Performance of an integrated modeling framework for spring wheat yield simulation in Saskatchewan, Canada

• A modeling framework combining AquaCrop-OS and SHAW is proposed. • Coupling SHAW improves AquaCrop-OS yield accuracy in drought years. • The improved model captures interannual variability and multi-year yield trends. Crop models depend on field experiments to provide a numerical representation of the complex interactions between crops, soil, and the atmosphere. However, when applied at a regional scale, these models require parameter recalibration to maintain their performance and accuracy. Here, we comprehensively evaluated the performance of a crop model, AquaCrop-OS, across the agricultural region of Saskatchewan from 1981 to 2016. The model was configured with 167 0.5° × 0.5° grids and initialized by daily meteorological data . To realistically reproduce soil moisture conditions in early spring, we coupled the SHAW (Simultaneous Heat and Water) model with AquaCrop-OS to physically simulate heat and water fluxes based on cultivation types, snow accumulation, and soil hydrological characteristics. The results show that the performance of AquaCrop-OS is improved significantly by coupling with SHAW. The degree of improvement varies depending on different climate conditions. In extreme drought years (Standardized Precipitation Evapotranspiration Index (SPEI)<-2), the normalized root-mean-square error (NRMSE) is reduced by 60 %. In moderate drought (-2<SPEI<-1), normal (-1<SPEI<1), and moderate wet (1<SPEI<2) years, the NRMSE is reduced by an average of 11.2 %. The coupled model captures the long-term area mean of spring wheat yield variability well, with the coefficient of determination (R 2 ) of 0.72 and the NRMSE of 11.16 %. The performance of AquaCrop-OS exhibits spatial variability across different regions, with the Pearson correlation coefficient (PCC) ranging from 0.1 to 0.95, while the NRMSE ranges from 9 % to 29 %. Most of the study area shows good performance, with an underestimation of yield in the northern region. This study has extended the application of AquaCrop-OS for spring wheat in western Canada to a regional scale. Our efforts have improved the regional-scale simulation of spring wheat yield under observed climatic, soil, and water conditions in the region, which is crucial for making seasonal projections and assessing the impact of climate change on spring wheat yield.