Impacts of Near-Term Climate Change on Irrigation Demands and Crop Yields in the Columbia River Basin
Abstract
Adaptation to a changing climate is critical to address future global food and water security challenges. While these challenges are global, successful adaptation strategies are often generated at regional scales; therefore, regional-scale studies are critical to inform adaptation decision making. While climate change affects both water supply and demand, water demand is relatively understudied, especially at regional scales. The goal of this work is to address this gap, and characterize the direct impacts of near-term (for the 2030s) climate change and elevated CO2 levels on regional-scale crop yields and irrigation demands for the Columbia River basin (CRB). This question is addressed through a coupled crop-hydrology model that accounts for site-specific and crop-specific characteristics that control regional-scale response to climate change. The overall near-term outlook for agricultural production in the CRB is largely positive, with yield increases for most crops and small overall increases in irrigation demand. However, there are crop-specific and location-specific negative impacts as well, and the aggregate regional response of irrigation demands to climate change is highly sensitive to the spatial crop mix. Low-value pasture/hay varieties of crops—typically not considered in climate change assessments—play a significant role in determining the regional response of irrigation demands to climate change, and thus cannot be overlooked. While, the overall near-term outlook for agriculture in the region is largely positive, there may be potential for a negative outlook further into the future, and it is important to consider this in long-term planning.
Key Points
- Near-term outlook for irrigated agriculture in the Columbia River basin is largely positive; however, there are negative impacts as well
- The regional responses of irrigation demand to climate change is sensitive to crop mix; low-value crops play a large disproportionate role
- While the overall near-term outlook is largely positive, this need not necessarily extend further out into the future
1 Introduction
Global growth in population and affluence are increasing the demand for food (Godfray et al., 2010); simultaneously, climate change is reducing irrigation water availability in regions where streamflow is dominated by snowmelt runoff (Barnett et al., 2005). Because irrigated agriculture is critical in meeting the world's food supply, these factors will result in increasing food-security challenges (Foley et al., 2011; Godfray et al., 2010; Tilman et al., 2011). Coadaptation of irrigated agricultural and water management systems is key to addressing these challenges in a sustainable manner (Grafton et al., 2017; Lobell et al., 2008), especially as irrigation is often in competition with other out-of-stream as well as in-stream water uses.
Irrigation for agriculture accounts for about 70% of global freshwater withdrawals (Döll, 2009) and constitutes the largest share of consumptive water use (Falkenmark & Lannerstad, 2004). Irrigated cropland in arid and semiarid areas has historically played a central role in increasing agricultural productivity and reliability, and meeting global food demands (Rosegrant et al., 2009; Sauer et al., 2010). At the same time, irrigation can also be associated with negative effects on the water resource system as a whole, through depleting groundwater and surface water resources, diverting water away from instream flow requirements that are necessary for ecological health, and creating water quality and salinity issues, etc. (De Fraiture & Wichelns, 2010). Irrigated areas are expected to expand in the future (Neumann et al., 2011) and consequently increase the competition for limited resources. Thus, a strong argument can be made that water resources, agricultural productivity, irrigation demands, and the relationships between them will to a large extent dictate how we address food-security challenges in the future.
Climate is an important factor that affects agricultural productivity (Mueller et al., 2012; Van Ittersum et al., 2013) and irrigation demands (Fischer et al., 2007). Climate change affects agricultural productivity directly by impacting potential crop yields (i.e., those achievable under stress-free conditions), and indirectly through changes in water availability, which is function of water supply and how this supply is managed for competing uses. (For example, under the doctrine of prior appropriation in the western U.S., water law will determine which user(s) receives the water during water shortages.) Climate change also impacts irrigation water demands often in substantial ways (Fischer et al., 2007; Tubiello et al., 2007), although few studies have attempted to quantify these changes (Fischer et al., 2007), especially at regional scales.
Although food security is a global issue, impacts of climate change on water resources and agricultural productivity are frequently site or regionally specific (McGrath & Lobell, 2013a; Nelson et al., 2009). However, models built for global studies intentionally prioritize incorporation of processes that are either important or can be parameterized at global scales (Lawrence & Fisher, 2013) and may be missing other processes that become important at regional scales. Furthermore, successful adaptation strategies are often generated at regional/local scales, and a regional-scale model may provide a more appropriate framework to inform decision making. Therefore, given this and the importance of understanding both agricultural productivity and irrigation water demand together, there is a knowledge gap that limits sustainable adaptation of irrigated agricultural systems to climate change.
The goal of this work is to characterize the direct impacts of near-term (2030s) climate change and increasing CO2 levels on irrigation demands and irrigated crop yields on a regional scale, thus providing the water demand context for regional agricultural and water resource decision making in the Columbia River basin (CRB) of the Pacific Northwest U.S. We define near-term as a 20 year outlook into the future as this is the time frame for which infrastructure and other water resources planning decisions are made by state and local agencies. We incorporate a dynamic crop growth component within a macroscale hydrology model and apply the coupled model to analyze the direct impacts of changes in climate and elevated CO2 levels on irrigation water demands and irrigated crop yields in the CRB. In addition to providing a regional-scale water demand context for decision making, this work also lays the foundation for future efforts that can look at water supply and demand in conjunction, and include other components such as competing uses, and water management (such as water rights curtailment) and human decision making in an internally consistent manner.
2 Background
2.1 Site Description
The Columbia River in the Pacific Northwest (PNW) is the fourth largest river in the U.S. The watershed drains approximately 688,000 km2, encompassing parts of Idaho, Montana, Nevada, Oregon, Utah, Washington, and Wyoming in the U.S. About a third of the basin is in southern British Columbia, Canada. The CRB is heavily managed with a system of dams and reservoirs. However, storage in the CRB is only about 30% of the annual runoff in an average year, while relative storage capacity of other large river systems can by much higher, e.g., 2–3 times the annual runoff in the Mississippi River system (BPA, 2002). Limited storage in the CRB implies that water managers and other decision makers have less control in managing water resources, making the system more sensitive to warming-related changes to the hydrological cycle.
Water resources of the CRB are heavily managed to satisfy multiple (often competing) objectives including hydropower production, flood control, agricultural withdrawals, instream flow requirements for fish, and recreational needs (Hamlet & Lettenmaier, 1999; Hamlet et al., 2010). There is conflict between in-stream water demands such as for hydropower and fish and wildlife (Leonard et al., 2015). Eight fish species protected under the U.S. Endangered Species Act have habitat in the Columbia River (Cosens & Williams, 2012). Hydropower produced by dams along the Columbia River and its tributaries provide for over 70% the region's energy demands, and account for 40% of the U.S. hydropower production (Hamlet et al., 2002). In addition, these in-streams demands compete with out-of-stream uses such as irrigation. Irrigated agriculture has a substantial impact on the CRB's water resources, and agricultural withdrawal is the largest consumptive user of Columbia River water with about 14,200 km2 of irrigated area in the CRB.
Agriculture is a vital part of the rural PNW economy, with an annual value over $9 billion in Washington State alone (Brady & Taylor, 2011). A large component of CRB agricultural production is supported by irrigation (Figure 1), and the basin supports very diverse agricultural production, including tree fruit, potatoes, hops, vegetables, cereal grains, wine grapes, and hay. Agriculture and related services account for more than 10% of the basin's employment (NRC, 2004). The diversity in crop mix, importance of agriculture to the regional economy, heavy human management of the river system, and limited storage that is sensitive to warming make the CRB an ideal test bed for integrated modeling of water resources and agricultural systems in the context of climate change.

The outline of the Columbia River basin and irrigated crop extent in the U.S. part of the Columbia River basin are shown. This is based on the 2008 Agricultural Land Use Layer from the Washington State Depart of Agriculture, and the 2009 Cropland Data Layer from U.S. Department of Agriculture. (Note that agricultural production in the Canadian part of the basin is not considered.) The irrigated extent is shown separately for three broad classes of crops: (a) perennials such as pasture/hay varieties of crops that are harvested as multiple cuttings, (b) other perennials such as tree fruit, and (c) annual crops such as potato, grain crops, and vegetables. The colors correspond to the percentage area of each grid cell that is covered by irrigated crops.
2.2 Climate in the CRB
The CRB receives the majority of its precipitation in the winter months between October and March; summers (the season of peak crop irrigation requirements) are relatively dry. Average annual precipitation is strongly variable across the region and ranges from less than 200 mm in central Washington to 500–760 mm near the mountain foothills and 1,000 mm or more in some high elevation areas. Surface water flows in the CRB are dominated by the temperature-sensitive cycle of snow accumulation and melt (Hamlet & Lettenmaier, 1999; Leung & Ghan, 1998); i.e., the snowpack stores winter precipitation for later use during the summer season of peak irrigation demand, thereby making summer water availability vulnerable to warming (Barnett et al., 2005; Mote et al., 2005).
The climate of the Pacific Northwest has changed over the last century. Average temperatures across the PNW have risen about 0.8°C, with some areas experiencing increases up to 2.2°C (Abatzoglou et al., 2014; Mote, 2003). Climate models suggest that precipitation and temperature changes will continue to intensify over the next century. Temperature changes are projected to be in the range of 0.6–2.8°C over the next 50 years (Mote & Salathe, 2010). Precipitation changes are much less certain than temperature projections and are unlikely to be distinguishable from natural variability until late this century (Mote & Salathe, 2010). However, changes in precipitation seasonality are anticipated; these include increasing precipitation during the cool season and decreasing precipitation during the warm season (Mote & Salathe, 2010). These projections are based on CMIP3 results. The latest CMIP5 projections for the PNW are similar to CMIP3 projections in reproducing the historical climate of the PNW (Rupp et al., 2013). For future climate simulations, CMIP3 and CMIP5 projections show similar trends in annual and seasonal changes of both temperature and precipitation (Rupp et al., 2013). The main difference is that CMIP5 projections are warmer than CMIP3 projections due to increased GHG forcing considerations. Any differences in precipitation are masked by the large interannual variability in precipitation.
2.3 Effect of Climate Change and Elevated CO2 Concentrations on Crop Yields
Higher growing season temperatures are generally expected to negatively affect crop production (Battisti & Naylor, 2009; Lobell & Asner, 2003; Lobell & Field, 2011). This could be due to accelerated time to maturity or increased heat stress (Alexandrov et al., 2002; Easterling et al., 2007; Travasso et al., 2006). The impacts are crop dependent with each crop having a unique response and some crops being more sensitive to temperature than others (Easterling et al., 2007). In addition, every crop has an optimal range of temperatures for growth which varies by growth stage, and impacts of higher temperatures on yields depend on the prevailing baseline temperatures (Hatfield et al., 2011). Therefore, although the general expected impact of increasing temperatures is negative, this is not true for all crops; some crops currently growing in suboptimal temperatures may experience a yield increase due to increasing temperatures, as long these benefits outweigh yield reductions due to shorter time to maturity. Therefore, the response of crop yields to temperature is nonlinear, nonmonotonic and varies regionally (Chang, 2002; McGrath & Lobell, 2013a; Schlenker & Roberts, 2008).
Elevated CO2 concentration levels generally have a positive impact on crop yields, especially crops that follow the C3 photosynthetic pathway (Hatfield et al., 2011). The increases in yields are facilitated through increased photosynthesis (Parry et al., 2004). Crop response to changes in CO2 concentrations are complex and depend upon the crop species as well as interactions with temperature, soil moisture, nutrient management, and acclimation to these factors (Hatfield et al., 2011; Long, 1991; Wolfe et al., 1998). Nutrient availability also moderates CO2 fertilization effects on crop yields with potentially insignificant impacts under low nitrogen availability (Tubiello et al., 2007). While not considered in our analysis, another emerging area of concern is the negative effect of CO2 fertilization on food nutrition value (Kirschbaum, 2011; McGrath & Lobell, 2013b). A meta-analysis by Taub et al. (2008) noted 10%–15% declines in protein content for CO2 concentrations between 540 and 960 ppm in major food crops, including wheat, barley, soybean, and potato.
