Spatio-temporal sensitivity analysis for flow and sediment load modeling using SWAT

Authors: TALEBIZADEH, MANSOUR - Teck Resources Limited, Vancouver, Canada; Moriasi, Daniel; STEINER, JEAN - Kansas State University; GOWDA, PRASANNA - US Department Of Agriculture (USDA); Starks, Patrick; Verser, Jerry

Submitted to: Water Resources Management
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/17/2024
Publication Date: 12/26/2024
Citation: Talebizadeh, M., Moriasi, D.N., Steiner, J.L., Gowda, P.H., Starks, P.J., Verser, J.A. 2024. Spatio-temporal sensitivity analysis for flow and sediment load modeling using SWAT. Water Resources Management. https://doi.org/10.1007/s11269-024-04066-6.
DOI: https://doi.org/10.1007/s11269-024-04066-6

Interpretive Summary: The soil and water assessment tool (SWAT) model is one of the USDA Agricultural Research Service hydrologic and water quality models used to quantify the impact of land management and conservation practices on soil and water resources under a changing climate. For the outputs of SWAT to be useful for agricultural production policy and decision making, it needs to be accurately set up by populating important parameters with realistic values. In this study we used a sensitivity analysis method to identify important parameters associated with streamflow and in-stream sediment. Results indicated that four parameters associated with surface runoff, evapotranspiration, and groundwater processes were the most sensitive for the streamflow. For sediments, the stream bank channel and surface runoff parameters were the most sensitive. The results also showed that the location within the study area and season of the year affect the selected parameters. This information will help make sure SWAT is accurately set up to ensure reliable model outputs. USDA is an equal opportunity provider and employer.

Technical Abstract: Sensitivity analysis can be used for identifying sensitive model parameter, driving simulation outputs which in turn can assist in model calibration. However, sensitivity analysis is usually performed on aggregated output or a model performance metric calculated over a simulation period. This paper studies spatial and temporal variations of SWAT model parameter sensitivities for flow and sediment load modeling in an agricultural watershed. To this end, parameters uncertainties at hydrologic response unit (HRU) levels were characterized through probability distributions for SWAT parameters. Parameter sensitivities along their spatial locations were calculated on monthly time scales and were used for their ranking using the maximum and median values of sensitivity indices calculated over an 11-year simulation period (2003-2013). The runoff parameter (CN2) in SCS method, Soil evaporation compensation coefficient (ESCO), Plant uptake compensation factor (EPCO), and Threshold depth of water in the shallow aquifer (GWQMN) were identified as top sensitive parameters for flow, based on the maximum calculated sensitivity indices. Manning's n parameter (CH_N2), driving the calculation of flow velocity, and channel erosion were identified as most sensitive parameters for sediment load simulation. Parameter sensitivities according to the median of the calculated sensitivity indices resulted in a different ranking of parameters, compared to the ranking based on the maximum of the calculated sensitivity indices. Temporal (calculated on monthly time scales) and spatial (calculated at HRU level) variations of parameter sensitivities were plotted for sensitive parameters. According to the results, some parameters tend to have higher sensitivities during most of the simulation periods, while some others indicated higher sensitivities during limited periods where their impact become more dominant in driving flow and sediment load values. This information can be used for identification of periods where certain model parameters indicate higher sensitivities and potentials to be estimated through calibration. USDA is an equal opportunity provider and employer.