0～3小时短时定量降水预报算法研究

STUDY OF 0～3 HOUR SHORT-TERM FORECASTING ALGORITHM FOR RAINFALL

利用广东省3km高度CAPPI雷达反射率因子拼图资料，对2012年6月22日一次降水个例进行0-3h短时定量降水预报算法研究，以期通过改进算法来提高降水短时预报的准确率。在利用交叉相关算法获取回波移动矢量场后，使用五点平滑及卡尔曼滤波的方法对其进行处理，利用处理后的矢量场对雷达回波进行3h外推，再进行0．3h降水量预报。研究结果表明，经过滤波处理后可得到时空连续性更好的回波移动矢量场，滤波后回波外推预报效果明显改善，其临界成功指数（CSI）有所提高，空报率（FAR）显著降低，提高了降水预报准确率。

In this paper, mosaics ofreflectivity from a CAPPI radar at the 3 km height in Guangdong province was used to study short-term forecasting algorithm for a rainfall case. The aim of this study was to improve the accuracy of forecasting. Five-point smoothing filter and Kalman filter were applied to treat the vector field of echo motion obtained through the method of cross-correlation. Then the vector field and radar echoes were combined to make short-term forecast and nowcast. The results show that spatial and temporal continuity of the vector field was better than that before the filtering. The accuracy of echoes and rainfall forecast after the filtering were improved significantly.