Within the CRB, a regional-scale study that considers impacts of climate change on crop yields for a full range of crops in a spatially explicit manner is lacking. Stöckle et al. (2010) used a point-scale crop growth model (CropSyst) to study the impacts of climate change on crop yields in a few locations in eastern Washington. The study showed decreases in potato yields and increases in apple yields for the 2040s. Vano et al. (2010) looked into impacts of climate change on tree fruit production for a point-scale in the Yakima River basin, which is a subbasin in the CRB, and noted negative effects. Tubiello et al. (2002) studied climate change impacts on wheat and potato yields for several sites in U.S., two of which are in the CRB. Their work indicated a positive effect of climate change on winter wheat yields and a negative effect on potato yields. Although, not within the CRB, Lobell et al. (2006) performed an empirical study focused on perennial crops in nearby California and determined that climate change is expected to have a negative effect on perennial crop yields in the region. Lee et al. (2011) project increases in hay yields, and decreases in annual field crop yields for the Central Valley region in California.
2.4 Effect of Climate Change and Elevated CO2 Concentrations on Irrigation Demands
Warming affects irrigation demands both through increases in evapotranspiration needs and by changing the crop calendar through the potential for an earlier planting and harvest, and decreased time between growing season start and harvest (Hatfield et al., 2011; Rosenzweig & Hillel, 1995). Changes in the crop calendar affects irrigation demand because both potential evapotranspiration and precipitation vary seasonally (Wada et al., 2013). Elevated CO2 levels improve plant water-use efficiency at a leaf level through partial stomatal closure (Rosenzweig & Hillel, 1995). However, elevated CO2 levels also increase primary production and can lead to overall increases in transpiration in semiarid regions (Leipprand & Gerten, 2006). These effects can also be moderated by limiting factors such as nutrient limitations (Konzmann et al., 2013; Tubiello et al., 2007).
Based on global studies (Doll, 2002; Fischer et al., 2007; Wada et al., 2013), the general climate change impact is a projected increase in irrigation demand. Konzmann et al. (2013) documents an exception, with projected decreases in demand. Although the importance of crop-specific differences in response is known (Easterling et al., 2007), these differences are not accounted for in the global studies, and crops are lumped into coarse groups, e.g., rice versus nonrice by Doll (2002), and a larger 12-group classification by Konzmann et al. (2013). Limited studies (Konzmann et al., 2013; Wada et al., 2013) have accounted for increased water-use efficiencies under elevated CO2 levels.
Select global studies provide maps of changes from which projected changes for the CRB can be visually approximated. Doll (2002) projects 5–30% increases in irrigation demands in the 2020s for the CRB. Wada et al. (2013) take a multimodel ensemble average approach and project increases of 5–10% in the CRB by the end of the century. Konzmann et al. (2013) project 20–40% decreases in irrigation demands in the CRB by the end of the century. Near-term impacts are not mapped by Wada et al. (2013) nor Konzmann et al. (2013); however, they emphasize the effect of shifts in crop growth stages on irrigation demands, and shifts in the seasonality of peak demand in different parts of the world. Regional-scale studies that consider the impacts of climate change on irrigation demands for a full range of crops in a spatially explicit manner is generally lacking. One study specific to a tributary of the CRB is described by Vano et al. (2010), who use a point-scale simulation to project decreases in irrigation demand for tree fruit in the Yakima River basin. Our study uses similar tools to those used by Vano et al. (2010) with the exception that the hydrology model is fully coupled to the cropping systems model and run in a spatially explicit manner over the entire region.
3 Methods
Hamlet et al. (2013) implemented, calibrated, and evaluated the large-scale Variable Infiltration Capacity (VIC) hydrologic model over the U.S. Pacific Northwest at a 1/16° grid cell resolution by building upon the implementation of Elsner et al. (2010). They applied the VIC model to assess the 21st century climate change impacts on the hydrology and water resources of the region, providing projections of future water supply. Building on this framework, we add a dynamic crop growth component to allow consideration of irrigation water demand (in addition to water supply), and consider impacts on agricultural production in a spatially and crop-specific manner.
3.1 Model Components in the Framework
The VIC 4.0.7 (Cherkauer et al., 2003; Liang et al., 1994) model is a physically based, spatially explicit, large-scale hydrology model that solves water and energy balance equations to generate runoff and baseflow at the grid cell scale. The VIC model captures subgrid heterogeneity in land cover; a percentage of each grid cell is assigned to each land cover class and the full water and energy balance routines are run for each class. A separate routing model (Lohmann et al., 1998) performs streamflow routing as an off-line process. The VIC model has been widely used for basins across North America (Christensen et al., 2004; Hayhoe et al., 2007; Maurer, 2007; VanRheenen et al., 2004). More specifically the VIC model has been implemented to assess climate change impacts over the CRB (Elsner et al., 2010; Hamlet et al., 2013; Hamlet & Lettenmaier, 1999; Payne et al., 2004) for different future climate scenarios. We have used the Hamlet et al. (2013) implementation of the VIC model (version 4.0.7) in this integrated framework. Outside croplands, the land cover characterization is static; i.e., natural vegetation growth and associated dynamics are not modeled.
CropSyst is a field-scale, multiyear, multicrop model developed to serve as an analytical tool to study the effects of climate, soils, and management on cropping systems productivity, nutrient cycling and fate, and the environment. Crop growth stages are based on crop-specific thermal time accumulation (growing degree days), where each crop has a base and cutoff temperature below and above which thermal accumulation does not take place. Management options include crop rotation, cultivar selection, irrigation, nitrogen fertilization, tillage operations, and residue management. CropSyst has been evaluated and used in the PNW (e.g., Pannkuk et al., 1998; Peralta & Stockle, 2002; Stöckle et al., 2010) and in many other locations worldwide (e.g., Badini et al., 2007; Benli et al., 2007; Sadras & Angus, 2006). For this integrated framework, a simplified version of CropSyst that focuses on water use and crop productivity was extracted for coupling with the VIC hydrology model to create the VIC-CropSyst-v1.2 (“VIC-CropSyst”) model.
3.2 Framework Integration
The integration between the VIC model and CropSyst is shown in Figure 2. The VIC model invokes CropSyst when it encounters a cropland land cover class within a grid cell (CropSyst may be invoked several times for a single VIC model grid cell as multiple crop types can exist in a single cell). The VIC model provides to CropSyst information related to weather, crop type and crop management on a daily basis. CropSyst grows the crop, keeps track of water stress, and passes back to the VIC model irrigation demand, transpiration and crop evaporation, also on a daily basis. The irrigation demand request from CropSyst is adjusted (increased) to account for efficiencies of specific irrigation methods (this adjusted demand is the “top of the crop” irrigation demand) before application to the soil. The VIC model subsurface drainage process then accounts for return flows from irrigation. Due to the differences in the way the VIC and CropSyst characterize the subsurface profile, the models maintain separate subsurface hydrologic processes in the coupled model. Feedback of transpiration and crop evaporation amounts from CropSyst to the VIC model are used to adjust the soil moisture levels in the VIC model and maintain consistency in the water balance between models. In summary, the coupled VIC-CropSyst framework provides natural (nonmanaged) runoff and baseflow, “top of the crop” irrigation demands and crop yields.

VIC-CropSyst v1.2 integration. Information exchanged between the VIC and CropSyst models are listed within the arrows connecting the model. Key inputs and/or variables simulated are also listed next to the corresponding models.
3.3 Input Data
Table 1 summarizes the sources of the inputs for models and are described in more detail in each of the sections below.
Parameter | Description |
---|---|
Dsmax | Maximum baseflow from the deepest soil layer (mm/d) |
Ds | Fraction of Dsmax at which nonlinear baseflow begins |
Ws | Fraction of maximum soil moisture (in the deepest soil layer) where nonlinear baseflow occurs |
b-inf | The parameter that defines the shape of the variable infiltration curve |
d | Soil depth of each layer (m) |
3.3.1 Land Cover Parameterization
Land cover and its parameterization are important drivers of physical processes related to hydrology and crop growth, including evapotranspiration, interception, infiltration, and runoff. We use the Elsner et al. (2010) land cover parameterization (which, in the absence of CropSyst coupling, used a static generic crop group parameterized as corn) and extend it to include an explicit cropland cover parameterization. Two different data sources were used: the 2008 Washington State Department of Agriculture agricultural land use layer (WSDA ALL) for the Washington State part of the study area, and the 2009 United State Department of Agriculture cropland data layer (USDA CDL) for parts of the study area in the U.S. outside Washington State. For the Canadian portion of the study area, the crop model was not invoked and we retained the Elsner et al. (2010) land cover characterization. USDA CDL has only crop distribution information, whereas WSDA ALL is more detailed with information on crop distribution as well as irrigation extent and method. In the absence of a data set to prescribe irrigation extent outside Washington State, we made the following assumptions related to irrigation extent and efficiencies. A list of high-value crops such as tree fruit, potatoes, and vegetables that are always irrigated within the Washington part of the CRB are assumed to be irrigated outside Washington State as well, while other crop types are parameterized as nonirrigated. If a crop is assumed to be irrigated outside Washington State, the dominant irrigation type for that crop in Washington was assigned outside Washington. Irrigation efficiencies for each irrigation method were assigned based on National Agricultural Statistics Service (NASS) and irrigation guide recommendations. Also, the distribution of pasture grass crop types is missing in the WSDA ALL data set, and was obtained from USDA CDL even for the Washington State portion of the study area. This crop type has a large spatial variability in irrigation extent, and we made the assumption to irrigate pasture if more than half of the other crop acreage in a grid cell is irrigated. The range of crops groups captured is listed in Appendix A, although results are shown for representative crops rather than for this comprehensive list.
3.3.2 Gridded Meteorological Data
We use the historical gridded data sets at 1/16° spatial resolution created by Elsner et al. (2010). The gridded data sets for temperature and precipitation are at a daily time step and based on methods described by Maurer et al. (2002) and Hamlet and Lettenmaier (2005) and include corrections for important systematic biases, such as the influence of orography when gridding temperature and precipitation observations. Wind speed values are based on the National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) reanalysis products (Kalnay et al., 1996). Other variables such as solar radiation and vapor pressure deficit were derived from the daily temperature range or minimum temperature as described by Maurer et al. (2002).
For future climate simulations, we use five different CMIP3 GCM/emission scenario combinations based on discussions by Mote and Salathe (2010), such that they captured the entire spread of temperature and precipitation change projections for the area for the 2030s. This includes four GCM/emission scenario combinations that represent the four possible combinations of (both low and high) extremes in projected P and T changes (PCM1 B1, IPSL A1B, CCSM3 B1, and CGCM3.1 t47 B1) as well as a GCM/emission scenario combination that represents more moderate projected changes in P and T (HADCM B1). The GCM results are downscaled to the 1/16° resolution using the hybrid-delta change method (see Hamlet et al., 2013; Tohver et al., 2014 for details) to create gridded data products for the 2030s for the five future climate scenarios. As noted earlier, the latest CMIP5 climate projections show a larger degree of warming in the area, but the general climate change signals are still consistent between CMIP3 and CMIP5. Therefore, key conclusions of climate change impacts should be consistent between CMIP3 and CMIP5 meteorological inputs.
3.3.3 Soil Characteristics
For the hydrology component, we use the gridded 1/16° resolution soil file developed by Elsner et al. (2010) which is based on Maurer et al. (2002) which, in turn, is based on gridded data sets developed as part of the Land Data Assimilation System (LDAS; Mitchell et al., 2004) project. Three soil layers are considered. The dynamic crop growth component uses its own soil layering system with a higher number of soil layers based on soil depth. The parameters are based on the STATSGO2 (NRCS, 2006) soil survey database provided by the USDA NRCS.
