摘要
在天气学分型的基础上 ,用神经网络Back -Propagation(简称B -P)算法 ,以 1990～ 1995年汛期 (6月下旬～ 9月上旬 )日本数值预报产品和实时资料若干气象要素作为预报因子 ,对黄河三角洲的汛期暴雨进行了试预报 ,预报准确率平均达 70 % ,比主观预报准确率提高了 5 0 %以上。
Based on the classification of synoptically meteorology,using the Back-Propagation algorithm of neural network,and by combining the numerical forecast products in Japan and the actual data from 1990 to 1995,so the rainstorm forecast of the Yellow River Delta is finally achieved,the average accuracy of forecasting can reach to 70%,which is enhanced 50% than subjective accuracy of forecasting .
出处
《河南气象》
2000年第3期19-21,共3页
Meteorology Journal of Henan
关键词
神经网络
黄河三角洲
暴雨
G-P算法
汛期
预报
Neural network
The Yellow River Delta
Rainstorm
Back-Propagation algorithm
