Evaluation of Soil Water Pedotransfer Functions for Use in the Apsim Biochar Model.

Accurate estimates of soil water parameters are needed for reliable yield predictions from agricultural crop models. Biochar, a soil amendment, is known to affect soil physical properties such as bulk density and soil water retention. A biochar model was recently developed within the Agricultural Production Systems sIMulator (APSIM) cropping systems model to predict the impacts of biochar on agro-ecosystems. A modified version of the Saxton and Rawls pedotransfer functions, which include quality modifiers that account for different biochar types, are currently used in the APSIM biochar model to estimate soil water parameters. However, these equations were developed after conducting a literature review and were never tested and validated using experimental data. The objectives of this study were, 1) to compare and evaluate multiple pedotransfer functions for estimating soil water parameters in soils with and without biochar, and 2) to calibrate APSIM biochar parameters and improve accuracy of simulated biochar impacts on soil water estimates. Forty-eight soil cores to a depth of 1.2m were collected from five climatically distinct regions across the state of Iowa. Each core was sectioned into five depth increments and analyzed for OM content, texture, and water retention parameters. Results show that for soils without biochar the Saxton and Rawls equations provided the best estimates for saturation point and field capacity, while the Web Soil Survey database most accurately estimated permanent wilting point and plant available water. For soils with biochar, the Web Soil Survey database estimates for saturation point, field capacity, permanent wilting point, and plant available water were more accurate than the estimates from the other soil water equations. During biochar model calibration we found that the quality modifiers are site-specific and local calibration is required to accurately predict the impacts of biochar on soil water parameters.