3.3.4 CO2 Concentrations
The baseline historical simulations use atmospheric CO2 concentrations of 370 ppm, while the 2030s simulations use concentrations of 437 and 461 ppm for B1 and A1B scenarios, respectively.
3.3.5 Crop Parameters
CropSyst crop parameters describe phenology, canopy growth, transpiration, biomass production, and yield. While over 90 crop types are simulated, parameters are provided for a set of 26 crops listed in Appendix B. These parameter values are taken from existing model applications in the region and elsewhere in the world. The other crops are described by approximation to this set of 26 crops (i.e., crop types that behave similarly are given the same parameter values). Biomass production and yield information for crops not part of Appendix B (these typically have small production acreage) are not readily available. For these crops, the primary parameterization emphasis is on canopy cover and water use by approximation to crops in Appendix B and thus their yield outputs should not be considered definitive.
3.4 Calibration and Sensitivity Analysis
We retain the Hamlet et al. (2013) calibrated (against naturalized stream flows) parameters, which were based on Matheussen et al. (2000), for this study. Calibrated parameters include soil related parameters such as middle and bottom soil layer depths, the infiltration curve shape parameter, and the baseflow shape parameters (see Table 1 for descriptions). Given that streamflow calibration is performed against naturalized stream flows, and not observed gage flows that account for human influence such as irrigation, it is reasonable to retain the Hamlet et al. (2013) calibrated parameters. However, we do investigate the sensitivity of irrigation demands and crop yield to the Hamlet et al. (2013) calibrated soil parameters. The list of parameters tested for are listed in Table 1. There is minimal sensitivity of irrigation demands and yields to these parameters; the changes in irrigation demands and yields where within 1% for a ±10% change in the calibrated soil parameters.
Crop yields and irrigation demands are sensitive to solar (shortwave) radiation and vapor pressure deficit. Given that these variables are estimated (Maurer et al., 2002; Thornton & Running, 1999) using often-used algorithms and are not input directly as observations, we perform a sensitivity analysis of irrigation demands and crop yields to solar radiation and vapor pressure deficit. We perform this for select grid cells that have a large irrigated crop intensity (more 15 irrigated crops grown in the grid cell). A ±10% range is applied to solar radiation and vapor pressure deficit. While the magnitudes of the crop irrigation demands and yields are somewhat sensitive to vapor pressure deficit and shortwave solar radiation (Table 2), differences in the change in irrigation demands and crop yields between historical and future simulations are negligible (Table 3). Therefore, while vapor pressure deficit and solar radiation inputs are a source of uncertainty to the magnitudes of crop irrigation demands and yields, there is less uncertainty with respect to climate change impacts, which is the focus of this study.
Variable | Range (%) | Historical irrigation demand sensitivity (%) | Historical crop yield sensitivity (%) |
---|---|---|---|
Vapor pressure deficit | +10 | −2.9 to −1 | 0 to 0.4 |
Vapor pressure deficit | −10 | 1 to 2.5 | −0.4 to 0 |
Solar radiation (short wave) | +10 | 3.8 to 5 | 1.2 to 4.6 |
Solar radiation (short wave) | −10 | −4.5 to −3.5 | −4.6 to −2.2 |
- Note. The numbers correspond to select grid cells that have a high crop intensity (more than 15 irrigated crops grown in the grid cell). The sensitivity is calculated as the % difference between irrigation demands (or yields) between original baseline value for vapor pressure deficit (or solar radiation), and the ±10% variable value range. For example, historical irrigation demand sensitivity for +10% vapor pressure deficit is calculated as (“historical irrigation demand under +10% vapor pressure deficit” – “historical irrigation demand under default vapor pressure deficit”)/“historical irrigation demand under default vapor pressure deficit” × 100. The sensitivity ranges shown are across multiple crops.
Variable | Range (%) | Climate change effect on irrigation demand (%) | Climate change effect on crop yields (%) |
---|---|---|---|
Vapor pressure deficit | +10 | 1.65 | 3.35 |
Vapor pressure deficit | −10 | 1.56 | 3.30 |
Solar radiation (short wave) | +10 | 1.50 | 3.32 |
Solar radiation (short wave) | −10 | 1.50 | 3.34 |
- Note. The numbers correspond to select grid cells that have a high crop intensity (more than 15 irrigated crops grown in the grid cell). The climate change effect is calculated as the % difference between irrigation demands (or yields) between future (hadcm B1 scenario) and historical simulations with vapor pressure deficit (or solar radiation) at ±10% variable values. For example, the climate change effect on irrigation demand for +10% vapor pressure deficit is calculated as (“hadcm B1 irrigation demand under +10% vapor pressure deficit” – “historical irrigation demand under +10% vapor pressure deficit”)/“historical irrigation demand under + 10% vapor pressure deficit” × 100. The effects shown are averaged across multiple crops.
Regional irrigation demands and yields are sensitive to crop phenological parameters, which can vary by crop. To account for site-specific and local variation in biomass production, the VIC-CropSyst-simulated crop yields are calibrated against observed county level National Agricultural Statistical Service (NASS) crop yield statistics using a calibration period of 1996–2006. The comparison between simulated and observed yields is used to derive an adjustment factor for canopy growth (which ultimately affects biomass production) for each crop/county combination and applied to the expected maximum canopy cover, and green and total canopy cover at maturity. Crop yields are evaluated by aggregating simulated crop yields to a county/crop level and comparing with 25 years (1970–1995) of survey-based NASS yield statistics. The NASS yields are detrended using a linear fit and adjusted to dry yield values for the comparison. NASS yields are adjusted to dry yields based on moisture rate assumptions (e.g., 15% for corn and 80% for potatoes as listed in Table C1 of Appendix C). These evaluation results are in section 4.1.
3.5 Framework Application
The modeling framework described above is run at a daily time step for a 30 year time frame corresponding to historical climate, and a 30 year time frame corresponding to future climate in the 2030s (centered around 2035) to understand the direct impacts of climate change and elevated CO2 levels on irrigation demand and crop yields in the CRB.
4 Results
4.1 Evaluation
Streamflow. Hamlet et al. (2013) evaluated streamflow at about 80 locations in the CRB; 50 out of the 80 sites have Nash Sutcliffe Efficiency (NSE) scores of approximately 0.7 (therefore a good to excellent fit). The remaining locations had marginal (0.3–0.7) or negative NSE values (negatives values indicate that a user would benefit more by using mean observed streamflow than the simulated time series), which could be related to precipitation errors or errors in baseflow due to groundwater contributions that are not simulated by the model.
Crop yields. The details of crop yield evaluation are in Appendix C. In general, the mean annual yields for observed and simulated values are in agreement for irrigated county/crop combinations with relative errors less than ±5% (see Table C1 in Appendix C). There is less agreement in the standard deviation between observed and simulated yields with the model underestimating the variance in yields in some cases and overestimating it in others (see Table C2 in Appendix C). The interannual variability in NASS yields is poorly captured in all county/crop combinations with negative NSE values for the majority of the cases (see Table C3 and Figures C1–C3 in Appendix C). This is not surprising because the NASS yields are a surrogate for actual yields, while the modeled yields represent potential maximum yields achievable under stress-free conditions and is capturing just the climate sensitivity. Year-to-year NASS yields are affected by several factors not captured by the model: management decisions, pests, diseases, and effects of extreme climate conditions, while our study is focused only on capturing the climate sensitivity aspect. In addition, there are uncertainties in the inputs to the model, particularly temperature and precipitation, and uncertainties associated with the NASS statistics themselves.

Differences in average growing season characteristics for 30 years of historical and future climate simulations are shown by crop for the CRB. Growing season start (crop emergence) dates, growing season end (harvest) dates, as well as actual growing season length (end date-start date) between the 2030s future climate simulations and historical climate simulation by crop for the CRB are shown. The y axis indicates the differences between future simulations and historical simulations in calendar days. The box plots capture spatial differences (across all of the CRB grid cells occupied by a specific crop) as well as multiple future climate scenarios.
Given that we use one set of crop parameters representative of the entire basin, the primary purpose of the calibration process was to capture the spatial heterogeneity in yields. There is good agreement between simulations and observations in this respect with relative errors less than ±5% for irrigated county/crop combinations over the evaluation period (see Table C1 in Appendix C).
Irrigation demands. Evaluation of the VIC-CropSyst-simulated irrigation water demand is based on actual diversion data at Bank's Lake, which stores and supplies water to the Columbia Basin Project, an irrigated area in central Washington. This reservoir stores water for irrigation purposes only and releases are for irrigation demands. Based on 2008, 2009, and 2010 data, actual irrigation diversions from Bank's Lake are in the range of 3.1–3.3 km3/yr. Simulated “top of the crop” irrigation demands for the period of 1977–2006 are on average about 2.7 km3. “Top of the crop” demands are crop evapotranspiration needs adjusted for irrigation system inefficiencies and are the amount of water applied; they do not include conveyance losses. The difference of 15–22% between the simulation results and actual diversions can be attributed to these conveyance losses. The WSDA ALL based irrigated acreage extent for this region is about 2,630 km2, which is also in the ballpark of the 2,711 km2 irrigated acreage in the Columbia Basin Project. This small difference in irrigated extent may also be partly responsible for the discrepancy between observed Bank's Lake diversions and simulated “top of the crop” irrigation demand. Unlike streamflow measurements, which are available at multiple locations, there is lack of appropriate data to evaluate irrigation demands at small watershed scales, and to evaluate the simulated seasonality of irrigation demands in this region.
4.2 Climate Change Impacts
In addition to considering the overall response of irrigation demands and irrigated crop yields to climate change, we consider the response to temperature, precipitation and CO2 changes separately, so we can identify the dominant effect as well complimentary and opposing effects, the balance between which can change over time. To highlight the differences in crop-specific responses, crops are grouped into three broad classes: (a) annual crops that are planted and harvested each year (e.g., corn, potatoes, and wheat), (b) perennial crops that are harvested as multiple cuttings in a year (e.g., pasture, alfalfa, and timothy), and (c) other perennial crops (e.g., apples, grapes, and hops). Crops within these groups show a similar pattern of response, and results of representative crops from each of these three groups are used for illustration.
Changes in growth season characteristics. Growing seasons and shifts in them will have implications for both the magnitude and seasonality of irrigation demands. While warming can result in a longer frost-free available growing season, it can also result in crop-specific and location-specific shifts in actual (time between emergence of crop and harvest) crop growth season characteristics. There are numerous factors that cause warming to influence these characteristics. Planting and growth can start early, and thereby require earlier irrigation. The longer available frost-free growing season has potential for irrigation demands to extend later into the season. Warming can result in accelerated growing degree day accumulation leading to earlier crop maturity, thereby resulting in a shorter actual growth season despite a longer available growing season, and thereby reduction in late season demands. In Figure 3, the pasture/hay varieties of crops emerge 6.6–9.5 days earlier on average in the 2030s as compared to current conditions. They are harvested as multiple cuttings and can make use of the entire available growing season. On average, the results show 7.3–9.7 days increase in the length of the actual growing season for these crops. Other perennial crops emerge earlier, but also mature earlier and we see small decreases or no change in the actual growing season length. Annual crops require limited growing degree day accumulation for emergence and emerge immediately after sowing. Because the sowing dates are constant between historical and future simulations, we do not simulate potential early emergence for these crops. However, they mature earlier, resulting in a shorter actual growing season (5.7–19.6 days earlier on average depending on the crop).
