Multiobjective sensitivity analysis of sediment and nitrogen processes with a watershed model

[1] This paper presents a computational analysis for evaluating critical non-point-source sediment and nutrient ( specifically nitrogen) processes and management actions at the watershed scale. In the analysis, model parameters that bear key uncertainties were presumed to reflect the importance of natural processes and/or management actions that they represent. The multiobjective generalized sensitivity analysis and the tree-structured density estimation procedures were used in combination to investigate correlation structure in the parameter space while accounting for multiple objectives. Using the Soil and Water Assessment Tool, this framework was applied to the Dreisbach watershed in Indiana in the Midwestern portion of the United States. Results showed that incorporation of parameter interactions and multiple objectives is essential to obtaining conclusive information about critical system processes and management actions. Interactions between surface runoff volume and within-channel processes were critical to describe transport of sediments in the study watershed. Key management actions for total nitrogen control were found to be nitrogen fertilizer application and upland farming practices. The sensitivity analysis reported herein could be used to derive a list of key non-point-source best management practices for development of watershed management plans. Implications of the analysis that are relevant to calibration of complex watershed models are discussed.