Multiobjective sensitivity analysis and model parameterization approach for coupled streamflow and groundwater table depth simulations using SHETRAN in a wet humid tropical catchment

Abstract

Physically based Distributed Hydrological Models (PDHM) are complex and exhibits extreme challenges for simulating the dynamics of coupled streamflow and groundwater table depths. The present work assesses the capabilities of a PDHM, SHETRAN in capturing streamflow and depth to groundwater table to represent the hydrologic processes within a wet tropical humid catchment. The most significant model parameters for the coupled simulations of streamflow and groundwater table depths are assessed through a multiobjective sensitivity analysis using Morris screening method and local sensitivity analysis. Horizontal and vertical soil saturated hydraulic conductivities, specific storage, as well as Actual/Potential evapotranspiration and Strickler overland coefficient, are highly sensitive for both the model responses considered. In order to arrive at the optimal model parameters for simulating depth to water table and streamflow, the effect of three model parametrization approaches were analyzed: with no spatial variation in soil properties, elevation based variation of only middle and bottom soil layer saturated hydraulic conductivity and elevation based variation of horizontal (kx), vertical (kz) soil saturated hydraulic conductivities and specific storage (Ss). Results indicate that for the studied watershed, the elevation based variation in kx, kz, and Ss resulted in a better prediction of depth to water table variations (in terms of correlation coefficient) with no significant improvements in streamflow simulation compared to the other two parameterization cases. The study suggests incorporating preferential flows in SHETRAN model to improve prediction of the water table dynamics in catchments belonging to the humid tropical region.