Holistic parameter sensitivity: A new perspective to address the risk brought by multi-source deep uncertainty on...

Watershed models are widely applied to support watershed management, but their applications are commonly hindered by the presence of multi-source deep uncertainty (MSDU). To mitigate the impact on watershed management, this study categorizes deep uncertainties into three sources and develops a holistic sensitivity analysis to quantitatively characterize changes in parameter sensitivity influenced by MSDU. The method has been applied in Upper Chishui River Basin with (Soil and Water Assessment Tool) SWAT model, and the results indicate that the holistic sensitivity indices (HSI) of parameters show spatial heterogeneity and are significantly affected by MSDU, with the maximum increase and decrease in sensitivity indicators reaching 73.3% and 36.4% under different sources of uncertainty, respectively. The study also reveals the impact mechanism between changes in driving conditions and sensitivity indices. Based on the values and characteristics of HSI of parameters under 307,152 MSDU scenarios, the study identifies 34 sensitive parameters out of 49, including 13 extremely sensitive parameters and three spatially sensitive parameters. The results highlights that our method is more effective than traditional methods in addressing changes in parameter sensitivity due to MSDU, can detect the increase in parameter sensitivity caused by climate change and human actions, and reduces the risk of identification bias in parameter values.