Application Of Semidistributed Hydrological Model Based On Different Data Conditions In Flood Simulation

Precipitation and evaporation are very important input data in flood simulation.At present,there are two main methods to measure precipitation and evaporation:ground station observation and remote sensing observation.Generally speaking,the observation accuracy of the ground station is high,but only the data on the point scale can be obtained,which is difficult to reflect the spatial distribution of rainfall and evaporation completely.Remote sensing observation can obtain the data on the surface scale.However,there are still some problems in remote sensing observation,such as high uncertainty,space-time scale and so on.Taking the Qixing watershed of Sanjiang Plain as the study area,the tropical rainfall measurement mission(TRMM)data are downscaling corrected by using the methods of multiple linear regression and geographical weighted regression.The results show that the correction effect of geographical weighted regression is better,which can reduce the average relative deviation between the corrected precipitation and the measured precipitation series and improve the correlation coefficient;while the multiple linear regression correction,the average relative deviation and correlation coefficient improved slightly.The remote sensing evapotranspiration data of the Qixing watershed are assimilated by using the Kalman Filter(KF)and the Ensemble Kalman Filter(EnKF)according to the measured evapotranspiration data of the reference station(Baoqing station),and the assimilated remote sensing evapotranspiration data are input into the semi distributed hydrological model for flood simulation.The results show that the remote sensing evapotranspiration data can provide the effective evapotranspiration data for the area without the ground evaporation station.In order to make better use of the area rainfall and area evaporation data,a semi distributed hydrological model based on full storage runoff(SHMSR)is constructed by referring to the Xin'anjiang model.The results show that the model can be applied to flood simulation of Qixing River Basin in Heilongjiang Province.Comparing the results of flood simulation using rainfall station data and downscaling corrected TRMM data,it is found that using downscaling corrected TRMM data can improve the accuracy of flood simulation.Comparing the results of flood simulation using the evaporation data of the nearby evaporation station with the assimilation remote sensing evapotranspiration data,it is found that the assimilation remote sensing evapotranspiration data can improve the accuracy of flood simulation.At the same time,the rational formula is gridded,and 10 flood are selected for simulation.The results show that the grid rational formula model can simulate the single peak flood process in a large watershed.