Evaluating Scenario Based Performance of DSSAT Response to Soil Depth, Initial Soil Water Content and Choice of Zea mays L. Cultivar Selection in Semi‐Arid North West Province in South Africa

ABSTRACT Process‐based crop models are widely used to assess crop responses to climate variability, yet their performance is highly sensitive to assumptions regarding soil properties, initial soil water content and cultivar selection, particularly in spatially heterogeneous, rainfed systems. This study evaluates the performance of the DSSAT‐CERES‐Maize model across the North West Province of South Africa using a fine‐scale, quinary catchment‐based framework. Four scenario simulations were developed to examine the influence of soil depth, pre‐season soil moisture and cultivar choice on simulated maize yields. Model outputs were evaluated against district‐level reported yields for the 1981–1999 period using a comprehensive multi‐criteria assessment framework incorporating distributional tests, correlation analysis, weighted regression and multiple performance metrics. Results indicate that DSSAT effectively reproduces inter‐annual yield variability across spatial scales, with stronger agreement at the district level than at the provincial scale. Scenario performance was highly sensitive to soil depth and initial soil water assumptions, with the scenario incorporating deeper effective rooting depth and intermediate pre‐season soil moisture consistently achieving superior agreement across most evaluation criteria. Cultivar selection influenced yield variability, highlighting the importance of representative genetic parameterisation in regional applications. While simulated and reported yield medians did not differ significantly at the district scale, error magnitudes and efficiency metrics varied spatially, reflecting the dominant influence of climate variability under rainfed conditions. These findings demonstrate that spatially explicit, scenario‐based evaluation enhances confidence in crop model applications and provides valuable insights for agrometeorological assessments, climate adaptation planning and decision support in semi‐arid, water‐limited agricultural systems.