The Optimization And Evaluation Of Quantitive Precipitation Estimation Based On Multi-radar Mosaic

With the frequent occurrence of the disastrous weather in the recent years, thesurface rainfall spatial distribution information with high resolution and high accuracyis more and more important in the short-term weather nowcasting and fine forecastingoperational work. The QPE (Quantitive Precipitation Estimation) algorithm based onmulti-radar mosaic has presented good stability and real-time performance in thepreliminary operational applications. However, it also suffers from the error sourcessuch as radar beam partial blockage, Z-R relationships and radar-gauge merge method,etc, which limited the accuracy of radar QPE seriously. This paper evaluates theoperational effectiveness of the multi-radar mosaic QPE and mainly optimizes andimproves on the key factors of the radar QPE algorithm using the new technologyprocessing scheme at the same time, the main work of the study and the main resultsincluded are presented in the following four sections:(1) Evaluate and analyze the operational effectiveness, the applicability and mainerror souces of the multi-radar mosaic QPE algorithm，using the radar VPR (VerticalProfiles of Reflectivity) as the main means, with four different rainfall events. Theresults show that: with respect to the stratiform cloud precipitation, the coexistence ofabsolute blockage of the radar beam and bright-band is prone to overestimation of theradar QPE and the echoes from the top of the cloud is related to the underestimationof the radar QPE. The coexistence of different rainfall types makes uniform Z-Rrelationship obviously inapplicable in the in Meiyu front and typhoon, which resultsin serious local overestimation and underestimation of radar QPE. The severeconvection among the squall line and the asymmetry of typhoon are also importantreasons reducing the accuracy of the radar QPE.(2) For the situation of the coexistence of different rainfall type in the multi-scaleweather events but the Z-R relationship is uniform, a new grouping fitting scheme,which is based on the storm identification, is proposed to improve the accuracy of theinitial field of radar QPE. With the evaluation and comparison using four differentrainfall events, it shows that the proposed fitting scheme is obviously superior to theuniform dynamical Z-R fitting scheme and simple-grouping Z-R fitting scheme and itcan derive more accurate radar QPE field, which makes it have better rainfallestimation effects and operational capabilities.(3) For the situation of radar partial blockage effects which is related to the underestimation of the corresponding radar QPE, a partial blockage identificationmethod based on the probability features of radar echoes is proposed. With the help ofthe identification results of radar beam’s partial blockage, the weak radar echoes areeliminated before constructing the multi-radar mosaic. The evaluation and verificationresults using four different rainfall events show that: with the new data processingscheme, the discontinuity of multi-radar mosaic and the weak radar echoes can beeffectively enhanced and the influence on Z-R fitting scheme is reduced effectively,which further increases the accuracy of the radar QPE derived within and out of theregion affected, and the accuracy of radar QPE on the whole is also raised.(4) For the optimization of radar-gauge merge algorithm, a new merge methodunited storm-based grouping Z-R relationships and optimal interpretation is proposed,and it is compared between the dynamical unique Z-R fitting scheme and optimalinterpretation scheme, the simple grouping Z-R fitting scheme and optimalinterpretation scheme, and the gauge-only IDW（Inverse Distance Weights）method.The results show that merge method with storm-based grouping Z-R relationships isobviously superior to the other two merge methods; the three merge methods are allsuperior to the gauge-only IDW method during the Meiyu front, typhoon and squallline. However, the latter two merge method show obvious inferiority to thegauge-only IDW method, only the merge method with storm-based grouping Z-Rrelationships is superior to the gauge-only IDW method.