sheffner, E.J., Nemani, R.R., 2012. Satellite irrigation
management support with the terrestrial observation and prediction system: a
framework for integration of satellite and surface observations to support im-
provements in agricultural water resource management. Selected topics in
applied earth observations and remote sensing. IEEE J. 5, 1709e1721. http://
dx.doi.org/10.1109/JSTARS.2012.2214474.
Mulligan, M., Fisher, M., Sharma, B., Xu, Z.X., Ringler, C., Mahe , G., Jarvis, A.,
Ramírez, J., Clanet, J.-C., Ogilvie, A., Ahmad, M.-D., 2011. The nature and impact
of climate change in the Challenge Program on Water and Food (CPWF) basins.
Water Int. 36, 96e124. http://dx.doi.org/10.1080/02508060.2011.543408.
National Research Council, 2011. A Review of the Use of Science and Adaptive
Management in California's Draft Bay Delta Conservation Plan. National Acad-
emies Press.
Nemani, R., Hashimoto, H., Votava, P., Melton, F., Wang, W., Michaelis, A., Mutch, L.,
Milesi, C., Hiatt, S., White, M., 2009. Monitoring and forecasting ecosystem
dynamics using the terrestrial observation and prediction system (TOPS).
Remote Sens. Environ. 113, 1497e1509. http://dx.doi.org/10.1016/
j.rse.2008.06.017.
Parry, M., Rosenzweig, C., Livermore, M., 2005. Climate change, global food supply
and risk of hunger. Philosophical Trans. R. Soc. B Biol. Sci. 360, 2125e2138.
http://dx.doi.org/10.1098/rstb.2005.1751.
Pen~ a-Barragan, J.M., Ngugi, M.K., Plant, R.E., Six, J., 2011. Object-based crop identi-
fication using multiple vegetation indices, textural features and crop phenology.
Remote Sens. Environ. 115, 1301e1316. http://dx.doi.org/10.1016/
j.rse.2011.01.009.
Piontek, F., Müller, C., Pugh, T.A.M., Clark, D.B., Deryng, D., Elliott, J., Colo n
Gonza lez, F., de, J., Flo
€rke, M., Folberth, C., Franssen, W., Frieler, K., Friend, A.D.,
Gosling, S.N., Hemming, D., Khabarov, N., Kim, H., Lomas, M.R., Masaki, Y.,
Mengel, M., Morse, A., Neumann, K., Nishina, K., Ostberg, S., Pavlick, R.,
Ruane, A.C., Schewe, J., Schmid, E., Stacke, T., Tang, Q., Tessler, Z.D.,
Tompkins, A.M., Warszawski, L., Wisser, D., Schellnhuber, H.J., 2014. Multi-
sectoral climate impact hotspots in a warming world. In: Proceedings of the
National Academy of Sciences 111, 3233e3238.
Purkey, D.R., Joyce, B., Vicuna, S., Hanemann, M.W., Dale, L.L., Yates, D., Dracup, J.A.,
2008. Robust analysis of future climate change impacts on water for agriculture
and other sectors: a case study in the Sacramento Valley. Clim. Change 87,
109e122. http://dx.doi.org/10.1007/s10584-007-9375-8.
Raes, D., Steduto, P., Hsiao, T.C., Fereres, E., 2009. AquaCropdthe FAO Crop Model to
Simulate Yield Response to Water: II. Main Algorithms and Software Descrip-
tion All rights reserved. No part of this periodical may be reproduced or
transmitted in any form or by any means, electronic or mechanical, including
photocopying, recording, or any information storage and retrieval system,
without permission in writing from the publisher. Agron. J. 101, 438e447.
http://dx.doi.org/10.2134/agronj2008.0140s.
Rosenzweig, C., Elliott, J., Deryng, D., Ruane, A.C., Müller, C., Arneth, A., Boote, K.J.,
345
Folberth, C., Glotter, M., Khabarov, N., Neumann, K., Piontek, F., Pugh, T.A.M.,
Schmid, E., Stehfest, E., Yang, H., Jones, J.W., 2014. Assessing agricultural risks of
climate change in the 21st century in a global gridded crop model intercom-
parison. Proc. Natl. Acad. Sci. 111, 3268e3273. http://dx.doi.org/10.1073/
pnas.1222463110.
