Sensitivity analysis of parameters and calibration methods in the Liuxihe distributed model for flood simulation

Parameters are key to hydrological models, and they have a significant impact on the performance of the model. In this study, the new-generation distributed physical Liuxihe model, a physically-based hydrological model is selected as an example. The model parameters are classified into six major categories on the basis of their respective physical properties: meteorological, hydrological, vegetation, geological, soil and landform. The sensitivity influence of each parameter on the flood prediction performance of the model is analyzed. The results show that the parameter optimization method can increase the model calibration efficiency by 3.4 times. Moreover, after the model is calibrated, the average BIAS of the flood peak flow is 12.1245%. The classification of model parameters and the evaluation of their impact on flood simulations for the Liuxihe model provide key results for model calibration. The proposed approach provides important technical support for flood warning and forecast.