Spatio-Temporal Sensitivity Analysis for Flow and Sediment Load Modeling Using SWAT in a Heterogeneous Agricultural Watershed

Sensitivity analysis can be used to identify model parameters driving simulation outputs. However, sensitivity analysis is usually performed on an aggregated output or a model performance metric, calculated over a simulation period. This paper studies the spatial and temporal variations of the SWAT model parameter sensitivities for flow and sediment load modeling in an agricultural watershed. The Willow Creek Sub-watershed in central Oklahoma, characterized by diverse land use and agricultural management, was used as the case study. Parameter uncertainties at the hydrologic response unit (HRU) level were characterized through probability distributions, while parameter sensitivities were calculated using Standardized Regression Coefficients (SRCs) at a daily timescale. Significant temporal variations were observed for different parameters, influenced by changes in climate and management conditions. Additionally, the results indicated certain parameters exhibited high sensitivities only during specific periods, while others maintained high sensitivities throughout most of the simulation period. This study underscores the necessity of integrating both temporal and spatial sensitivity analyses to capture the dynamic nature of parameter sensitivities at a finer level of spatial granularity than the watershed scale.