Random Forest-Based Global Sensitivity Analysis for SWAT+: Identifying Key Hydrological Inputs

To improve decision-making in watershed management, numerical models, e.g. the Soil and Water Assessment Tool Plus (SWAT+) are crucial. Determining key input variables via Global Sensitivity Analysis (GSA) aids in minimizing model uncertainty and enhancing precision. This research utilizes a Random Forest (RF)-based GSA to evaluate the significance of SWAT+ input variables in modeling river flow and sediment load. By employing Permutation Variable Importance (PVI) metrics the sensitivity of input variables, such as climate data, is assessed. Findings suggest that rainfall is the primary factor affecting both flow and sediment load simulation results. The RF-based GSA framework not only preserves computational efficiency, but also offers an effective approach for ranking important input variables. This research provides important perspectives for enhancing hydrological model effectiveness and refining data collection methods.

کلیدواژه ها:

Global sensitivity analysis ، Random forest ، Permutation variable importance ، SWAT+ ، Input variables