Abstract:

Based on monthly temperature, sea surface temperature over Pacific, 500 hPa geopotential height field and circulation data in July and August from 53 stations in Northeast China during 1961 - 2004, the temporal and spatial temperature pattern over Northeast China from July to August in recent 44 yearswas analyzed using Emp iricalOrthogonal Function ( EOF) analysismethod. The canonical correlation analysis (CCA) method was used to forecast and testmonthly temperature in midsummer. Results show that the spatial pattern ofmonthly temperature in July and August ismainly classified into three types: the whole region consistency, north - south type and east - west type. The CCA method handles better in p rediction of monthly temperature in July and August of 2005 and 2006 in Northeast China.

Key words: northeast China, EOF, temporal and spatial characteristic, canonical correlation analysis, forecast

摘要：

利用1961～2004年东北地区53个测站7～8月逐月气温资料及太平洋海温场、北半球500hPa高度场和环流特征量资料,采用场经验正交函数展开和主成分分析方法,研究近44 a东北地区盛夏月尺度气温的时空分布特征;借助典型相关系数,对场与场的关系进行分析,利用多因子场预测未来要素场的典型相关方法,对东北地区盛夏逐月气温进行了预测及评估检验。结果表明:东北地区7、8月份平均气温分布的一致性比较好,分别占总方差的65%和72%;东北地区盛夏7、8月气温主要有全区一致型、南北型和东西型等几种主要空间类型。通过实践检验,典型相关分析方法对2005～2006年东北地区盛夏7、8月份月尺度气温趋势有较好的预测能力。

关键词: 东北地区, EOF分析, 时空特征, 典型相关分析, 预测