Evaluation of daily precipitation in China from ECMWF and NCEP reanalyses

Correlation analysis, hidden period analysis and complex Morlet wavelet transform were used with daily rainfall in China from observation, ECMWF (European Centre for Medium-Range Weather Forecast) and NCEP (National Center for Environmental Prediction) reanalyses to evaluate the validation of precipitation estimates. The results showed that Fisher's test and wavelet analysis, specially the latter, are useful tools for statistical analysis of daily precipitation datasets. Daily rainfall data obtained from ECMWF reanalysis are obviously better than those from NCEP reanalysis in terms of long period daily mean, local correlation, variation amplitude, fluctuation pattern and frequency. Although there is still room for improvement, ECMWF reanalysis is the best available dataset with global coverage and daily variability. In both of the reanalyzed daily mean precipitation fields, the higher estimations of Sichuan Basin rainfall are most likely caused by the topography of the basin, where small scale mountains in the southeast could not be represented by the reanalysis grid points, the typhoon and summer monsoon rainbelt could reach this region.