Process Importance Identification for the SPAC System Under Different Water Conditions: A Case Study of Winter Wheat

Modeling the soil–plant–atmosphere continuum (SPAC) system requires multiple subprocesses and numerous parameters. Sensitivity analysis is effective to identify important model components and improve the modeling efficiency. However, most sensitivity analyses for SPAC models focus on parameter-level assessment, providing limited insights into process-level importance. To address this gap, this study proposes a process sensitivity analysis method that integrates the Bayesian network with variance-based sensitivity measures. Four subprocesses are demarcated based on the physical relationships between model components revealed by the network. Applied to a winter wheat SPAC system under different water conditions, the method effectively and reliably identifies critical processes. The results indicate that, under minimal water stress, the subprocesses of photosynthesis and dry matter partitioning primarily determine agricultural outputs. As the water supply decreases, the subprocesses of soil water movement and evapotranspiration gain increasing importance, becoming predominant under sever water stress. Throughout the crop season, the subprocess importance and its response to water stress are modulated by the crop phenology. Compared to conventional parameter sensitivity analysis, our method excels in synthesizing divergent parameter importance changes and identifying influential subprocesses, even without high-sensitivity parameters. This study provides new insights into adaptive SPAC modeling by dynamically simplifying unimportant subprocesses in response to environmental changes.