Rosenzweig, C., Jones, J.W., Hatfield, J.L., Antle, J.M., Ruane, A.C., Mutter, C.Z., 2015.
The Agricultural Model Intercomparison and Improvement Project: Phase I
Activities by a Global Community of Science. Handbook of Climate Change and
Agroecosystems: the Agricultural Model Intercomparison and Improvement
Project (AgMIP) Integrated Crop and Economic Assessments 3.
Rosenzweig, C., Jones, J.W., Hatfield, J.L., Ruane, A.C., Boote, K.J., Thorburn, P.,
Antle, J.M., Nelson, G.C., Porter, C., Janssen, S., Asseng, S., Basso, B., Ewert, F.,
Wallach, D., Baigorria, G., Winter, J.M., 2013. The agricultural model intercom-
parison and improvement Project (AgMIP): protocols and pilot studies. Agric.
For. Meteorology 170, 166e182. http://dx.doi.org/10.1016/
j.agrformet.2012.09.011.
Ruane, A.C., Goldberg, R., Chryssanthacopoulos, J., 2015. Climate forcing datasets for
agricultural modeling: merged products for gap-filling and historical climate
series estimation. Agric. For. Meteorology 200, 233e248. http://dx.doi.org/
10.1016/j.agrformet.2014.09.016.
Ruane, A.C., Major, D.C., Yu, W.H., Alam, M., Hussain, S.G., Khan, A.S., Hassan, A.,
Hossain, B.M.T.A., Goldberg, R., Horton, R.M., Rosenzweig, C., 2013. Multi-factor
impact analysis of agricultural production in Bangladesh with climate change.
Glob. Environ. Change 23, 338e350. http://dx.doi.org/10.1016/
j.gloenvcha.2012.09.001.
Sandoval-Solis, S., McKinney, D., Loucks, D., 2010. Sustainability index for water
resources planning and management. J. Water Resour. Plann. Manage 137,
381e390. http://dx.doi.org/10.1061/(ASCE)WR.1943-5452.0000134.
Sau, F., Boote, K.J., Bostick, W.M., Jones, J.W., Mínguez, M.I., 2004. Testing and
improving evapotranspiration and soil water balance of the DSSAT crop models.
Agron. J. 96, 1243e1257.
Schaible, G.D., Aillery, M.P., 2012. Water Conservation in Irrigated Agriculture:
Trends and Challenges in the Face of Emerging Demands. USDA Economic
Research Service.
Scow, K., 2009. Cover Crop Biomass Yields. Davis, CA: Russell Ranch Electronic Data
Archives: CC-001 (Accessed 23 April 15). http://russellranch.ucdavis.edu/data.
Solley, W.B., Pierce, R.R., Perlman, H.A., 1998. Estimated Use of Water in the United
States in 1995 (Report No. 1200), Circular.
Taylor, R.G., Scanlon, B., Doll, P., Rodell, M., van Beek, R., Wada, Y., Longuevergne, L.,
Leblanc, M., Famiglietti, J.S., Edmunds, M., Konikow, L., Green, T.R., Chen, J.,
Taniguchi, M., Bierkens, M.F.P., MacDonald, A., Fan, Y., Maxwell, R.M., Yechieli, Y.,
Gurdak, J.J., Allen, D.M., Shamsudduha, M., Hiscock, K., Yeh, P.J.-F., Holman, I.,
Treidel, H., 2013. Ground water and climate change. Nat. Clim. Change 3,
322e329. http://dx.doi.org/10.1038/nclimate1744.
Tubiello, F.N., Soussana, J.-F., Howden, S.M., 2007. Crop and pasture response to
climate change. Proc. Natl. Acad. Sci. 104, 19686e19690. http://dx.doi.org/
10.1073/pnas.0701728104.
University of California Cooperative Extension, 2015. Delta Crops Resource Man-
agement [WWW Document] (Accessed 13 April 15). http://ucanr.edu/sites/
deltacrops/Rice/.
University of California Division of Agriculture and Natural Resources, 2006. Small
Grain Production Manual (No. 8208).
USDA NASS, 2014. Census of Agriculture, p. 2012.
