Automatic Calibration Method for Flood Evolution Model Parameters Based on Data Assimilation Technology: A Case Study of the Wuqiangxi Basin

This study addresses the challenge of achieving accurate and timely flood forecasting in the face of increasingly frequent extreme flood events. We propose an automated calibration method for flood evolution model parameters based on hydrodynamic simulation and a three-dimensional variational data assimilation algorithm. The Wuqiang River basin, characterized by complex terrain and concentrated rainfall, is used as a case study for model validation. Firstly, a two-dimensional hydrodynamic mode l of the Wuqiang River was constructed to accurately capture the basin’s complex topography and flood evolution process. Subsequently, a three-dimensional variational data assimilation technique was introduced to integrate observed water levels and other o bservational data with model simulation results, enabling the dynamic optimization of the model’s roughness coefficient. Research results demonstrate a significant improvement in flood forecasting accuracy at Yujiacun and Fenkoucun stations after data assimilation calibration. The agreement between simulated and observe d water levels has been greatly enhanced, with notable improvements in the Nash-Sutcliffe Efficiency (NSE) coefficient and maxim um water depth error.