Changes in crop yields. We us the term “crop yield” as indicative of the potential yield attainable under no stress for a certain climatic condition. It is important to understand changes to crop yields as it is tightly connected to irrigation demands and water resources in a region. Increases in crop yield could require increased irrigation demands to facilitate that. Decreases in crop yields could lead producers to make adaptations to mitigate against these effects (e.g., switching the crop variety or crop mix, or double cropping) which in turn will have implications for irrigation demand. In general, the near-term (2030s) projections are for an increase in yields for pasture/hay varieties and perennials, while the direction of yield change is crop dependent for annuals (see Figure 4). The net response for each annual crop is dependent on the relative magnitudes of two competing effects on yields: the positive effect of CO2 fertilization and the negative effect of warming. CO2 fertilization leads to increases in yields for most crops. The exception is for C4 varieties of crops (such as corn) that have highly efficient CO2 photosynthetic pathways and are not as greatly impacted by increases CO2 levels (as compared to C3 crops which is the photosynthetic pathway used by the majority of the other crops grown in the region). Our model parameterization of CO2 fertilization on yields is not crop specific. There is one set of parameterization for C4 crops (e.g., corn types), and another for C3 crops (all other crops in the region). Therefore, the CO2 fertilization effect looks similar within these crop groups. The warming effect on yield is generally negative due to accelerated growth and earlier maturity. Conversely, warmer temperatures lead to higher yields in the pasture/hay varieties of crops that can make use of the longer available growing season (through multiple cuttings). The combination of increases in both temperature and CO2 levels results in relatively large overall increases in yields for the pasture/hay varieties of crops. Similarly, warming can lead to higher yields for perennials. The exception is for perennials where the harvested part is a fruit such as apples and grapes. To reflect on-the-ground practices, the model constrains the fruit load to be below a prescribed value. This practice is prevalent because large fruit loads can adversely affect flowering potential in future years. Therefore, increases in yield for these crops are limited to the prescribed load, and the potential for simulated yield increase exists only where the prescribed fruit load is higher than modeled fruit loads.

Crop-specific percent changes in irrigated crop yields between historical and the 2030s. The individual effects of elevated CO2 levels, changes in precipitation patterns, and increases in temperature are separated.
To illustrate spatial differences in yield change, we produce spatial maps for potato (chosen as a representative annual crop) and alfalfa (chosen as a representative perennial pasture/hay crop; Figure 5). For a given crop, the spatial variation in yield is largely a function of the temperature effect (Figure 5 focuses on the temperature effect). Potatoes (and other annual crops) experience larger negative yield changes in the Snake Plain region in the southeastern portion of the CRB, as compared to the rest of the basin. As shown in Figure 5, this spatial pattern in yield change follows the spatial patterns in the diurnal temperature change (DTR). The Snake Plain region has a relatively small DTR due to higher daily minimum temperatures, leading to relatively higher heat accumulation potential during the growing season. Accelerated growth under these conditions leads to relatively larger detrimental effects on yields. In contrast, the spatial patterns in alfalfa (and other pasture/hay varieties of crops), do not follow the spatial patterns of DTR; instead, they are more aligned with the spatial patterns in daily maximum temperatures during the summer months. This is because higher temperatures limit the growth of the crop, and the level of increased biomass production is constrained at these temperatures.

(a, d) Average changes in potato and alfalfa yields between 2030s middle climate scenario and historical condition, expressed as a fraction of historical yields; (b, e) average annual diurnal temperature range (DTR, calculated as Tmax-Tmin) in the 2030s for potato-growing and alfalfa-growing areas; and (c, f) 2030s daily maximum temperature (Tmax) averaged over the summer (June–August) for potato-growing and alfalfa-growing areas.
Our results are consistent with results found by others for the Western U.S. in some cases, but not always. For example, Lobell et al. (2006) characterized yield changes between +10% and −20% for perennial crops (oranges, walnuts, and grapes) in California around the 2030s with larger adverse effects further into the future. In comparison, the yield changes we observe for perennial fruit crops in the CRB (although different crops) are zero to +5%. Being warmer than the CRB, the negative effect of additional warming on yields can be expected to commence earlier in California. With respect to hay and annual crops, our direction of change in yields is consistent with observations made by Lee et al. (2011) for the California Central Valley: increases in hay yields and decreases in annual crop yields. Our range of changes in yield for potatoes, wheat, and apples are consistent with the results of Stöckle et al. (2010) for four locations in eastern Washington.
Changes in irrigation demands. Annual average irrigation demands for the CRB are about 16 km3. Overall, we see a ∼4% increase in average annual irrigation demands in the CRB in the 2030s. This varies between −1% and +12% by subwatershed (results not shown). Figure 6 shows the overall effect as well as individual effects annually and by season. Elevated CO2 levels result in small decreases in irrigation demand when considered in isolation due to increased water-use efficiencies (i.e., increased stomatal resistance). Increases in early season precipitation result in small decreases in irrigation demands; decreases in precipitation during the other parts of the growing season result in small increases in irrigation demands (i.e., precipitation deficits are compensated by additional irrigation application). Temperature changes are the dominant effect describing overall changes in irrigation demands. In addition to affecting potential evapotranspiration, warming impacts the seasonality of irrigation demands by accelerating the crop's phenological stages, resulting in larger increases in early season demand and potential decreases in late season demand. Annual changes mirror summer season changes because that is the season with the relatively largest demands. In addition, demand increases in the early season are offset by decreases in the late season. There is variation in late season demand changes due to crop-specific differences in crop maturity/harvest timings and spatial differences in the crop mix, as described below.

Percent changes in irrigation demand between 2030s and historical conditions. Changes are separated by early season (March–May) demands, summer (June–August) demands, and late season (September–November) demands. The individual effects of elevated CO2 levels, changes in precipitation patterns, and change in temperature are also separated.
Figure 7 illustrates crop-specific responses for representative crops within the three classes of crops described above: pasture/hay varieties, other perennials, and annuals. Annual crops are projected to have a shorter growing season length due to accelerated growth and earlier maturity and are also associated with decreases in irrigation demands, while pasture/hay crops (harvested as multiple cuttings through the season) can make use of the longer available growing season under warmer conditions and have increases in irrigation demands. Perennial crops also have increased irrigation demands (even if their harvest timing is earlier under warming) because they are irrigated through the end of the season (i.e., postharvest) to keep them alive. There are some differences in the magnitudes of irrigation demand change across perennial crops. For example, in perennials where the harvested part is a fruit (such as apples), the fruit is a carbon sink, and once the carbon sink is lost at harvest, the plant growth mechanism starts shutting down and the stomata constrict, with reduced irrigation needs after harvest. In other perennials such as hops, where this response is not applicable, irrigation demands continue to be high after harvest until the end of the season.

Crop-specific percent changes in irrigation demands between 2030s and historical simulations. The individual effects of elevated CO2 levels, changes in precipitation, and changes in temperature are separated and shown individually along with the overall effect.
There are some spatial differences in the overall effect of climate change on irrigation demands (see Figure 8). Within Washington State, the Yakima River tributary of the CRB sees 7% increases in demand at the annual scale, while the Columbia Basin Project area sees negligible changes or decreases in demand at the annual scale. Differences in the crop mix (i.e., the relative proportion of annual versus perennials versus pasture/hay varieties of crops) largely explain this (see Figure 9). Decreases (negative changes) in average annual irrigation demands correspond to grid cells with crop area comprising mostly annuals. Large increases in average annual irrigation demands correspond to grid cells with crop area comprising negligible amounts of annual crop extent and large amounts of pasture/hay (that can make use of the longer available growing season) crop extent.

Spatial differences in average changes in irrigation demand between 2030s and historical conditions. Changes are separated by early season (March–May), summer (June–August), and late season (September–November) demands. Annual demands are also shown.

Average percentage change in irrigation demand in the 2030s as compared to historical conditions. Each point is an irrigated grid cell in the CRB. The x axis corresponds to the fraction of irrigated crop area that is occupied by annuals while the colors on the grey scale correspond to the fraction of irrigated crop area that is occupied by irrigated pasture/hay varieties. Note that grid cells with the highest fractions occupied by pasture/hay varieties are associated with the largest increases in irrigation demand.
There is limited work characterizing the implications of climate change for irrigation demands. Our results are consistent with irrigation demand increase projections of global scale studies (Doll, 2002; Fischer et al., 2007; Wada et al., 2013), although the models, crop categorization, climate, and socioeconomic factors considered are all different and a direct comparison is not relevant. Our process-based methodology is more closely aligned with the study performed by Konzmann et al. (2013) who projects decreases in demands by the end of the century due to elevated CO2 levels and shifts in the seasonality of crop growth stages and time to maturity. While there are differences in the models (global scale versus regional scale) and climate scenarios themselves, one important difference between Konzmann et al. (2013) and our work is our detailed characterization of crops, and the absence of explicit consideration of pasture/hay varieties of crops by Konzmann et al. (2013). Konzmann et al. (2013) lumps pasture/hay varieties together with several other crops including potatoes which have very different responses, and this distinction is important in our region. Within the CRB, Vano et al. (2010) project decreases in irrigation demand for tree fruit in the Yakima River basin, while in this study we project increases in irrigation demand. There are multiple differences between Vano et al. (2010) and our simulation: geographical extent, the climate scenarios, simulation time period, crop model and parameters used, and the fact that crop model simulations by Vano et al. (2010) are made for a few specific locations in the basin, rather than distributed across the entire basin (as we have done).
5 Discussion
Because each crop type responds differently to climate change, crop mix plays a large role in determining the net subregional and overall regional responses of irrigation demands to changes in climate. The near-term ∼4% net increase in irrigation demands expected at the CRB scale is a function of the irrigated crop mix being primarily comprised of pasture/hay crop varieties, which have a positive response to climate change. Given that the irrigation extent of this crop group is approximated (see section 3.3), this sensitivity has implications for the estimates of regional changes in irrigation demands presented here. The modeled results could be an overestimate or underestimate depending on the actual irrigated pasture/hay extent. This work, however, underscores the importance of considering the crop mix in a spatially explicit manner to better capture the regional response of irrigation demand to changes in climate. The resources to meet irrigation demands, implications of changes in irrigation demand, and plans to address them are made at subregional watershed scales. At these scales, the average changes in irrigation demands in the near term vary from −1% to 12%, again a reflection of the crop mix. Therefore, different subregions are impacted at varying levels, and thus have different planning needs. For example, one of the watersheds that has a relatively larger average increase in irrigation demand (7%) is the Yakima River watershed. Historically, this watershed experiences shortages of surface water supply to meet irrigation demands, and can be more adversely affected than other regions by increases in irrigation demand.
Low-economic-value crops such as pasture/hay varieties play an important role in the assessment of the regional impacts of climate change on agriculture. Historically, these assessment have primarily focused on cereal grains (maize, wheat, soybean, and rice), which are primary contributors of human caloric consumption (Lobell et al., 2013). Some recent work has included high-economic-value perennial crops on a local/regional scale (Lobell and Field, 2011; Stöckle et al., 2010; Vano et al., 2010). However, as seen in these results, low-value crops such as pasture/hay varieties play an important role in characterizing overall regional impacts of climate change (especially with respect to irrigation demands) and should not be ignored in assessing regional-scale impacts. In addition, while pasture/hay varieties of crops have increased demands, they can also be more flexibly managed by producers (as compared to higher-value crops) to improve water management during periods of water shortages. For example, the fact that these crop varieties are often harvested through multiple cuttings allows producers the flexibility to avoid having to irrigate during the late irrigation season through one less cutting, allowing for the water to be used to irrigate higher-value crops or to supplement competing uses such as instream flows for fish and/or hydropower generation. An active water market can provide the incentive for and facilitate these types of decisions (Wheeler et al., 2014).
Shifts in the seasonality of irrigation demands can also have implications for competing uses of surface water. For example, instream flows for fish are typically critical in both early and late seasons. Because we project a shift in irrigation demand to earlier in the season, early season (March–May) competition for water between uses will be exacerbated, while late season (September–November) water competition may be alleviated, in some cases (at least for near-term climate change). This outcome has important implications for reservoir operations, particularly when the reservoirs are managed to meet multiple objectives.
