Understanding surface-water availability in the Central Valley as a means to projecting future groundwater storage with climate variability

California's Central Valley (CV) relies heavily on diverted surface water and groundwater pumping to supply irrigated agriculture. However, understanding the spatiotemporal character of water availability in the CV is difficult because of the number of individual farms and local, state, and federal agencies involved in using and managing water. Here we use the Central Valley Hydrologic Model (CVHM), developed by the USGS, to understand the relationships between climatic variability, surface water inputs, and resulting groundwater use over the historical period 1970-2013. We analyzed monthly surface water diversion data from >500 CV locations. Principle components analyses were applied to drivers constructed from meteorological data, surface reservoir storage, ET, land use cover, and upstream inflows, to feed multiple regressions and identify factors most important in predicting surface water diversions. Two thirds of the diversion locations ( 80% of total diverted water) can be predicted to within 15%. Along with monthly inputs, representations of cumulative precipitation over the previous 3 to 36 months can explain an additional 10% of variance, depending on location, compared to results that excluded this information. Diversions in the southern CV are highly sensitive to inter-annual variability in precipitation (R2 = 0.8), whereby more surface water is used during wet years. Until recently, this was not the case in the northern and mid-CV, where diversions were relatively constant annually, suggesting relative insensitivity to drought. In contrast, this has important implications for drought response in southern regions (eg. Tulare Basin) where extended dry conditions can severely limit surface water supplies and lead to excess groundwater pumping, storage loss, and subsidence. In addition to fueling our understanding of spatiotemporal variability in diversions, our ability to predict these water balance components allows us to update CVHM predictions before surface water data are compiled. We can then develop groundwater pumping and storage predictions in real time, and make them available to water managers. In addition, we are working toward future projections by coupling the regional CVHM to downscaled GCM output to assess future scenarios of water availability in this critical region.