Enhancing spring wheat growth simulation and yield estimation in arid regions: A SWAP–IES optimization approach

Abstract Accurate simulation of the crop growth process was the foundation for the development of smart agriculture. However, the uncertainty of crop growth models limits their practical application. This study integrates the Soil Water Atmosphere Plant (SWAP) model with the Iterative Ensemble Smoother (IES) algorithm to develop the SWAP–IES optimization approach and explores various uncertainty factors of the system, including the ensemble size, observational errors setting, combination of observation variables and their corresponding observation stages, and uncertain parameters selection. The results suggested that, under water stress conditions, an ensemble size of 50 was recommended. It was advisable to choose leaf area index (LAI) and soil moisture content (SW) as observation variables, focusing on monitoring data from the flowering to the milk stage. The suitable observational error settings for LAI and SW were 0.3–0.5 m 2 m − 2 and 0.03–0.05 cm 3 cm − 3 , respectively. For uncertain parameters, it was recommended to select the five crop parameters (RGRLAI, SPAN, CVO, EFF, and CVL) and three soil parameters ( θ s , K s , and n ) for simulation. The SWAP‐IES, validated with 2020 and 2021 spring wheat ( Triticum aestivum L.) experiments, demonstrated high accuracy in simulating yields, with root mean square error values of 0.56 and 0.61 t ha −1 , respectively. The SWAP–IES optimization approach could significantly reduce the uncertainty in the simulation process and improve simulation accuracy by optimizing the system settings strategy.