3D Wind Field Reconstruction Based On Multiple Radar Data

The three-dimensional structure information of wind field has important reference significance for the weather short-term forecast and the disaster weather warning.At present,the research about retrieving wind field based on Doppler weather radar mainly focus on single and two radars.In the research of wind field reconstruction by single radar,the structure information of horizontal wind field is obtained at some assumptive conditions.In the research of wind field reconstruction by two radars,the three-dimensional wind field structure is obtained by using a specific scanning strategy or at many assumptive conditions.In theory,at least three radars are needed to reconstruct the three-dimensional structure of wind field.Limited by the number,cost and deployment location of radar,the joint observation of wind field by three or more radars is seldom studied.Currently,the construction of a new generation of Doppler radar network in China has been basically completed.The three-dimensional wind field retrieval by multiple Doppler radars is helpful to improve the monitoring and forecasting technology of severe convective weather.The research of this paper focuses on the reconstruction of three-dimensional wind field based on multiple Doppler weather radar data,including the quality control algorithm of Doppler velocity data,interpolation algorithm for radar data,wind field inversion algorithm.The quality control algorithm of Doppler velocity data includes removing isolated data points,smoothing filter,making up missing points and velocity dealiasing.In terms of velocity dealiasing,this paper studies two-dimensional multi-path velocity dealiasing algorithm and multi-scale cross search box algorithm,the former is for general data of one elevation layer,and the latter is for isolated echo area.Through the experiment,it is found that the combination of these radar data quality control algorithms can effectively improve the data quality,reduce the impact of various kinds of noise,and basically corrected all the aliasing data.The distribution of radar volume scanning data is unevenly distributed in space and the simultaneous scanning of multiple radars is not practical.This paper studies several interpolation algorithms of radar data,and proposes to apply Delaunay triangulation to radar data interpolation.Some radar data within 3 minutes were selected for the test.The advantages and disadvantages of each interpolation algorithm were summarized by comparing the reflectivity CAPPI image at 3km height.Experimental results show that Delaunay triangulation has excellent interpolation effect in a small range,and can get the data of any surface in the interpolation region.In the stage of wind field inversion,this paper studies three Doppler radar wind field inversion algorithm based on three-dimensional variation,which establishes loss function including observation term limit,continuity limit and smoothing limit,and uses conjugate gradient method to obtain the optimal wind field component with minimum loss function.By comparing the direct inversion method with the three-dimensional variation method,it is found that the three-dimensional variation method can improve the inversion accuracy of each wind field component,especially the vertical wind component.Finally,the actual radar detection data is used to further test and analyze the three-dimensional variational method.