Agricultural producer adaptations can impact projected changes to irrigation demands. Our irrigation demand projections are likely a lower bound because we consider only one producer adaptation to warming: an earlier planting which (in combination with warming) leads to earlier emergence. Other adaptations can also lead to increased irrigation demands, especially during the late season. These include (1) adoption of more slowly maturing crops to address yield decreases due to a shortened time to harvest, (2) double cropping, which is planting a second crop after the first has been harvested, because the available growing season is longer, and (3) increases in irrigation extent (although this is not straightforward given water right limitations and the fact that most nonmarginal lands are already under cultivation). One adaptation that could decrease demand is changes in irrigation efficiency through technology adoption or management (e.g., more efficient irrigation infrastructure or precision agriculture). This, however, has implications for downstream users that rely on return flows from upstream inefficient irrigation systems to meet their needs (i.e., improvements in irrigation efficiency can reduce deep percolation losses that return to the stream through enhanced baseflow later in the irrigation season; Perry et al., 2009). Finally, macroscale trends in crop mix can have a large impact on regional irrigation demand. One important trend in the region is an increase in the areal extent of wine grapes in Washington State. There has been a fivefold increase in acreage in the last 30 years as per the Washington State Wine Commission. Wine grapes are typically less water intensive as compared to other crops, and this could reduce the overall irrigation demand if the increase in wine acres is a result of a switch from more water intensive crops. Another potential macroscale trend is a migration of certain crops (that are normally grown in warmer climates) into the PNW. Near-term climate change is projected to benefit PNW agriculture in general (although, as we have mentioned, this is very site and crop specific), and crops historically only grown in warmer regions may migrate into the PNW and affect irrigation demands in potentially unexpected ways.
While herein we present the near-term overall response of irrigation demands and crop yields to climate change, these responses can change over time. The factors affecting these responses (e.g., elevated CO2 levels and warming) are not independent of each other and the response is crop-specific, competing, nonlinear (Schlenker & Roberts, 2008) and nonmonotonic (Chang, 2002), with regional differences. While the balance between these effects generally result in a positive overall effect for near-term crop yields, this balance can shift further into the future, with the potential for the overall response and producer adaptations to these responses to change over time. While the near-term projection is generally positive in terms of crop yield (and even irrigation demand), longer-term projections may not be nearly as optimistic as temperature effects began to dominate over the positive effect of CO2 fertilization. This has long-term planning implications for both agricultural producers and water managers within the region.
6 Conclusions
The CRB as a whole has a generally positive outlook in the near-term with respect to irrigated agriculture, with increases in yields for several crops, small yield decreases for annual crops (which can potentially be addressed by varietal adaptation), and small increases in overall irrigation demands. However, there are specific instances and locations (primarily dependent on the crop mix) where negative impacts are expected. For example, increased early season stress with respect to competing uses and specific subwatersheds (some that are already stressed with overallocation of resources) having increasing competition for resources. Therefore, to better inform regional agricultural resource decision making, there is a need for spatially explicit studies that can account for location-specific and crop-specific factors, including crop mix and other crop management decisions.
Uncertainties in climate impact projections for regional-scale irrigated agriculture relate not just to uncertainties in climate projections and crop response to changing conditions, but also to human decisions. These include the crop mix choice, crop varietal adaptations, irrigation extent, and irrigation technology and management, all of which have implications for regional-scale water demand and water resources planning.
Furthermore, this near-term positive outlook may not continue further out into the future. As the CO2 levels increase, their positive effect on yield will likely diminish, although this response is uncertain (Leipprand & Gerten, 2006). With increased warming, heat stress and other detrimental effects of higher temperatures will start to dominate. This nonlinear and competing nature of the factors that impact crop yield response to climate change may eventually result in a switch from positive to negative, although there is considerable uncertainty on when (and for which crops) this will happen. This highlights the importance of long-term planning at the nexus of agriculture and water management to ensure a sustainable and economically viable agricultural production system in the CRB.
While regional-scale models (such as the one described here) help characterize impacts at scales in which policy decisions are made, their utility is not restricted to regional-scale assessments. This is especially true for a region like the CRB where the majority of the agricultural production is exported out of the region. With increased global connectivity and global commodity trading, regional impacts can have a significant effect on global impacts and have the potential to inform (and evaluate) global assessments.
This work fills a gap in our regional-scale understanding of the impacts of climate change on irrigation demands. As an immediate extension of this work, we are pursuing three future directions. The first is an understanding of how the net response of crop yields and irrigation demands to climate change evolve over time, and the thresholds at which near-term positive responses can switch signs and become a negative response. The second is quantifying the impacts of climate change on surface water availability used for irrigation (in conjunction with the impacts on irrigation water demand addressed herein), the extent to which water rights curtailment (restricting water rights during periods of water shortages) may change in the future, and how changes in water rights curtailment impacts agricultural production. This involves consideration of potential changes to both water supply and demand in conjunction, and incorporation of regional water management and regulations (such as instream flow requirements for fish) into the modeling framework. The third is an assessment of the synergies and tradeoffs between multiple short-term adaptations such as fallowing (not planting) and deficit irrigation (applying less than optimal amounts of irrigation water); and long-term adaptations such as the adoption of more efficient irrigation technology, and selection of crop variety (e.g., those that mature more slowly) and/or crop mix, in response to the impacts of climate change on crop yields and irrigation demands. Together, these will inform coadaptation of food and water systems to ensure the sustainability of irrigated agriculture in the region, while meeting competing needs.
Acknowledgments
This research is funded from the BioEarth project—Department of Agriculture, National Institute of Food and Agriculture grant 2011–67003-30346, and the Forecast project—Washington Department of Ecology grant C1400281. The authors thank two reviewers and our stakeholders who have contributed invaluable insights to the two projects. Model code and data associated with this work is available here: https://doi.org/10.6084/m9.figshare.5890735.v1. The authors can also be contacted at [email protected] or [email protected] for additional clarification.
Appendix A
Table A1 lists the major crop groups that are modeled.
Grains/cereal/beans/lentils | Pasture/hay | Vegetables/berries/herbs/other | Tree fruit and grapes | Seed |
---|---|---|---|---|
Winter wheat | Pasture | Potatoes | Apple | Sod grass seed |
Barley | Alfalfa | Garlic | Cherry | Bluegrass seed |
Dry beans | Timothy | Carrots | Peach | Corn seed |
Grain corn | Clover hay | Sweet corn | Juice grapes | Ryegrass seed |
Spring wheat | Triticale | Onions | Pear | Pea seed |
Peas | Hay | Pea green | Wine grapes | Bean seed |
Chickpea | Grass for hay | Mint | Alfalfa seed | |
Oats | Herbs | Safflower seed | ||
Buckwheat | Caneberry | Grass seed | ||
Blueberry | Mustard seed | |||
Sugar beets | ||||
Bean green | ||||
Asparagus | ||||
Hops |
- Note. Related to each crop group, there could be multiple crop codes in the USDA CDL and WSDA ALL. For example, there are five crop codes related to the broad category of “pasture.”
Appendix B
Tables B1–B4 list the crop parameters used in the model.
Alfalfa | Alfalfa seed | Apple | Cherry | Walnut | Other tree fruit | Barley | Dry bean | |
---|---|---|---|---|---|---|---|---|
[Crop] | ||||||||
Canopy_growth | Canopy cover | Canopy cover | Canopy cover | Canopy cover | Canopy cover | Canopy cover | Canopy cover | Canopy cover |
Harvested_part | Leaf | Seed | Fruit | Fruit | Fruit | Fruit | Grain | Grain |
C_species | C3 | C3 | C3 | C3 | C3 | C3 | C3 | C3 |
Life_cycle | Perennial | Perennial | Perennial | Perennial | Perennial | Perennial | Annual | Annual |
[Emergence] | ||||||||
Model | Thermal time | Thermal time | Thermal time | Thermal time | Thermal time | Thermal time | Thermal time | Thermal time |
[Growth] | ||||||||
TUE_equation | TUE_curve | TUE_curve | TUE_curve | TUE_curve | TUE_curve | TUE_curve | TUE_curve | TUE_curve |
TUE_scaling_coef | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 |
TUE_scaling_coef_veg | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 |
TUE_at_1pKa_VPD | 2.9 | 2.9 | 4.8 | 4.8 | 1.48 | 4.8 | 4.8 | 4.3 |
TUE_at_1pKa_VPD_veg | 2.9 | 2.9 | 4.8 | 4.8 | 1.48 | 4.8 | 4.8 | 4.3 |
RUE_PAR | 1.7 | 1.7 | 2.8 | 2.7 | 0.86 | 2.8 | 2.8 | 2.5 |
LWP_reduces_canopy_expansion | −1,000 | −1,000 | −800 | −800 | −800 | −800 | −1,000 | −1,000 |
LWP_stops_canopy_expansion | −1,300 | −1,300 | −1,200 | −1,200 | −1,200 | −1,200 | −1,300 | −1,300 |
Early_growth_limit_temp | 8 | 8 | 10 | 10 | 10 | 10 | 8 | 8 |
[Transpiration] | ||||||||
ET_crop_coef | 1.23 | 1.23 | 1.1 | 1.1 | 1.15 | 1.1 | 1.14 | 1.18 |
Max_water_uptake | 13 | 13 | 12 | 12 | 12 | 12 | 13 | 12 |
Stomatal_closure_leaf_water_pot | −1,300 | −1,300 | −1,300 | −1,000 | −1,300 | −1,300 | −1,300 | −1,000 |
Wilt_leaf_water_pot | −2,000 | −2,000 | −2,000 | −1,500 | −2,000 | −2,000 | −2,000 | −1,500 |
[Canopy_cover] | ||||||||
Initial_cover | 0.05 | 0.05 | 0.05 | 0.05 | 0.05 | 0.05 | 0.05 | 0.05 |
Maximum_cover | 0.85 | 0.85 | 0.5 | 0.5 | 3.5 | 0.5 | 0.8 | 0.75 |
Mature_green_cover | 0.2 | 0.2 | 0.5 | 0.5 | 3.5 | 0.5 | 0.3 | 0.2 |
Mature_total_cover | 0.7 | 0.7 | 0.5 | 0.5 | 0.5 | 0.5 | 0.7 | 0.7 |
[Phenology] | ||||||||
Maturity_significant | TRUE | TRUE | TRUE | TRUE | TRUE | TRUE | TRUE | TRUE |
Emergence | 130 | 120 | 130 | 130 | 24 | 50 | ||
Flowering | 850 | 850 | 70 | 50 | 70 | 70 | 630 | 574 |
Peak_LAI | 800 | 800 | 240 | 450 | 240 | 240 | 600 | 813 |
Filling | 950 | 950 | 200 | 470 | 200 | 200 | 680 | 670 |
Maturity | 1,500 | 1,500 | 1,700 | 1,250 | 1,800 | 1,800 | 1,115 | 1,366 |
Senescence | 1,000 | 1,000 | 2,000 | 1,400 | 2,000 | 2,000 | 720 | 1,047 |
Tuber_init | ||||||||
Resolution | Day | Day | Day | Day | Day | Day | Day | |
Base_temp | 5 | 5 | 5 | 5 | 5 | 5 | 3 | 4 |
Cutoff_temp | 25 | 25 | 25 | 25 | 25 | 25 | 25 | 25 |
Maximum_temp | 25 | 25 | 25 | 25 | 25 | 25 | 25 | 25 |
[Dormancy] | ||||||||
Chill requirement | 800 | 800 | 800 | 800 | ||||
Deg_day_bud break chill req sat | 130 | 120 | 130 | 130 | ||||
Deg day bud break chill req not sat | 150 | 140 | 150 | 150 | ||||
[Root] | ||||||||
Root_sensitivity_water_stress | 0.2 | 0.2 | 0 | 0 | 0 | 0 | 0.2 | 0.2 |
Max_root_depth | 1.8 | 1.8 | 0.8 | 1.5 | 0.8 | 0.8 | 1.5 | 0.7 |
Root_length_at_emergence (cm) | 150 | 150 | 12 | 7 | ||||
Sow_depth | 0.08 | 0.08 | ||||||
Root_density_distribution_curvature | 0.00001 | 0.00001 | 0.001 | 0.00001 | 0.00001 | 0.00001 | 0.00001 | 0.00001 |
[Morphology] | ||||||||
Max_canopy_height | 0.7 | 0.7 | 3 | 1 | 0.4 | |||
[Inactive_period] | ||||||||
Consider_inactive_days | 7 | 7 | 7 | 7 | 7 | 7 | 7 | |
Inducement_temperature | 5 | 5 | 5 | 5 | 5 | 5 | 5 | |
Start_DOY | 330 | 330 | 330 | 330 | 330 | 330 | ||
Minimum_duration | 90 | 90 | 90 | 90 | 30 | 90 | ||
[Season] | ||||||||
Start_DOY_WA | Not applicable | Not specified | Not specified | Not specified | 105 | 140 | ||
Start_DOY_ID | 105 | 140 | ||||||
Start_DOY_OR | 105 | 140 | ||||||
Duration_WA | Not applicable | 100 | 104 | |||||
Duration_OR | 100 | 104 | ||||||
Duration_ID | 100 | 104 | ||||||
[Harvest] | ||||||||
Unstressed | 0.85 | 0.45 | 0.45 | |||||
Translocation_max | 0.35 | 0.3 | ||||||
[Fruit] | ||||||||
Grape | FALSE | FALSE | FALSE | FALSE | ||||
Fract_tot_solids | 0.3 | 0.25 | 0.8 | 0.3 | ||||
Max_fruit_load | 39,200 | 22,000 | 3,363 | 72,000 | ||||
[CO2] | ||||||||
Growth_ratio | 1.2 | 1.2 | 1.2 | 1.2 | 1.2 | 1.2 | 1.2 | 1.2 |
Elevated_reference_conc | 600 | 600 | 600 | 600 | 600 | 600 | 600 | 600 |
Baseline_reference_conc | 360 | 360 | 360 | 360 | 360 | 360 | 360 | 360 |
- Note. There are 26 sets of crop parameters that are applied to complete list of crop types in the region. If the crop parameters for a particular crop type in the USDA CDL or WSDA ALL are not available, the parameters corresponding to the closest crop from this list are used. This is the first of set of 4 tables with parameter values.
