Classification of Precipitation Radar Reflectivity Echo with Back-Propagation ANN

In this work, the data from the China new generation S band A series radar (CINRAD/SA) in Hefei during 2001 to 2003 was used to study the precipitation echo classification with a back-propagation (BP) model of artificial neural network. Three types of precipitation echo were considered in this study: stratiform, convective and mixed rain. Based on a case study with the trained BP ANN, it was proved that the single hide-layer BP model of ANN could be used to classify the different precipitation echoes at a high succeed-rate. Moreover, it was also found that the succeed-rate could be influenced by following factors: the amount and the input-order of training samples, the nerve cell number of the hide-layer and the choice of the learning rate.