The Clear Lake release schedule specifies how much water is available annually and monthly to the District during the peak agricultural season from April to September. The decree’s “Quantity” criteria sets allowable seasonal withdrawal limits based on April 1 water levels recorded at Rumsey, known as the Rumsey gauge. If the Rumsey gauge is at or above 7.54 feet, then 150,000 acre feet of water is available for the growing season from April 1 to October 31. Monthly percentages of the ASW are available for release each month. If Rumsey levels are below 3.22 feet, no water can be released that year apart from flood flows. For in‐between levels, ASW are set in the release schedule that increases to a maximum of 150,000 acre feet in what is known as the quantity criteria. As per these stipulations, the District did not make any releases in the severe drought of 1976–1977, as well as in 1990 at the end of several dry years. The Solano Decree also stipulates “Stage criteria” that set limits to draw down, posing an additional constraint to the District’s withdrawal of water in any given month. Clear Lake releases in the winter are also controlled by the 1920 Gopcevic Decree for flood control operation. The highly controlled nature of this lake can be attested by the historical monthly average lake levels which have varied only 5.7 ft on average within a water year, with a maximum range of 10.9 ft and a minimum of only 2.3 ft.The Cache Creek model, run at a monthly time step, uses climate and land cover information to simulate the water balance. It uses the results to simulate the management of Clear Lake and Indian Valley Reservoirs and water supply for irrigation downstream. The model simulates irrigation demand for 20 crop types within Yolo County, grow bags garden which is met through surface and groundwater sources .
The model was calibrated to a historical run from 1971–2000, which formed the baseline scenario. The calibrated model was then run under various combinations of climate and agricultural land use projections as described below. Figure 3.1 shows the study area along with the spatial discretization of the model. The spatial domain of the model covers 5027 square kilometers and includes the Cache Creek watershed up to Capay , and all of Yolo County. The focus of the irrigation water demand and supply analysis is on the District service area , although the model can also simulate irrigation demand for the rest of the county. Table 3.1summarizes each catchment’s characteristics. A water balance simulated for each catchment. Spatial data on elevation, watersheds, and land use were acquired and used to define and characterize each catchment. Elevation data were extracted from the Digital Elevation Model provided by the U.S. Geological Survey. Land cover information was assembled from two sources. For the non‐agricultural landscape, the National Landcover Data Set was used . For the agricultural areas, county reports and DWR Land Use Surveys were used . Upstream catchments were aggregated from the DWR watersheds layer. This aggregation was based on climate considerations, the locations of major infrastructure , in‐stream flow requirements, and flow gauges. Parameters of the rainfall‐runoff module were calibrated against the longest available continuous data from gauges in unimpaired watersheds. These were at Kelsey Creek and Hough Springs on the north fork of Cache Creek , in the headwaters of Clear Lake and Indian Valley, respectively. Goodness of fit metrics were computed for each set of simulated and observed hydrographs. Two groundwater objects were defined and conceptually aligned to the groundwater sub‐ basins delineated by DWR: one below Capay Valley receiving recharge as infiltration from the Capay Valley catchment, the other below the Yolo Valley floor, receiving recharge from the catchments downstream of Capay.
Our model’s treatment of groundwater is similar to the Central Valley application . It is capable of relative comparison among scenarios of groundwater recharge and extraction volumes, but not of simulating absolute groundwater depths. The model simulates the operations of Clear Lake, Indian Valley, and the water delivery through canals. Detailed description of how WEAP simulates reservoir releases through conservation storage and flood rules is available in Yates et al. . Reservoir physical characteristics were obtained from California Department of Water Resouces California Data Exchange Center and the District. Indian Valley operating rules were obtained from the District. Clear Lake operating rules were obtained from the District, and from documentation of the Solano and Gopcevic Decrees described earlier . Details, including the stepwise procedure on implementing the Solano Decree, are available in public documents and through the District. Clear Lake releases during the wet season are controlled by the Gopcevic Decree, for which target storage levels come into play from January to March. These target storages were set as WEAP’s “Top of Conservation” in the model’s Clear Lake reservoir object. The second operating constraint, also from the Solano Decree, is its stage limitation criteria. These criteria were programmed and set as “Top of Buffer” in the reservoir object. The third constraintis the hydraulic capacity of Clear Lake’s outlet channel. Hydraulic capacity varies by the stage; data obtained from the District was used to develop a hydraulic capacity constraint as a function of stage. This expression was set as a hydraulic constraint on the releases from Clear Lake in the model. Outlet flows were then constrained to be a minimum of the hydraulic capacity constraint, and the allowable monthly withdrawal as determined by the Solano decree’s Quantitative criteria—the latter also entirely encoded within WEAP. Clear Lake does not provide carryover storage for irrigation demand. Although Indian Valley does provide carryover storage, typically it is operated with no carryover storage . In general, the District attempts to utilize all its Clear Lake allocation each year. This means that Clear Lake usage is prioritized over Indian Valley as much as possible. In the model’s setting of supply priorities, this translates to a lower filling priority for Clear Lake over Indian Valley. Simulation of reservoir operations was verified by comparing simulated versus observed reservoir levels.
