Sensitivity and Uncertainty Analysis of the APEX Model to Water Status

【Background and Objective】 The APEX model is a comprehensive watershed-scale model for simulating the effects of management practices on agricultural systems and their impacts on water quality, soil erosion, and nutrient cycling. This paper analyzes the sensitivity of its parameters to water status in soil. 【Method】 The analysis is based on data measured from 2016 to 2019 from an irrigation experiment conducted in Jiaodong in Shandong province. Winter wheat was used as the model plant; the Sobol, Morris, and FAST methods were used to analyze the sensitivities of the APEX model parameters associated with crop growth and water stress. We considered the influences of groundwater depth, rainfall and irrigation. 【Result】 When groundwater depth was 1.25 m, the maximum root depth (RDMX) was the most sensitive parameter affecting evapotranspiration, biomass, and yield, while the maximum potential leaf area index (DMLA) was the most sensitive parameter impacting leaf area index (LAI). When the groundwater depth was increased to 5 m, the sensitive parameters influencing crop evapotranspiration and yield differed, with PARM38 (weight coefficient of water stress calculation) and RWPC1 (proportion of root biomass during germination) becoming the most sensitive parameters. Results calculated from all three methods indicated that as irrigation water increased, the sensitivity of RDMX decreased, while the sensitivities of DMLA, DLAI (peak point in growth season), and WA (potential light energy utilization) increased. The sensitivity of RDMX was significantly higher in dry years than in humid years, as opposed to the sensitivity of DMLA. Uncertainty analysis demonstrated that wheat biomass, yield, and evapotranspiration fell within the 5% to 95% confidence interval of the simulated data. 【Conclusion】 The most sensitive parameters identified by the Sobol, Morris, and FAST methods were consistent, although their sensitivity indexes varied with irrigation treatments, rainfall patterns, and groundwater depth. Considering computational efficiency and accuracy, the Morris method is more suitable for parameter sensitivity analysis of the APEX model. These findings provide valuable insights into the application of the APEX to analyze the impact of environmental conditions on crops.