Parameter sensitivity analysis for a physically based distributed hydrological model based on Morris' screening method

Abstract

Achieving accurate parameter calibration is a relatively difficult problem in development and application of physically based distributed hydrological models. In the current research, Morris' screening method (one factor at a time [OAT]) and an error evaluation calibration were jointly adopted for parameter sensitivity analysis (PSA) for a distributed hydrological model developed for a single mountainous terrain. The results indicated that (1) the order of sensitivity of the parameters for rainfall−runoff calculation is thickness of the first soil layer, coefficient of permeability, infiltration velocity, porosity, and coefficient of roughness; (2) the order of the parameters for snowmelt calculation is bulk transfer coefficient, temperature correction factor, and albedo of snow cover layer surface; and (3) the value change of each parameter of the first layer in the soil vertical profile mainly impacts the numerical value of discharge of the surface flow, while the value change of each parameter of the second layer impacts the numerical value of discharge of the surface flow less, but results in the peak misplacement to some extent. The results are expected to provide methodology references for PSA and calibration for the physically based distributed hydrological model which developed for mountainous or otherwise single terrain.