中文摘要：

针对2003年7月3—4日淮河流域梅雨暴雨过程,利用AREM模式,在分析暴雨预报对不同来源的初始资料和不同要素初始误差的敏感性的基础上,重点研究了降水过程中误差的演变特征和发展机理。分析结果表明,初始小振幅误差增长最快,而伴随着降水的发生和发展,误差演变特征表现为由局地增长发展为全局传播的过程,且误差最优增长总是出现于雨区,这意味着雨带是误差增长的敏感区域。雨区内存在的初始误差对降水预报误差具有重要贡献,初始湿度条件不仅影响误差的传播特征,还使雨带上中小尺度误差迅速增长并造成更大尺度的误差。基于误差能量公式的计算结果表明,误差增长的能量来源主要由凝结加热提供,因此,从能量角度而言,误差增长和降水增大是同＂源＂的,从而使暴雨可预报性受到固有的限制。

英文摘要：

The prediction of heavy rainfall along the meiyu front in China in July 2003 is studied.Based on the study results of the sensitivity of precipitation prediction in the AREM to the different initial data analyses and the different initial errors of variables,the evolution of error growth and its mechanism are decribed and discussed in details in this paper.The results indicate that an initial error with smaller amplitude will grow faster and afterwards the evolution of error growth following the occurrence and development of rainfall is characterized by a transition from local growth to global spreading to the overall domain with the remarkable error growth occurring in the rain belt,suggesting that the rain belt is a sensitive area for error growth.The initial errors in the rain belt contribute significantly to the error of precipitation prediction.The distribution of initial moisture has not only important effect on the propagation of error growth,but also drives meso-and small-scale errors to grow rapidly and subsequently causes the error growth in larger-scale motions. Since the computational results based on the error energy formula show that the energy for error growth is largely provided by latent heat,it is therefor concluded in terms of energy that the precipitation and error growths have the same＂energy source＂,which imposes intrinsic limitation on the predictability of heavy rainfall.