Abstract: Based on the daily precipitation data of 115 rain gauging stations from 1963 to 2013 in the Yangtze River Basin, the multi-scale characteristics of spatial-temporal variability of precipitation in Yangtze River Basin during 1963-2013 are revealed by combining the wavelet multi-resolution analysis and the information entropy theory. The results show that, spatially, the variability of daily precipitation series exists obvious local clustering, significant heterogeneity along the latitudinal direction, and quasi uniformity along the longitudinal direction. There exists an obvious dividing line at approximately 103° longitude, and the variability in the west of this line is significantly higher than that in the east. The variability of daily precipitation has a strong linear correlation with elevation, and a weak linear correlation with mean annual precipitation and average number of rain days. Temporally, an obvious dividing line can be observed at approximately 103° longitude from the spectra of the relative contribution of each subseries to the total variability of the original daily precipitation series. In the west of the line, the contribution of each subseries at different temporal scales is higher than that in the east. And there exists an abruptness at the time scale of 256 days in the mutual information spectra. In addition, the time scale of the precipitation variability can be divided into three distinct scaling regions:(1) synoptic (2-16 days), (2) monthly to annual times (16-256 days) and (3) inter-annual (more than 256 days).

Key words: wavelet transform, information theory, daily precipitation, multiscale variability, Yangtze River Basin