Estimation of groundwater pumping as closure to the water balance of a semi-arid, irrigated agricultural basin

Groundwater pumping is frequently the least measured water balance component in semi-arid basins with significant agricultural production. In this article, we develop a GIS-based water balance model for estimating basin-scale monthly and annual groundwater pumping and apply it to a 2300 km2 semi-arid, irrigated agricultural area in the southern San Joaquin Valley, California. Both, annual groundwater storage changes and pumping are estimated as closure terms. The local hydrology is dominated by distributed surface water supplies, limited precipitation, and large crop water uses; whereas basin-scale runoff generation and groundwater-to-surface water discharges are negligible. Groundwater represents a terminal long-term storage reservoir with distributed inputs and outputs. To capture the spatio-temporal variability in water management and water use, the study area is delineated into 26 water service areas and 9611 individual fields or land units. The model computes conveyance seepage losses external to districts; seepage losses within districts; and net applied surface water of each district. For each land unit, the model calculates the applied water demand; its allotment of delivered surface water; the groundwater pumping required to meet the balance of its applied water demand; and aquifer recharge resulting from deep percolation of applied water and precipitation. These spatially distributed components are aggregated to the basin scale. Estimated annual groundwater storage changes compared well to those computed by the water-table fluctuation method over the 30-year study period, providing an independent verification of the consumptive use estimation. Pumping accounted for as much as 80% of the total applied water in ‘Critical’ water years and as little as 30% in ‘Wet’ years. Pumping estimates are most sensitive to estimation uncertainty of soil available water. They show little sensitivity to estimation errors in effective root depth, irrigation efficiencies, and intra-district seepage losses, although the cumulative sensitivity is significant.