Ecohydrologic Processes of Irrigated Agriculture Control Lake-Groundwater Interactions in a Highly Managed Watershed

Abstract:

Understanding ecohydrologic processes of irrigated agriculture, the world's largest consumer of freshwater, and its environmental impacts is important as droughts continue to affect agroecosystems. Coevolving feedbacks between food production, hydroclimatic variability, and changes in water and agricultural management practices in agroecosystems impact surface water-groundwater interactions, making hydrologic predictions challenging. The Salton Sea watershed in southern California and northern Mexico is one of the most productive agricultural regions in the United States and encompasses the Salton Sea, the largest inland terminal lake in California. Lake-groundwater interactions in this watershed are tightly coupled to the imported Colorado River inflows which indirectly provide about 1.4 km³/yr of water to the lake via irrigated agricultural runoff and drainage. While expansion of irrigated agriculture during the 1950s raised lake water levels, a significant declining trend in the lake level is observed in the last two decades without substantial changes in the lake inflows. To understand ecohydrological processes of the Salton Sea watershed, we set-up a semi-distributed hydrologic model over an area of 21,801 km² using the Soil and Water Assessment Tool (SWAT). For model calibration, a two-step strategy is utilized: an automatic calibration approach at a monthly scale for the headwater sub-basins, and a manual calibration for the agricultural sub-basins. Results indicate variable performance metrics across multiple headwater sub-basins with NSE values of 0.3-0.8 and 0.4-0.7 during calibration and validation, respectively. This variable performance is a result of ephemeral flows in this dry watershed, and lack of high resolution precipitation data in the headwater sub-basins. Simulated evaporation ratio in the natural and irrigated catchments are 53% and 75% of total water input, respectively. Because agricultural areas have not changed significantly since 2000, changes in subsurface flow processes caused by groundwater pumping in the northern part of the watershed and improving irrigation efficiency might explain decline in the lake water levels. Future work will focus on setting up a coupled SWAT-MODFLOW model to capture subsurface flow dynamics and lake-groundwater interaction.