Testing the Performance of the Apsim Model in Iowa.

In the US Midwest, the Agricultural Production Systems sIMulator (APSIM) has been used to analyze yield variability caused by a number of factors, however, evaluation of the APSIM soil modules that provide constraints to crop growth is still lacking. In this study we provide a parallel assessment of both crop and soil modules, and in particular we evaluated the following modules: maize, soil water, soil N, residue, manure and soil temperature. Published data were used for model calibration and evaluation. After appropriate calibration of key model parameters, the model captured sufficiently well dynamics of soil water (root mean square error, RMSE=0.032 mm mm^-1, R²=0.69), soil temperature (RMSE=2.1 °C, R²=0.89), soil inorganic N dynamics (RMSE=12.6 kg N ha^-1, R²=0.59), contrasting soil net N mineralization patterns under fresh and compost swine manure applications (RMSE=13.6 kg N ha^-1, R²=0.88), crop phenology (RMSE=1.52 days, R²=0.99), leaf area index (RMSE=0.60 m²m^-2, R²=0.90), leaf N concentration (RMSE=0.28 g 100g^-1, R²=0.90), canopy N uptake (RMSE= 9.0 kg N ha^-1, R²=0.84), biomass production (RMSE=0.77 Mg ha^-1, R²=0.98) and grain yield (RMSE=0.53 Mg ha^-1; R²=0.84). The calibrated model was evaluated against independent data on grain yield (RMSE=0.65 Mg ha^-1, R²=0.88) and biomass production (RMSE = 1.1 Mg ha^-1; R²=0.97), showing very acceptable performance, especially in addressing yield-N relationships. It was concluded that APSIM can be used with confidence in Iowa to assist multiple productivity tasks.