Classification Of Short-time Heavy Rain Using Radar And Lightning Observations

With the continuously development of economic and social, people have gradually paid more attention on the disasters brought by the local short-time heavy rain. Higher prediction accuracy is also needed. Because there are significant differences of precipitation efficiency among different types of heavy rain, classification of precipitation is crucial. Therefore, accurate quantitative precipitation forecast based on heavy rain type classification techniques will be helpful to establish the heavy rain early warning capacity, to minimize the casualties and economic losses brought by the short-time heavy rain, and finally to protect the national security and the reality of economic development’s needs.In this paper, WSR-88D Doppler weather radar’s reflectivity data, lightning localization data, and temperature profile data are used to compare the evolution, convective structure, and lightning activity in the short-time heavy rain cases with one hour rainfall≥20mm) for the period of June to November2010. Three major types of brief heavy rain producing storms are found in the Shanghai area:warm stratiform clouds, warm convective cloud-dominated, and cold convective cloud-dominated. The comparisons are made for the radar reflectivity and VTL(vertically integrated liquid) between several levels, lightning activities, the height of several temperature. The results show:(1) VIL (below0℃, height of0-lkm,0-2km,0-3km and0-4km), average reflectivity (height of1-2km,2-3km and3-4km) in the low layers, and the maximum reflectivity have a close positive correlation with rainfall rate;(2) VIL (above0℃,-10℃and-20℃), average reflectivity (height of5-6km,6-7km and7-8km) in the high layers, VIL, total cloud-to-ground lightning (CG) flashes, negative cloud-to-ground lightning (NCG) flashes play an important role in identifying the storm types.(3) Rain rate has a close correlation with CG and positive CG frequency in cold convective cloud-dominated brief heavy rainfall, in which Negative CG is dominant and the higher the proportion of negative CG the stronger precipitation.Thirty four forecasting factors are selected in a multiple stepwise regression method to create prediction equation for the three types of precipitation. Three verification scores:CS I(Critical Success Index), POD (Probability of Detection), and FAR (False Alarm Ratio) are used. The verification results show the regression of warm convective cloud-dominated has the best recognition performance with a highest CSI of78.93%. On this basis, a approach to automatic classification and recognition technology is developed. After automatic identification and classification of the short-time heavy rain cases, rain rates are calculated using precipitation type-based Z-R relationships. The mean relative error of rain rate has a drop of3.19%compared to the default WSR-88D precipitation quantitative estimation, and has an decrease of3.19%,5.26%,22.56%,0.29%and1.84%relative to precipitation estimation by the current operational Z-R relationships respectively.