Field Validation and Comparison of LEACHM and NCSWAP Models for Predicting Nitrate Leaching

The abilities of the LEACHM and NCSWAP models to simulate nitrate leaching were compared using field data collected from a three-year nitrate leaching experiment conducted in central Pennsylvania on Hagerstown silt loam soil (fine, mixed, mesic, Typic Hapludalf). Nitrate leaching losses below the 1.2-m depth from N-fertilized and manured corn were measured with zero-tension pan lysimeters. Four nitrogen and manure treatments were modeled for the growing seasons of 1988, 1989, and 1990 using the LEACHM and NCSWAP models. The cumulative simulations were then compared with the cumulative pan efficiency corrected measured data for these three years. Both models were calibrated to the site conditions using the growing season data of 1989. After the models were calibrated for the 1989 year, they were evaluated using 1988 and 1990 nitrate leaching data. Simulated results for the calibration year for both models were reasonably accurate. Statistical criteria were established from the calibration data set (1989) to evaluate the simulations from both models for the two validation years (1988 and 1990). Based on this statistical criteria, both models generally did not successfully predict nitrate leaching below the 1.2-m depth for most of the treatments for the validation years. Much of the simulation error seemed to be related to the inability of both models to simulate the macropore influenced water flow in the well-structured soil and/or the sub-model controlling soil nitrogen rate constants. The overall performance of both models was compared and it was concluded that the LEACHM model (Md = 0.38 kg ha–1) statistically performed better than the NCSWAP model (Md = –3.44 kg ha–1) in simulating nitrate leaching.