Predicting the Impacts of Climate Change on Irrigation Demand and Erosion Processes in Agricultural Watersheds

Climatic conditions play a major role in physical processes impacting soil and agrochemical detachment and transportation from/in agricultural watersheds. In addition, these climatic conditions are projected to significantly vary spatially and temporally in the 21st century, leading to vast uncertainties about the future of soil erosion and transport and non-point source pollution in agricultural watersheds. In this study, we selected the sunflower basin in the lower Mississippi River basin, USA to contribute in the understanding of how climate change affects watershed processes and the transport of pollutant loads. The watershed agricultural lands cover more than 94% of the total area, given the prolonged growing season in comparison to the Midwest, access to the Mississippi River Valley alluvial aquifer for irrigation, and fertile flood-plain soils. To achieve our goal, topographic, climate, soil, farming management, land cover/land use datasets were collected and the AnnAGNPS model databases were generated, and calibrated to generate a simulation of the watershed conditions from 2000 until 2015. The calibrated model Nash-Sutcliffe efficiency coefficient is 0.61 in comparison to -0.77 before calibration. Then climate projections were retrieved from the archive of World Climate Research Programme's (WCRP) Coupled Model Intercomparison Phase 5 (CMIP5) project. Four models were selected for each of the four RCPs (2.6, 4.5, 6.0 and 8.5): Miroc 5.1, GLDL-esm2g.1, CCSM4.1 and GFDL-cm3.1 for representing four various ranges of temperature and precipitation and covering the equilibrium climate response and transient response spectrum. Statistics derived from downscaled GCM output representing the 1981-2010 (base period), 2011-2040, 2041-2070 and 2071-2100 time periods were used to generate maximum/minimum temperature and precipitation on a daily time step using the USDA Synthetic Weather Generator, SYNTOR. The downscaled climate data were then utilized as inputs to run in AnnAGNPS. A total of 64 AnnAGNPS models with 100 realizations for each model were simulated. This study is a step toward proactive management of water and sediment resources in agricultural watersheds under a changing climate.