Based on the concept of the multiscale wavelet edge detection, an algorithm is presented to process the radar reflectivity data, so to identify convective cloud echoes effectively and retrieve convective precipitation. Taking a severe thunderstorm with heavy precipitations happened in Nanjing between 6 and 7 July 2009 as an example, a case study is conducted, focusing on the 180 minutes with relatively dense lightning flashes and severe precipitation. The convective cloud echo areas are then recognized by means of the multiscale wavelet edge detection algorithm with the radar reflectivity data from Longwang station every 6 minutes. The corresponding convective precipitation is inversed through ZI relation. The lightning flashes in the same 6 minutes are calculated, so the relation between convective precipitation and lightning can be analyzed. Results show that Rain Yields per flash (RPF) ranges from 7.47×106 to 4.46×107  kg, with average value being 1.72×107 kg; the nonlinear curve fitting equation of precipitation intensity and lightning frequency is established.