Parameter Uncertainty and Sensitivity Analysis for Nutrient Modelling in a Forested Catchment Using the Sequential Uncertainty Fitting (SUFI-2) Algorithm in SWAT-CUP

Abstract

Watershed models are useful in estimating the effects of management practices on both water quantity and quality. These models aid stakeholders and policymakers in managing water resources by informing them of the basic model structure and parameter interactions. For this study, the soil and water assessment tool (SWAT) was used to simulate streamflow, total nitrogen (TN), and total phosphorus (TP) in the steep and forested Ogouchi catchment. The parameter uncertainty and sensitivity of the models were then assessed using the sequential uncertainty fitting (SUFI-2) algorithm. This study demonstrated how SWAT can be used to estimate the nutrient load and hydrologic condition in a small archipelagic watershed where the entire hydrologic process is different from continental catchments. Results showed that the parameter uncertainty is highest for simulating TP (p-factor = 0.58-0.46, r-factor = 2.00-9.71) as compared with TN (p-factor = 0.85-0.54, r-factor = 0.91-3.27). Meanwhile, the sensitivity analysis revealed that nutrient loads are most sensitive to the SCS runoff curve number and average steepness, which are streamflow-related parameters. Overall, this study showed that nutrient loads are highly sensitive to streamflow-related parameters, which suggests that stabilizing runoff and sedimentation rates along the watershed is necessary to maintain good water quality in the reservoir.