USDA NASS, 2010. Field Crops Usual Planting and Harvesting Dates (No. 628).
Varela-Ortega, C., Blanco-Gutie rrez, I., Swartz, C.H., Downing, T.E., 2011. Balancing
groundwater conservation and rural livelihoods under water and climate un-
certainties: an integrated hydro-economic modeling framework. Glob. Environ.
Change 21, 604e619. http://dx.doi.org/10.1016/j.gloenvcha.2010.12.001.
Wada, Y., Wisser, D., Eisner, S., Flo €rke, M., Gerten, D., Haddeland, I., Hanasaki, N.,
Masaki, Y., Portmann, F.T., Stacke, T., Tessler, Z., Schewe, J., 2013. Multimodel
projections and uncertainties of irrigation water demand under climate change.
Geophys. Res. Lett. 40, 4626e4632. http://dx.doi.org/10.1002/grl.50686.
Walthall, C.L., Hatfield, J., Backlund, P., Lengnick, L., Marshall, E., Walsh, M.,
Adkins, S., Aillery, M., Ainsworth, E.A., Ammann, C., Anderson, C.J., Bartomeus, I.,
Baumgard, L.H., Booker, F., Bradley, B., Blumenthal, D.M., Bunce, J., Burkey, K.,
Dabney, S.M., Delgado, J.A., Dukes, J., Funk, A., Garrett, K., Glenn, M.,
Grantz, D.A., Goodrich, D., Hu, S., Izaurralde, R.C., Jones, R.A.C., Kim, S.-H.,
Leaky, A.D.B., Lewers, K., Mader, T.L., McClung, A., Morgan, J., Muth, D.J.,
Nearing, M., Oosterhuis, D.M., Ort, D., Parmesan, C., Pettigrew, W.T., Polley, W.,
Rader, R., Rice, C., Rivington, M., Rosskopf, E., Salas, W.A., Sollenberger, L.E.,
Srygley, R., Sto€ckle, C., Takle, E.S., Timlin, D., White, J.W., Winfree, R., Wright-
Morton, L., Ziska, L.H., 2013. Climate Change and Agriculture in the United
States: Effects and Adaptation. (Technical Bulletin No. 1935). USDA, Washington,
DC.
White, J.W., Hoogenboom, G., Kimball, B.A., Wall, G.W., 2011. Methodologies for
simulating impacts of climate change on crop production. Field Crops Res. 124,
357e368. http://dx.doi.org/10.1016/j.fcr.2011.07.001.
Winter, J.M., Lopez, J.R., Ruane, A.C., Young, C.A., Scanlon, B.R., Rosenzweig, C., 2017.
Representing water scarcity in future agricultural assessments. Anthropocene
18, 15e26. http://dx.doi.org/10.1016/j.ancene.2017.05.002.
Yates, D., Galbraith, H., Purkey, D., Huber-Lee, A., Sieber, J., West, J., Herrod-Julius, S.,
Joyce, B., 2008. Climate warming, water storage, and Chinook salmon in Cal-
ifornia's Sacramento Valley. Clim. Change 91, 335e350. http://dx.doi.org/
10.1007/s10584-008-9427-8.
Yates, D., Purkey, D., Sieber, J., Huber-Lee, A., Galbraith, H., 2005a. WEAP21da
346 J.M. Winter et al. / Environmental Modelling & Software 96 (2017) 335e346

demand-, priority-, and preference-driven water planning model: Part 2: aiding
freshwater ecosystem service evaluation. Water Int. 30, 501e512. http://
dx.doi.org/10.1080/02508060508691894.
Yates, D., Purkey, D., Sieber, J., Huber-Lee, A., Galbraith, H., West, J., Herrod-Julius, S.,
Young, C., Joyce, B., Rayej, M., 2009. Climate driven water resources model of the
Sacramento basin. Calif. J. Water Resour. Plann. Manage 135, 303e313. http://
dx.doi.org/10.1061/(ASCE)0733-9496(2009)135:5303.
Yates, D., Sieber, J., Purkey, D., Huber-Lee, A., 2005b. WEAP21da demand-, priority-,
and preference-driven water planning model: Part 1: model characteristics.
Water Int. 30, 487e500. http://dx.doi.org/10.1080/02508060508691893.