Canola | Generic berry | Generic vegetable | Hops | Juice grapes | Wine grapes | |
---|---|---|---|---|---|---|
[Crop] | ||||||
Canopy_growth | Canopy cover | Canopy cover | Canopy cover | Canopy cover | Canopy cover | Canopy cover |
Harvested_part | Grain | Treat as leaf | Leaf | Flower | Fruit | Fruit |
C_species | C3 | C3 | C3 | C3 | C3 | C3 |
Life_cycle | Annual | Annual | Annual | Perennial | Perennial | Perennial |
[Emergence] | ||||||
Model | Thermal time | Thermal time | Thermal time | Thermal time | Thermal time | Thermal time |
[Growth] | ||||||
TUE_equation | TUE_curve | TUE_curve | TUE_curve | TUE_curve | TUE_curve | TUE_curve |
TUE_scaling_coef | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 |
TUE_scaling_coef_veg | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 |
TUE_at_1pKa_VPD | 3.1 | 4.8 | 4.8 | 4.8 | 4.8 | 4.8 |
TUE_at_1pKa_VPD_veg | 3.1 | 4.8 | 4.8 | 4.8 | 4.8 | 4.8 |
RUE_PAR | 1.8 | 2.8 | 2.8 | 2.8 | 2.8 | 2.8 |
LWP_reduces_canopy_expansion | −1,000 | −1,000 | −1,000 | −1,000 | −800 | −800 |
LWP_stops_canopy_expansion | −1,300 | −1,300 | −1,300 | −1,300 | −1,200 | −1,200 |
Early_growth_limit_temp | 8 | 8 | 8 | 12 | 10 | 10 |
[Transpiration] | ||||||
ET_crop_coef | 1.18 | 1.05 | 0.9 | 1.11 | 1.1 | 1.1 |
Max_water_uptake | 12 | 10 | 10 | 14 | 12 | 12 |
Stomatal_closure_leaf_water_pot | −1,200 | −800 | −800 | −1,300 | −1,300 | −1,300 |
Wilt_leaf_water_pot | −2,000 | −1,200 | −1,200 | −2,000 | −2,000 | −2,000 |
[Canopy_cover] | ||||||
Initial_cover | 0.05 | 0.05 | 0.05 | 0.05 | 0.05 | 0.05 |
Maximum_cover | 0.8 | 0.5 | 0.6 | 0.85 | 0.8 | 0.65 |
Mature_green_cover | 0.4 | 0.5 | 0.2 | 0.7 | 0.8 | 0.65 |
Mature_total_cover | 0.7 | 0.5 | 0.6 | 0.8 | 0.8 | 0.65 |
[Phenology] | ||||||
Maturity_significant | TRUE | TRUE | TRUE | TRUE | TRUE | TRUE |
Emergence | 70 | 96 | 96 | 200 | 300 | |
Flowering | 720 | 350 | 350 | 930 | 370 | 400 |
Peak_LAI | 750 | 960 | 960 | 977 | 850 | 850 |
Filling | 760 | 365 | 365 | 970 | 1,400 | 1,300 |
Maturity | 1,700 | 2,150 | 2,150 | 2,218 | 2,000 | 1,870 |
Senescence | 900 | 1,096 | 1,096 | 2,000 | 1,900 | 1,900 |
Tuber_init | ||||||
Resolution | Day | Day | Day | Day | Day | Day |
Base_temp | 1 | 3 | 3 | 5 | 5 | 5 |
Cutoff_temp | 25 | 25 | 25 | 30 | 30 | 30 |
Maximum_temp | 25 | 25 | 25 | 30 | 30 | 30 |
[Dormancy] | ||||||
Chill requirement | 100 | 100 | ||||
Deg_day_bud break chill req sat | 200 | 300 | ||||
Deg day bud break chill req not sat | 230 | 350 | ||||
[Root] | ||||||
Root_sensitivity_water_stress | 0.2 | 0.5 | 0.5 | 0 | 0 | 0 |
Max_root_depth | 1.5 | 1.2 | 0.5 | 1.5 | 1 | 0.8 |
Root_length_at_emergence (cm) | 19.2 | 12 | 5 | 150 | 10 | 10 |
Sow_depth | 0.08 | 0 | 0.05 | 0 | 0 | 0 |
Root_density_distribution_curvature | 0.00001 | 0.00001 | 0.00001 | 0.00001 | 0.00001 | 0.00001 |
[Morphology] | ||||||
Max_canopy_height | 0.6 | 1 | 0.4 | 5 | 2 | 1.2 |
[Inactive_period] | ||||||
Consider_inactive_days | 7 | 7 | 7 | 7 | ||
Inducement_temperature | 5 | 5 | 5 | 5 | ||
Start_DOY | 330 | 330 | 330 | 330 | ||
Minimum_duration | 30 | 90 | 90 | 90 | ||
[Season] | ||||||
Start_DOY_WA | 105 | 76 | 76 | |||
Start_DOY_ID | 105 | 76 | 76 | |||
Start_DOY_OR | 105 | 76 | 76 | |||
Duration_WA | 122 | 165 | 165 | |||
Duration_OR | 122 | 165 | 165 | |||
Duration_ID | 122 | 165 | 165 | |||
[Harvest] | ||||||
Unstressed | 0.24 | 0.9 | 0.1 | 0 | ||
Translocation_max | 0.2 | 0 | 0 | |||
[Fruit] | ||||||
Grape | FALSE | TRUE | TRUE | |||
Fract_tot_solids | 0.16 | 0.25 | ||||
Max_fruit_load | 34,375 | 15,000 | ||||
[CO2] | ||||||
Growth_ratio | 1.2 | 1.2 | 1.2 | 1.2 | 1.2 | 1.2 |
Elevated_reference_conc | 600 | 600 | 600 | 600 | 600 | 600 |
Baseline_reference_conc | 360 | 360 | 360 | 360 | 360 | 360 |
- Note. There are 26 sets of crop parameters that are applied to complete list of crop types in the region. If the crop parameters for a particular crop type in the USDA CDL or WSDA ALL are not available, the parameters corresponding to the closest crop from this list are used. This is the second of set of four tables with parameter values.
Lentil | Mint | Oat | Pasture seed | Potato | Pea seed | |
---|---|---|---|---|---|---|
[Crop] | ||||||
Canopy_growth | Canopy cover | Canopy cover | Canopy cover | Leaf-area-index | Canopy cover | Canopy cover |
Harvested_part | Grain | Leaf | Grain | Seed | Tuber | Grain |
C_species | C3 | C3 | C3 | C3 | C3 | C3 |
Life_cycle | Annual | Perennial | Annual | Perennial | Annual | Annual |
[Emergence] | ||||||
Model | Thermal time | Thermal time | Thermal time | Thermal time | Thermal time | Thermal time |
[Growth] | ||||||
TUE_equation | TUE_curve | TUE_curve | TUE_curve | TUE_curve | TUE_curve | TUE_curve- |
TUE_scaling_coef | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 |
TUE_scaling_coef_veg | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 | 0.6 |
TUE_at_1pKa_VPD | 3.4 | 5.5 | 4.8 | 2.5 | 5.8 | 4.5 |
TUE_at_1pKa_VPD_veg | 3.4 | 5.5 | 4.8 | 2.5 | 5.8 | 4.5 |
RUE_PAR | 2 | 3.2 | 2.8 | 1.5 | 3.4 | 2.6 |
LWP_reduces_canopy_expansion | −1,000 | −1,000 | −1,000 | −1,000 | −700 | −1,000 |
LWP_stops_canopy_expansion | −1,300 | −1,300 | −1,300 | −1,300 | −1,000 | −1,300 |
Early_growth_limit_temp | 8 | 8 | 8 | 8 | 8 | 8 |
[Transpiration] | ||||||
ET_crop_coef | 1.13 | 1.23 | 1.14 | 1.23 | 1.18 | 1.18 |
Max_water_uptake | 12 | 14 | 13 | 13 | 13 | 12 |
Stomatal_closure_leaf_water_pot | −1,000 | −1,300 | −1,300 | −1,300 | −800 | −1,000 |
Wilt_leaf_water_pot | −1,500 | −2,000 | −2,000 | −2,000 | −1,200 | −1,500 |
[Canopy cover] | ||||||
Initial_cover | 0.05 | 0.05 | 0.05 | 0.05 | 0.05 | 0.05 |
Maximum_cover | 0.75 | 0.97 | 0.8 | 0.7 | 0.95 | 0.8 |
Mature_green_cover | 0.2 | 0.97 | 0.3 | 0.2 | 0.7 | 0.05 |
Mature_total_cover | 0.7 | 0.97 | 0.7 | 0.5 | 0.7 | 0.7 |
[Phenology] | ||||||
Maturity_significant | TRUE | TRUE | TRUE | TRUE | TRUE | TRUE |
Emergence | 50 | 24 | 100 | 75 | ||
Flowering | 574 | 1,100 | 630 | 850 | 1,500 | 545 |
Peak_LAI | 813 | 700 | 600 | 800 | 1,350 | 720 |
Filling | 670 | 2,000 | 680 | 950 | 1,200 | 635 |
Maturity | 1,366 | 1,250 | 1,115 | 1,500 | 2,100 | 1,328 |
Senescence | 1,047 | 1,300 | 720 | 1,000 | 1,600 | 780 |
Tuber_init | 1,200 | |||||
Resolution | Day | Day | Day | Day | Day | Day |
Base_temp | 4 | 5 | 3 | 5 | 3 | 4 |
Cutoff_temp | 25 | 30 | 25 | 25 | 25 | 25 |
Maximum_temp | 25 | 30 | 25 | 25 | 25 | 25 |
[Dormancy] | ||||||
Chill requirement | ||||||
Deg_day_bud break chill req sat | ||||||
Deg day bud break chill req not sat | ||||||
[Root] | ||||||
Root_sensitivity_water_stress | 0.2 | 0 | 0.2 | 0.5 | 0.3 | 0.2 |
Max_root_depth | 0.6 | 1 | 1.5 | 1.5 | 0.6 | 0.7 |
Root_length_at_emergence (cm) | 7 | 1 | 12 | 15 | 12 | 7 |
Sow_depth | 0.08 | 0 | 0.08 | 0.08 | 0.08 | |
Root_density_distribution_curvature | 0.00001 | 0.00001 | 0.00001 | 0.00001 | 0.00001 | 0.00001 |
[Morphology] | ||||||
Max_canopy_height | 0.4 | 0.8 | 1 | 0.5 | 0.06 | 0.7 |
[Inactive_period] | ||||||
Consider_inactive_days | 7 | 7 | 7 | 7 | 7 | |
Inducement_temperature | 5 | 5 | 5 | 5 | 5 | |
Start_DOY | 330 | 330 | ||||
Minimum_duration | 90 | 90 | ||||
[Season] | ||||||
Start_DOY_WA | 120 | 105 | 120 | 91 | ||
Start_DOY_ID | 120 | 105 | 120 | 91 | ||
Start_DOY_OR | 120 | 105 | 120 | 91 | ||
Duration_WA | 114 | 100 | 156 | 105 | ||
Duration_OR | 114 | 100 | 156 | 105 | ||
Duration_ID | 114 | 100 | 156 | 105 | ||
[Harvest] | ||||||
Unstressed | 0.43 | 1 | 0.35 | 0.8 | 0.45 | |
Translocation_max | 0.3 | 0 | 0.3 | 0 | 0.3 | |
[Fruit] | ||||||
Grape | ||||||
Fract_tot_solids | ||||||
Max_fruit_load | ||||||
[CO2] | ||||||
Growth_ratio | 1.2 | 1.2 | 1.2 | 1.2 | 1.2 | 1.2 |
Elevated_reference_conc | 600 | 600 | 600 | 600 | 600 | 600 |
Baseline_reference_conc | 360 | 360 | 360 | 360 | 360 | 360 |
- Note. There are 26 sets of crop parameters that are applied to complete list of crop types in the region. If the crop parameters for a particular crop type in the USDA CDL or WSDA ALL are not available, the parameters corresponding to the closest crop from this list are used. This is the third of set of four tables with parameter values.
