Sensitivity Analysis of the VVP Wind Retrieval Method for Single-Doppler Weather Radars

Abstract

Abstract The sensitivity of the ill-conditioned coefficient matrix (CM) and the size of the analysis volume on the retrieval accuracy in the volume velocity processing (VVP) method are analyzed. By estimating the upper limit of the retrieval error and analyzing the effects of neglected parameters on retrieval accuracy, the simplified wind model is found to decrease the difficulty in solving and stabilizing the retrieval results, even though model errors would be induced by neglecting partial parameters. Strong linear correlation among CM vectors would cause an ill-conditioned matrix when more parameters are selected. By using exact coordinate data and changing the size of the analysis volume, the variation of the condition number indicates that a large volume size decreases the condition number, and the decrease caused by increasing the number of volume gates is larger than that caused by increasing the sector width. Using the spread of errors in the solution, a demonstration using mathematical deduction is provided to explain how a large analysis volume can improve retrieval accuracy. A test with a uniform wind field is used to demonstrate these conclusions.