Abstract:

Based on SOM (self-organizing maps) and EOF (empirical orthogonal function) methods, the dailymean and seasonal-mean summer precipitation patterns in Eastern China are analyzed and compared. Results show that the SOM method, compared with EOF, describes precipitation distributions more realistically. The spatial structures of summer precipitation patterns can be well explained by 850 hPa wind anomaly and 700 hPa specific humidity anomaly. As a posterior problem, 850 hPa height anomaly can well explain the corresponding precipitation area. However, as a prediction problem, due to other conditions such as moisture and temperature anomaly, the negative height anomaly in the numerical weather forecasting does not always indicate strong precipitation.

Key words: China summer precipitation, self-organizing maps (SOM), empirical orthogonal function (EOF)

摘要：

使用自组织映射(SOM)和经验正交函数(EOF)两种方法, 对比分析中国东部地区夏季日平均和季节平均的降水分型。结果表明, SOM比EOF能够更真实地描述降水分布。SOM得到的夏季降水分型的空间结构能够很好地用850 hPa的风扰动和700 hPa的比湿扰动解释。作为事后问题, 850 hPa的高度扰动负值中心能够很好地解释对应的降水落区; 但作为预报问题, 强降水不仅由数值模式预报中的负值高度扰动决定, 还受制于水汽和温度扰动等因素。

关键词: 中国夏季降水, 自组织映射(SOM), 经验正交函数(EOF)