A Radar-based Technique for the Identiflcation of Convective and Stratiform Precipitation using the FLNN with Its Application to Estimation of Precipitation

摘要

@@ 1. Introduction The improvement for radar data derived quantitative precipitation estimation (QPE) has been a challenge task in radar meteorology since the first weather radar was used to measure precipitation in 1940s (Cheng,1994). This is due to many factors affecting the accuracy for radar derived precipitation. One of most important factors is the variation of the relationship (Z-I) between radar observed reflectivity (Z) and precipitation intensity (l). It is generally accepted that the Z-l relationships may be much different for different precipitation types. Therefore, there have been many methods presented to classify the convective and stratiform precipitation echo of radar, which were use to estimate precipitation using different Z-/ relationships (Steiner et al., 1995, Rosenfeld et al. ,1995, Houze 1997, Biggerstaff et al. , 2000). Following this methodology in this paper, fuzzy neural network (FNN) was presented to identify precipitation types from radar echoes, which were applied to QPE using different Z-l relationships.