Evaluating the performance of DRAINMOD using soil hydraulic parameters derived by various methods

The objective of this study was to determine which level of soil information would be sufficient to use with DRAINMOD in predicting subsurface drainage. The model was evaluated by predicting and comparing with observed monthly and annual subsurface drainage for 14 years (1990-2003) from Webster silty clay loam soil experimental plots located near Gilmore City, in Pocahontas County, Iowa, USA. Three groups of input soil hydraulic parameters were obtained by: (1) determining the soil texture and bulk density (BD) from the ISPAID7.1 Soil Survey Database, then inputting them into a pedotransfer function model (ROSETTA) to determine soil hydraulic parameters (denoted as SP_1); (2) analyzing the soil texture and organic matter (OM) content in laboratory and deriving the BD, field capacity (θ−33 kPa) and wilting point (θ−1500 kPa) from literature, then inputting them into ROSETTA to determine soil hydraulic parameters (SP_2); and (3) calibrated soil hydraulic parameters based on initial inputs from the Soil Survey Database plus ROSETTA (SP_3). All methods resulted in an acceptable level of accuracy in predicting monthly and annual subsurface drainage volume from the study site and the differences in model performance using all these methods were subtle. This suggests that ROSETTA in combination with the Soil Survey Database offers a quick and easy way to derive the soil hydraulic parameters for running DRAINMOD to simulate subsurface drainage systems in Iowa's subsurface drained landscape where site-specific soil hydraulic properties may not be available. This warrants further evaluation of the model performance at other sites to support the findings in this study.