Pasture | Sugar beet | Sweet corn | Grain corn | Spring wheat | Winter wheat | |
---|---|---|---|---|---|---|
[Crop] | ||||||
Canopy_growth | Leaf-area-index | Canopy cover | Canopy cover | Canopy cover | Canopy cover | Leaf_area_index |
Harvested_part | Leaf | Tuber | Grain | Grain | Grain | Grain |
C_species | C3 | C3 | C4 | C4 | C3 | C3 |
Life_cycle | Perennial | Annual | Annual | Annual | Annual | Annual |
[Emergence] | ||||||
Model | Thermal time | Thermal time | Thermal time | Thermal time | Thermal time | Thermal time |
[Growth] | ||||||
TUE_equation | TUE_curve | TUE_curve- | TUE_curve- | TUE_curve | TUE_curve | TUE_curve |
TUE_scaling_coef | 0.6 | 0.6 | 0.5 | 0.5 | 0.6 | 0.6 |
TUE_scaling_coef_veg | 0.6 | 0.6 | 0.5 | 0.5 | 0.6 | 0.6 |
TUE_at_1pKa_VPD | 2.5 | 4.8 | 8 | 8 | 4.8 | 4.8 |
TUE_at_1pKa_VPD_veg | 2.5 | 4.8 | 8 | 8 | 4.8 | 4.8 |
RUE_PAR | 1.5 | 2.8 | 3.6 | 3.6 | 2.8 | 2.8 |
LWP_reduces_canopy_expansion | −1,000 | −1,000 | −1,000 | −1,000 | −1,000 | −1,000 |
LWP_stops_canopy_expansion | −1,300 | −1,500 | −1,300 | −1,300 | −1,300 | −1,300 |
Early_growth_limit_temp | 8 | 8 | 12 | 12 | 8 | 8 |
[Transpiration] | ||||||
ET_crop_coef | 1.23 | 1.23 | 1.2 | 1.25 | 1.14 | 1.19 |
Max_water_uptake | 13 | 13 | 13 | 13 | 13 | 13 |
Stomatal_closure_leaf_water_pot | −1,300 | −800 | −1,100 | −1,100 | −1,300 | −1,300 |
Wilt_leaf_water_pot | −2,000 | −1,200 | −1,600 | −1,600 | −2,000 | −2,000 |
[Canopy cover] | ||||||
Initial_cover | 0.05 | 0.05 | 0.05 | 0.05 | 0.05 | 0.05 |
Maximum_cover | 0.7 | 0.95 | 0.92 | 0.92 | 0.8 | 0.8 |
Mature_green_cover | 0.2 | 0.7 | 0.5 | 0.5 | 0.3 | 0.3 |
Mature_total_cover | 0.5 | 0.6 | 0.7 | 0.7 | 0.7 | 0.7 |
[Phenology] | ||||||
Maturity_significant | TRUE | TRUE | TRUE | TRUE | TRUE | TRUE |
Emergence | 116 | 85 | 80 | 24 | 97 | |
Flowering | 850 | 4,000 | 842 | 1,442 | 630 | 1,360 |
Peak_LAI | 800 | 1,260 | 734 | 1,350 | 600 | 1,200 |
Filling | 950 | 4,000 | 905 | 1,589 | 680 | 1,420 |
Maturity | 1,500 | 3,059 | 1,150 | 1,820 | 1,115 | 1,990 |
Senescence | 1,000 | 2,530 | 1,050 | 1,650 | 720 | 1,540 |
Tuber_init | 4,000 | |||||
Resolution | Day | Day | Day | Day | Day | Day |
Base_temp | 5 | 1.1 | 5 | 5 | 3 | 0 |
Cutoff_temp | 25 | 30 | 30 | 30 | 25 | 25 |
Maximum_temp | 25 | 30 | 30 | 30 | 25 | 25 |
[Dormancy] | ||||||
Chill requirement | ||||||
Deg_day_bud break chill req sat | ||||||
Deg day bud break chill req not sat | ||||||
[Root] | ||||||
Root_sensitivity_water_stress | 0.5 | 0.3 | 0.2 | 0.2 | 0.2 | 0.2 |
Max_root_depth | 1.5 | 1.2 | 1.8 | 1.8 | 1.5 | 1.5 |
Root_length_at_emergence (cm) | 15 | 12 | 12 | 12 | 12 | 12 |
Sow_depth | 0.08 | 0.08 | 0.08 | 0.08 | 0.08 | |
Root_density_distribution_curvature | 0.00001 | 0.00001 | 0.00001 | 0.00001 | 0.00001 | 0.001 |
[Morphology] | ||||||
Max_canopy_height | 0.5 | 0.5 | 2.2 | 2.2 | 1 | 1 |
[Inactive_period] | ||||||
Consider_inactive_days | 7 | 7 | 7 | 7 | 7 | 7 |
Inducement_temperature | 5 | 5 | 5 | 5 | 5 | 5 |
Start_DOY | 330 | |||||
Minimum_duration | 90 | |||||
[Season] | ||||||
Start_DOY_WA | 80 | 135 | 135 | 105 | 280 | |
Start_DOY_ID | 80 | 135 | 135 | 105 | 280 | |
Start_DOY_OR | 80 | 135 | 135 | 105 | 280 | |
Duration_WA | 193 | 80 | 125 | 100 | 285 | |
Duration_OR | 193 | 80 | 125 | 100 | 285 | |
Duration_ID | 193 | 80 | 125 | 100 | 285 | |
[Harvest] | ||||||
Unstressed | 0.85 | 0.5 | 0.5 | 0.5 | 0.45 | 0.45 |
Translocation_max | 0.45 | 0.45 | 0.35 | 0.4 | ||
[Fruit] | ||||||
Grape | ||||||
Fract_tot_solids | ||||||
Max_fruit_load | ||||||
[CO2] | ||||||
Growth_ratio | 1.2 | 1.2 | 1.05 | 1.05 | 1.2 | 1.2 |
Elevated_reference_conc | 600 | 600 | 600 | 600 | 600 | 600 |
Baseline_reference_conc | 360 | 360 | 360 | 360 | 360 | 360 |
- Note. There are 26 sets of crop parameters that are applied to complete list of crop types in the region. If the crop parameters for a particular crop type in the USDA CDL or WSDA ALL are not available, the parameters corresponding to the closest crop from this list are used. This is the fourth of set of four tables with parameter values.
Appendix C
Figures C1–C3 and Tables C1–C4 have crop evaluation details.
State | County | Commodity | Calibration ratio | Mean of average annual yields | ||
---|---|---|---|---|---|---|
Simulated (1) | NASS observed (2) | % Difference (2 – 1/2) × 100 | ||||
Idaho | Ada | Corn | 0.70 | 8.74 | 8.52 | −2.63 |
Idaho | Canyon | Corn | 1.00 | 12.08 | 11.89 | −1.57 |
Idaho | Gem | Corn | NA | 12.04 | 11.80 | −2.07 |
Idaho | Gooding | Corn | 0.71 | 8.67 | 8.50 | −2.04 |
Idaho | Jerome | Corn | 1.00 | 12.22 | 11.91 | −2.62 |
Idaho | Owyhee | Corn | NA | 12.03 | 11.86 | −1.44 |
Idaho | Payette | Corn | 0.74 | 9.04 | 8.90 | −1.64 |
Idaho | Twin Falls | Corn | 1.00 | 11.93 | 11.59 | −3.00 |
Oregon | Malheur | Corn | 1.00 | 11.85 | 11.72 | −1.12 |
Oregon | Morrow | Corn | NA | 12.18 | 12.17 | −0.07 |
Oregon | Umatilla | Corn | 1.00 | 12.16 | 11.93 | −1.89 |
Washington | Grant | Corn | 1.00 | 11.57 | 11.44 | −1.20 |
Washington | Yakima | Corn | 1.00 | 12.49 | 12.36 | −1.00 |
Idaho | Bannock | Potatoes | 0.53 | 9.28 | 9.10 | −2.01 |
Idaho | Bingham | Potatoes | NA | 16.62 | 16.32 | −1.82 |
Idaho | Bonneville | Potatoes | NA | 16.24 | 15.89 | −2.17 |
Idaho | Canyon | Potatoes | 1.00 | 16.48 | 16.27 | −1.30 |
Idaho | Caribou | Potatoes | 0.73 | 12.02 | 12.01 | −0.04 |
Idaho | Cassia | Potatoes | 0.42 | 7.76 | 7.55 | −2.71 |
Idaho | Elmore | Potatoes | 0.37 | 6.39 | 6.33 | −0.86 |
Idaho | Fremont | Potatoes | 0.77 | 11.81 | 11.49 | −2.78 |
Idaho | Gooding | Potatoes | 1.00 | 16.20 | 15.91 | −1.82 |
Idaho | Jefferson | Potatoes | 0.54 | 15.90 | 15.46 | −2.82 |
Idaho | Jerome | Potatoes | 0.53 | 9.20 | 8.88 | −3.54 |
Idaho | Madison | Potatoes | 0.58 | 9.22 | 8.99 | −2.59 |
Idaho | Minidoka | Potatoes | 0.47 | 8.39 | 8.15 | −3.04 |
Idaho | Owyhee | Potatoes | NA | 16.22 | 16.08 | −0.86 |
Idaho | Power | Potatoes | 0.68 | 11.60 | 11.40 | −1.75 |
Idaho | Teton | Potatoes | 1.00 | 15.67 | 15.32 | −2.26 |
Idaho | Twin Falls | Potatoes | 0.35 | 6.20 | 6.04 | −2.54 |
Washington | Adams | Potatoes | 1.00 | 15.15 | 14.93 | −1.50 |
Washington | Franklin | Potatoes | 1.00 | 16.40 | 16.29 | −0.67 |
Washington | Grant | Potatoes | 1.00 | 15.35 | 15.22 | −0.86 |
Washington | Skagit | Potatoes | NA | 17.83 | 18.08 | 1.35 |
Washington | Yakima | Wheat, irrigated | NA | 3.87 | 3.80 | −1.70 |
- Note. Crop and county combinations with at least 30 years of NASS observed yield estimates are used. The validation time period is 1970–1995. The calibration time period is 1996–2006. NASS observed yields are converted to dry matter yield (for comparison with modeled yields which are dry yields) by assuming a moisture content of 15% in corn and wheat, and 20% in potatoes. The calibration ratio is the calibrated adjustment factor applied to modeled yields in the county (for a particular crop) to account for spatial differences in yields.
