中文摘要：

利用1961--2008年广东逐日降水、NCEP／NCAR逐日再分析资料，先采用EOF分析、相关分析定义了影响广东前汛期降水的850hPa关键区——“南海北部风场指数”，再利用功率谱分析、Lanczos时间滤波器等方法研究了近48年广东前汛期降水与南海北部风场指数的低频振荡特征，并研究了广东暴雨与南海北部风场指数准双周振荡的关系。结果表明，广东前汛期降水与南海北部风场指数均以准单周、准双周振荡为主，而30—60天的季节内振荡不显著。南海北部风场指数与广东前汛期降水准双周振荡序列存在显著正相关，且明显高于准单周振荡序列的相关。近48年4—6月南海北部风场指数的准双周振荡波峰前后3天（个别4天）的广东省前汛期暴雨出现概率为78．2％。最后利用合成分析方法分析了南海北部风场指数准双周振荡的波峰附近有典型西南风暖区暴雨发生的大气环流场演变特征，可为广东暖区暴雨过程的中期预报提供参考依据。

英文摘要：

A wind field index in the north of South China Sea is defined using EOF and correlation analysis and with the daily precipitation data of observaton stations and NCEP/NCAR datasets for 1961--2008. A key area is identified at 850 hPa which affects rainfall during the annually first rainy season in Guangdong. Using the methods of power spectrum and Lanczos filter, low-frequency oscillation characteristics of the rainfall and wind field index in the north of South China Sea, and the relationship between rainstorm in Guangdong province and the quasi-biweekly oscillation of wind field index in the north of South China Sea are investigated. The results show that both the rainfall and the wind field index mainly exhibit quasi-weekly and quasi-biweekly oscillation, but the intraseasonal oscillation with 30-60 days period is not significant. The positive correlation between quasi-biweekly oscillation series of the rainfall and the wind field index is significant, and the correlation coefficient is significantly more than that between quasi-weekly oscillation series. The possibility for rainstorms to occur in Guangdong province within three days （four days in less cases） before or after the peaks of the quasi-biweekly oscillations is 78.2% from April to June for the last 48 years. By composite analysis, the common evolution characteristics are analyzed of the atmospheric circulation for typical rainstorms in the warm sector. They originate from the southwesterly with the peak of the quasi-biweekly oscillations. The characteristics can be used as reference for medium-range prediction of rainstorms in the warn section in Guangdong.