Abstract: In short-range climatic prediction, the transformations for summer precipitation usually have been made in single station or in stations in some area in China. There are linear transformations, such as anomaly, anomaly percent,and normalized for single station. Otherwise, the average and principal component analysis based on stations in the area has been usually made. Sometime the nonlinear transformations are also used on the preapitation series for impnwing the prediction, for example, square root, cubice root, or class transformation and so on. In order to investigate spectral structure and normality on summer precipitation in the original and transformed series, the data of summer (June to August) precipitation in Sanxia in China during the period 1952-1997 are selected in this paper. The transformations of precipitation are completed by nonlinear transformation. Those are square root,cubic root,probability (by Gamma distribution),class and rank transformation. Otherwise,the average, principal comonent of the variables in the 16 stations in the area is also analyzed. The method of power spectral analysis is used for the series. The parameters of skewness and kurtosis in distribution are calculated for the variables.