Precipitation Estimation And Fusion Over The Middle And Upper Reaches Of The Heihe River Basin Based On Doppler Weather Radar

In northwestern China,ground rainfall stations and meteorological stations are distributed sparsely and unevenly,so that although they can provide high-accuracy precipitation on a single point,the ability of ground stations estimating the spatial distribution of rainfall is insufficient.However,the new generation of Doppler weather radar deployed in China can measure the precipitation with high spatial and temporal resolution in a wide range,which can better reflect the spatial distribution of precipitation.However,its accuracy of precipitation estimation is not as good as that of ground stations.In order to obtain precipitation data with high precision and high spatial and temporal resolution,and provide better data reference for water resources management in this area,this thesis takes the middle and upper reaches of Heihe River in Northwest China as an example to carry out the optimization study of Z-I relation of quantitative estimation of local topographic cloud precipitation by Zhangye Doppler weather radar and realize the fusion of ground rainfall and quantitative estimation of precipitation by radar.The main research contents are as follows:The PyCINRAD package in Python is used to analyze the structure of basic data obtained by scanning Zhangye C-band Doppler radar.Combined with professional data analysis and file processing packages such as pandas,numpy and os,the echo data in radar basic data files are extracted in batches,singular values and missing values in time and space are quality controlled,and rainfall station-radar water data pairs are constructed.The Z-I relationship of Zhangye radar applied to local precipitation measurement is optimized by using the global optimization and the subsection optimization method according to distance,and the estimated results are compared with those obtained by using the default formula and typical formula.The results show that the estimated precipitation directly obtained by using the default formula or typical formula seriously underestimates the real precipitation in this area.The quantitative estimation of precipitation by Z-I relationship obtained by the two optimization methods can obviously improve this underestimation,and can improve the accuracy of radar quantitative estimation of precipitation both on the whole and at the single point scale.However,due to the limitation of precipitation estimation by Z-I relationship method,the estimation of precipitation by Z-I relationship obtained by the two optimization methods has limited optimization degree at the single point scale.In order to further improve the accuracy of radar precipitation estimation and combine the advantages of ground rainfall with precipitation estimated by radar,the methods of Geographic Ratio Analysis(GRA),Geographic Differential Analysis,GDA)and Conditional Merging(CM)were used to construct the data fusion model of precipitation estimated by radar and ground rainfall,and the accuracy of the fusion precipitation is tested by the method of leaving one cross-validation.The results show that,in terms of the data accuracy of precipitation,the fusion precipitation obtained by GRA method can not improve the accuracy of precipitation estimated by radar,while the GDA and CM methods can effectively improve the accuracy of precipitation estimated by radar.The data accuracy obtained by CM method is slightly better.From the aspect of describing the temporal and spatial distribution characteristics of precipitation,both the GDA and CM methods can improve the accuracy of the spatial distribution of precipitation monitored by a single Z-I relationship,while the GRA method not only can't accurately monitor the distribution of precipitation centers,but also has a large error in estimating extreme precipitation,which is the worst.It is not applicable to the fusion of ground station and radar monitoring precipitation in the study area.