Based on divergence characteristics in weather radar velocity PPI cha rts, a divergence information extracting technique is suggested. The median filt ering method is used to eliminate noise contamination and one by one data symmet ry technique is used to reduce the errors because of echo lacking or range foldi ng. Considering the effect of raindrop descending speeds, fuzzy searching method is adopted for extracting wind speed determined divergence feature and histogra m is used for wind direction determined divergence feature. Comparative results show that abstracting method is effective and can well display the divergence in formation. There is a good corresponding relationship between atmospheric dynami cs features showed by the divergence and largescale precipitation process.