The District’s main conveyance is in the form of 175 miles of mostly unlined canals and arterial ditches that run off the West Adams and Winters Canals from Capay Diversion Dam on Cache Creek. In the model, these conveyances are aggregated into a single transmission link object, with capacity set to the total distribution’s capacity of 750 cubic feet per second , and with an estimated leakage of 40 percent of conveyance flows obtained from calibration attempts and informed by District estimates of mass balances . Seventeen crop categories were modeled for the catchments dominated by agriculture. Table 3.3 lists the different crop categories considered along with county‐wide acreages from four selected years. The crop categories are informed by DWR’s irrigated crop acres and water use portfolio,grow bag for tomato taking into consideration both the crop categories and corresponding acreages available through the county reports as well as estimates of the District scale cropping pattern. An annual time series of total irrigated acreage and irrigated crop areas was assembled at the county level . Individual crop acreages were spatially distributed among the four agricultural catchments using GIS datasets available for 1989 and 1997 through the DWR Land Use Surveys . This allowed a cropping pattern to be represented in the model for the historical period for each agricultural catchment. Each crop’s irrigation water needs were simulated using crop‐specific crop coefficients, irrigation schedules, and irrigation thresholds. Crop‐specific parameters pertaining to irrigation were adapted from the Central Valley application by Joyce et al. , who calibrated the crop and irrigation parameters at the spatial scale of the DWR Planning Area level against four annual estimates of applied water published by DWR for 1998,1999, 2000, and 2001 . In our model, we also used DWR portfolio data available for the same years, but at a finer spatial level—the Detailed Analysis Unit . The irrigation threshold parameter in WEAP was calibrated for each crop to match DWR’s applied water estimates for 1998, 1999, 2000, and 2001 for the DWR’s Lower Cache Creek DAU which closely follows the county boundaries. Figure 3.2 presents the calibrated irrigation schedules and thresholds for each crop. The model’s estimation of water demand represents a departure from the operations of the District. The District solicits water demands from its customers every year in March, and then decides by April how much total quantity will be available. This decision is based on water levels in the two reservoirs and a projection of the season ahead. Since our goal was to look to the future, we used a simulation approach instead of hard‐coding the historical demand based on the District’s historical roster. The latter would not have provided us the means of projecting demand into the future.Yolo County based on the relationship between historical crop acreage, a set of economic variables , and climate variables . To forecast cropping area from the present to 2050, climatic variables were calculated from daily climate projections for the A2 and B1 scenarios generated by the GFDL climate model described above. The second land use projection was based on a hypothetical scenario envisioning an agricultural landscape which adapts to climate change in two ways: by allocating a smaller fraction of land to crops that require large amounts water; and by increasing crop diversity. For example, the acreage of rice, alfalfa, and other water intensive field crops were gradually reduced to the lows observed during a period of severe drought in the mid‐1970s . Likewise, an increase in crop diversity over time was simulated by progressively allocating a larger fraction of land to vineyards, winter grains, almonds, deciduous orchards, subtropical orchards, tomatoes, cucurbits, and truck crops . Since this crop diversification projection is a hypothetical construct, rather than a statistically derived forecast, a future time frame of 2009–2099 was used.
It should also be noted that this approach assumes gradual changes in crop acreage and did not attempt to capture the year to year variability reflected in the historic record. Statewide there has been a notable shift in irrigation methods from surface water applied using flood or furrow irrigation towards low‐volume sprinkler and drip irrigation, particularly for vegetable crops, orchards, and vineyards . These methods can potentially reduce soil evaporation and applied water . Furthermore, a recent survey of grower perspectives on water scarcity and climate change in Yolo County indicates a strong inclination to expand their use of drip and low‐volume irrigation among local farmers . Likewise, incentive programs to promote adoption of improved irrigation technology are seen as a politically feasible water demand management strategy. However, one criticism is that, in some watersheds, such policies have failed to curtail groundwater extraction as some farmers use the “water savings” to expand irrigated acreage or grow more water‐ intensive crops . As such, we included a conceptual scenario which assumes that irrigation technology and efficiency will continue to improve in coming decades but overall irrigated acreage in the district will not. We reflect these trends in the model, by decreasing the irrigation threshold parameter, in a manner similar to the work of Joyce et al. and Purkey et al. . Beginning in 2010, irrigation thresholds for each crop, except for wine grapes, winter grains, and safflower, were assumed to decrease linearly so that by 2099 they reached 70 percent of the historic reference threshold. For the latter crops, no change in water‐saving irrigation technologies was assumed because vineyards are already on drip irrigation, winter grains are mostly supplied by rain and stored soil water, and safflower is already a low water consuming crop.Another measure of water shortage is the frequency of years receiving no water allocation from Clear Lake. For example, if the Rumsey gauge is below 3.22 feet, the initial ASW assessment is for no allocation of water that year. During the historical period the model predicted 6 such years . Model projections for the climate only scenario suggest that the number of years receiving no allocation will increase gradually with time, particularly during the latter half of the century. In the far term under A2, reservoir inflows are very low in some years in response to the warmer and drier conditions.