Parameter Assessment in the Nonpoint Source Pollutant Load Estimation Model Using Regression Trees

Key Points

• The research aims to analyze parameter sensitivities and extract feature importance in pollutant load estimation models using regression trees.
• Utilized distributed runoff models and Latin hypercube sampling for sensitivity analysis.
• Assessed the impact of parameters on pollutant load estimation precision.
• Employed regression trees to extract importance of parameters and identify effective ranges.
• Identified key parameters significantly affecting pollutant load estimation accuracy.
• Achieved a reduction in the parameter range for effective estimation to less than half of the initial exploration set.
• Observed improved estimation precision with resampling based on the parameter ranges identified by regression trees.