Distributed hydrological model parameter sensitivity analysis and calibration method based on LHS-GAS-BP

The invention discloses a distributed hydrological model parameter sensitivity analysis and calibration method based on LHS-GAS-BP, and the method employs a genetic algorithm to optimize the weight and threshold of a BP neural network, and improves the defect that a BP neural network learning algorithm is liable to fall into local minimum. And sensitivity definition according to a dependent variable change rate caused by independent variable coupling change is adopted, so that relatively sensitive parameters in the model can be identified, parameter dimensions are reduced, model calculation cost is reduced, and support is provided for model parameter optimization and uncertainty analysis. The subjective influence of a manual trial and error method on hydrological model parameter calibration is overcome, model operation and goodness-of-fit evaluation can be automated through model batch execution and result batch extraction, parameter adjustment efficiency is improved, and compared with other parameter calibra