Research On The Application Of Rainfall Estimation Technology Based On Radar In Flood Forecast Of Biliuhe River Basin

Precipitation is a key variable in hydrological cycle.The temporal and spatial variation of precipitation affects the runoff depth,peak value and peak time of flood,and the data error of precipitation also increases the uncertainty of hydrological model.Therefore,it is very important to obtain rainfall observation data with high accuracy,strong timeliness and good spatial-temporal resolution for basin flood prediction.The radar can obtain real-time,accurate,high spatial-temporal resolution and large-scale rainfall data.It is of great significance to combine the rainfall data estimated by radar with hydrological model in flood prediction.This paper takes Biliuhe River Basin as the research object,evaluates and analyzes the quality of the rainfall data estimated by radar combined with the measured rainfall data of the surface rainfall station,and calibrates the rainfall data estimated by weather radar with multi-point calibration method and Kalman filter calibration method,and performs accuracy evaluation of the rainfall data estimated by radar after calibration with absolute error,root mean square error and other indicators;Based on the flood forecast model of Xin'anjiang River,combined with the measured historical flood data,the multi-objective genetic algorithm is used to calibrate and verify the model parameters,and the simulation results are evaluated by using the relative errors of runoff and peak discharge;Finally,input the uncalibrated radar rainfall data,the radar rainfall data calibrated by multi-point calibration method and the radar rainfall data calibrated by Kalman filter into the Xin'anjiang model for runoff simulation,evaluate the application effect of the combination of the radar rainfall data and the Xin'anjiang Model in the flood prediction of the Biliuhe River Basin,and explore its applicability.The results show that:(1)With the increase of 1h cumulative rainfall,the rate of missed rainfall prediction by radar gradually decreases,while the critical index of accurate rainfall prediction gradually increases.The rainfall estimated by radar is underestimated as a whole,and the underestimate phenomenon is more and more serious with the increase of 1h cumulative rainfall.(2)Compared with the multi-point calibration method,the Kalman filter calibration method can not only effectively improve the accuracy of rain station radar pair in general,but also significantly reduce the dispersion of rain station radar pair and improve the accuracy of individual rain station radar pair.In terms of basin area rainfall calibration,multi-point calibration method and Kalman filtering method can effectively improve the accuracy of radar estimation of basin area rainfall,but the process line of radar estimation area rainfall calibrated by Kalman filtering method has better consistency with the process line of measured area rainfall of rainfall station.(3)10 floods were used to calibrate the parameters of Xin'anjiang model,and the qualified rate of the simulation results was 90%;3 floods were used to verify the model,and the simulation results were all qualified,indicating that the optimized parameters of Xin'anjiang model are suitable for the flood prediction of Biliuhe River Basin.(4)The combination of the radar rainfall data calibrated by Kalman filter method and the Xin'anjiang model shows that the flood prediction accuracy is high and it has good applicability in the Biliuhe River Basin.