Watershed scale evaluation of an improved SWAT auto-irrigation function

Abstract The SWAT model is a well-documented hydrologic model. However, some studies report that the existing SWAT auto-irrigation methods are unable to represent actual irrigation management, particularly in intensively irrigated regions. In the U.S. Great Plains, the SWAT model does not reproduce the management allowed depletion (MAD) irrigation scheduling commonly used by researchers and producers. To this purpose, the SWAT source code has been modified to include the MAD auto-irrigation function. This study evaluated the performance of the soil water content (SWC), corrected soil water content (CSWC), plant water demand (PWD), and MAD auto-irrigation methods in streamflow and irrigation simulations by comparison with observed data. The CSWC and MAD methods performed the best in streamflow simulations with NSE > 0.75 and PBIAS within ±11%. Comparisons of simulated irrigation with the field irrigation also indicated the CSWC and MAD methods outperformed other methods with the NSE > 0.75 and PBIAS within ±5%.