基于人工神经网络的GMS云图四类云系的识别  

IDENTIFICATION ABOUT FOUR KINDS OF CLOUD SYSTEMS IN GMS IMAGES BASED ON NEURAL NETWORK

作　　者：白慧卿[1] 方宗义[1] 吴蓉璋[1] 郑君里[2] 

机构地区：[1]国家卫星气象中心 [2]清华大学电子工程系

出　　处：《应用气象学报》1998年第4期402-409,共8页Journal of Applied Meteorological Science

摘　　要：应用人工神经网络方法较成功地识别了ＧＭＳ云图中的四类云系，并与传统的统计识别方法进行了对比，得出结论：神经网络方法更适合于云系的特征识别．还研究了神经网络本身的问题，取得一些网络设计的经验．Based on the theory of neural network,four kinds of cloud systems in GMS images are successfully identified. Comparing neural network with traditional maximum likelihood method,we can draw the conclusion that the neural network is even better for identifying cloud systems. Some problems about neural network itself are also studied,and some experiences for designing neural network are obtained.

关 键 词：神经网络 云系 云图 GMS云图 

分 类 号：P455[天文地球—大气科学及气象学]