State | County | Commodity | Calibration ratio | Standard deviation of average annual yields | ||
---|---|---|---|---|---|---|
Simulated (3) | NASS observed (4) | Observed/simulated | ||||
Idaho | Ada | Corn | 0.70 | 0.63 | 0.48 | 0.76 |
Idaho | Canyon | Corn | 1.00 | 0.46 | 0.67 | 1.47 |
Idaho | Gem | Corn | NA | 0.69 | 0.68 | 0.99 |
Idaho | Gooding | Corn | 0.71 | 0.67 | 0.61 | 0.91 |
Idaho | Jerome | Corn | 1.00 | 0.93 | 0.75 | 0.81 |
Idaho | Owyhee | Corn | NA | 0.58 | 0.70 | 1.22 |
Idaho | Payette | Corn | 0.74 | 0.90 | 0.58 | 0.64 |
Idaho | Twin Falls | Corn | 1.00 | 0.44 | 0.74 | 1.69 |
Oregon | Malheur | Corn | 1.00 | 0.90 | 0.68 | 0.76 |
Oregon | Morrow | Corn | NA | 1.11 | 0.29 | 0.26 |
Oregon | Umatilla | Corn | 1.00 | 0.95 | 0.34 | 0.36 |
Washington | Grant | Corn | 1.00 | 0.22 | 0.32 | 1.49 |
Washington | Yakima | Corn | 1.00 | 0.69 | 0.38 | 0.56 |
Idaho | Bannock | Potatoes | 0.53 | 0.36 | 0.69 | 1.90 |
Idaho | Bingham | Potatoes | NA | 0.38 | 1.26 | 3.33 |
Idaho | Bonneville | Potatoes | NA | 0.31 | 1.21 | 3.98 |
Idaho | Canyon | Potatoes | 1.00 | 0.51 | 0.89 | 1.74 |
Idaho | Caribou | Potatoes | 0.73 | 0.48 | 0.85 | 1.75 |
Idaho | Cassia | Potatoes | 0.42 | 0.46 | 0.52 | 1.15 |
Idaho | Elmore | Potatoes | 0.37 | 0.55 | 0.45 | 0.82 |
Idaho | Fremont | Potatoes | 0.77 | 0.53 | 0.95 | 1.78 |
Idaho | Gooding | Potatoes | 1.00 | 0.63 | 1.00 | 1.58 |
Idaho | Jefferson | Potatoes | 0.54 | 0.52 | 1.16 | 2.22 |
Idaho | Jerome | Potatoes | 0.53 | 0.53 | 0.54 | 1.01 |
Idaho | Madison | Potatoes | 0.58 | 0.31 | 0.70 | 2.21 |
Idaho | Minidoka | Potatoes | 0.47 | 0.39 | 0.55 | 1.41 |
Idaho | Owyhee | Potatoes | NA | 0.60 | 1.02 | 1.70 |
Idaho | Power | Potatoes | 0.68 | 0.33 | 0.91 | 2.74 |
Idaho | Teton | Potatoes | 1.00 | 0.41 | 1.29 | 3.12 |
Idaho | Twin Falls | Potatoes | 0.35 | 0.43 | 0.36 | 0.84 |
Washington | Adams | Potatoes | 1.00 | 0.76 | 0.60 | 0.79 |
Washington | Franklin | Potatoes | 1.00 | 0.51 | 0.72 | 1.41 |
Washington | Grant | Potatoes | 1.00 | 0.54 | 0.58 | 1.07 |
Washington | Skagit | Potatoes | NA | 0.70 | 0.46 | 0.66 |
Washington | Yakima | Wheat, irrigated | NA | 0.48 | 0.15 | 0.32 |
- Note. Crop and county combinations with at least 30 years of NASS observed yield estimates are used. The validation time period is 1970–1995. The calibration time period is 1996–2006. The calibration ratio is the calibrated adjustment factor applied to modeled yields in the county (for a particular crop) to account for spatial differences in yields.
State | County | Commodity | Yield time series Nash Sutcliffe Efficiency (evaluation period: 1970–1995) |
---|---|---|---|
Idaho | Ada | Corn | −0.71 |
Idaho | Canyon | Corn | −2.21 |
Idaho | Gem | Corn | −0.91 |
Idaho | Gooding | Corn | −0.64 |
Idaho | Jerome | Corn | −0.33 |
Idaho | Owyhee | Corn | −0.86 |
Idaho | Payette | Corn | −0.25 |
Idaho | Twin Falls | Corn | −2.15 |
Oregon | Malheur | Corn | −0.42 |
Oregon | Morrow | Corn | 0.16 |
Oregon | Umatilla | Corn | −0.14 |
Washington | Grant | Corn | −4.02 |
Washington | Yakima | Corn | 0.18 |
Idaho | Bannock | Potatoes | −4.50 |
Idaho | Bingham | Potatoes | −9.54 |
Idaho | Bonneville | Potatoes | −16.17 |
Idaho | Canyon | Potatoes | −3.06 |
Idaho | Caribou | Potatoes | −2.95 |
Idaho | Cassia | Potatoes | −1.67 |
Idaho | Elmore | Potatoes | −0.78 |
Idaho | Fremont | Potatoes | −3.55 |
Idaho | Gooding | Potatoes | −2.35 |
Idaho | Jefferson | Potatoes | −6.65 |
Idaho | Jerome | Potatoes | −1.93 |
Idaho | Madison | Potatoes | −6.27 |
Idaho | Minidoka | Potatoes | −2.07 |
Idaho | Owyhee | Potatoes | −2.72 |
Idaho | Power | Potatoes | −6.76 |
Idaho | Teton | Potatoes | −10.63 |
Idaho | Twin Falls | Potatoes | −0.77 |
Washington | Adams | Potatoes | −1.41 |
Washington | Franklin | Potatoes | −2.73 |
Washington | Grant | Potatoes | −1.91 |
Washington | Skagit | Potatoes | −0.56 |
Washington | Yakima | Wheat, irrigated | 0.04 |
- Note. This is calculated for crop and county combinations with at least 30 years of NASS observed yield estimates. The evaluation time period is 1970–1995. The calibration time period is 1996–2006.
State | County | Commodity | Calibration ratio applied | Spatial variation in average annual yields | ||
---|---|---|---|---|---|---|
Simulated validation (5) | NASS observed (6) | % Difference (6 – 5/5) × 100 | ||||
Idaho | Ada | Corn | 0.70 | 0.70 | 0.69 | −1.61 |
Idaho | Canyon | Corn | 1.00 | 0.97 | 0.96 | −0.56 |
Idaho | Gem | Corn | NA | 0.96 | 0.95 | −1.06 |
Idaho | Gooding | Corn | 0.71 | 0.69 | 0.69 | −1.03 |
Idaho | Jerome | Corn | 1.00 | 0.98 | 0.96 | −1.60 |
Idaho | Owyhee | Corn | NA | 0.96 | 0.96 | −0.44 |
Idaho | Payette | Corn | 0.74 | 0.72 | 0.72 | −0.63 |
Idaho | Twin Falls | Corn | 1.00 | 0.96 | 0.94 | −1.98 |
Oregon | Malheur | Corn | 1.00 | 0.95 | 0.95 | −0.12 |
Oregon | Morrow | Corn | NA | 0.97 | 0.98 | 0.93 |
Oregon | Umatilla | Corn | 1.00 | 0.97 | 0.97 | −0.88 |
Washington | Grant | Corn | 1.00 | 0.93 | 0.92 | −0.20 |
Washington | Yakima | Corn | 1.00 | 1.00 | 1.00 | 0.00 |
Idaho | Bannock | Potatoes | 0.53 | 0.56 | 0.55 | −2.01 |
Idaho | Bingham | Potatoes | NA | 1.01 | 0.99 | −1.82 |
Idaho | Bonneville | Potatoes | NA | 0.99 | 0.96 | −2.17 |
Idaho | Canyon | Potatoes | 1.00 | 1.00 | 0.99 | −1.30 |
Idaho | Caribou | Potatoes | 0.73 | 0.73 | 0.73 | −0.04 |
Idaho | Cassia | Potatoes | 0.42 | 0.47 | 0.46 | −2.71 |
Idaho | Elmore | Potatoes | 0.37 | 0.39 | 0.38 | −0.86 |
Idaho | Fremont | Potatoes | 0.77 | 0.72 | 0.70 | −2.78 |
Idaho | Gooding | Potatoes | 1.00 | 0.98 | 0.97 | −1.82 |
Idaho | Jefferson | Potatoes | 0.54 | 0.96 | 0.94 | −2.82 |
Idaho | Jerome | Potatoes | 0.53 | 0.56 | 0.54 | −3.54 |
Idaho | Madison | Potatoes | 0.58 | 0.56 | 0.55 | −2.59 |
Idaho | Minidoka | Potatoes | 0.47 | 0.51 | 0.49 | −3.04 |
Idaho | Owyhee | Potatoes | NA | 0.98 | 0.98 | −0.86 |
Idaho | Power | Potatoes | 0.68 | 0.70 | 0.69 | −1.75 |
Idaho | Teton | Potatoes | 1.00 | 0.95 | 0.93 | −2.26 |
Idaho | Twin Falls | Potatoes | 0.35 | 0.38 | 0.37 | −2.54 |
Washington | Adams | Potatoes | 1.00 | 0.92 | 0.91 | −1.50 |
Washington | Franklin | Potatoes | 1.00 | 1.00 | 0.99 | −0.67 |
Washington | Grant | Potatoes | 1.00 | 0.93 | 0.92 | −0.86 |
Washington | Skagit | Potatoes | NA | 1.08 | 1.10 | 1.35 |
Washington | Yakima | Wheat, irr | NA |
- Note. Yields as a fraction of a reference county (in bold) are calculated for each crop. Crop and county combinations with at least 30 years of NASS observed yield estimates are used. The validation time period is 1970–1995. The calibration time period is 1996–2006. The calibration ratio is the calibrated adjustment factor applied to modeled yields in the county (for a particular crop) to account for spatial differences in yields.

Scatterplots of modeled yields versus observed yields (NASS statistics) by crop and county combinations are shown. This is calculated for crop and county combinations with at least 30 years of NASS observed yield estimates. The evaluation time period is 1970–1995. The calibration time period is 1996–2006. This is this is the first of a set of three plots that show this information.

Scatterplots of modeled yields versus observed yields (NASS statistics) by crop and county combinations are shown. This is calculated for crop and county combinations with at least 30 years of NASS observed yield estimates. The validation time period is 1970–1995. The calibration time period was 1996–2006. This is this is the second of a set of three plots that show this information.

Scatterplots of modeled yields versus observed yields (NASS statistics) by crop and county combinations are shown. This is calculated for crop and county combinations with at least 30 years of NASS observed yield estimates. The validation time period is 1970–1995. The calibration time period was 1996–2006. This is this is the third of a set of three plots that show this information.