The Multi-timescale Features for Guangxi Summer Precipitation and the Related Predictors

Based on NCEP/NACR reanalysis data and Guangxi summer precipitation (GSP) station data, using the correlation analysis, composite analysis, empirical orthogonal function (EOF), empirical mode decomposition (EMD), abrupt change test and the statistic significant test methods, GSP multi-timescale characteristics and their related circulation as well as the external forcing features are analyzed. According to the diagnostic analysis, the fitting and the prediction equation of GSP are proposed by the multivariate linear regression method.GSP is mainly influenced by the mid-latitude height field anomaly in Lake Baikal region, the subtropical high and monsoon trough (MonTr) in the subtropical region, the low level jet (LLJ) and upper level jet (ULJ) in the same season, as well as the sea surface temperature (SST) anomaly in the eastern of the South Indian Ocean in the pre-winter and pre-spring.The possible physical concept model for GSP is that, when MonTr, LLJ, and the easterly to the south of the subtropical high (ESTH) occur at 850 hPa wind field, the blocking high (BH) over Lake Baikal at 500 hPa potential height, as well as ULJ over South China at 200 hPa wind field are stronger (weaker) than normal, and the subtropical high ridge location is northward (southward) to its normal position, the rainfall is more. The influences of circulation may impact summer rainfall anomaly through the multi-timescale features.Using EMD method, there are 5 principle modes for the summer rainfall. The variance contributions from the first to the fourth intrinsic mode function (IMF1—IMF4) are 55%, 18%, 12% and 12%, respectively. The periods over the statistic significant test are quasi-2 years, 7.6 years, 12.7 years and 19 years. On the scale of quasi-2 years, the summer rainfall is affected by the corresponding IMF1 components of the MonTr, LLJ, ULJ, BH over Lake Baikal, SST anomaly in the east of the South Indian Ocean. The summer rainfall has high relationship with the other influenced indexes on the different time scales.Using IMF1—IMF4 components of circulation factors and the multivariate linear regression method, the summer precipitation equation is fitted. The results show that the multiple correlation coefficients reach 0.73 with the significant level over 0.05. The tests verify that the summer precipitation is really influenced by the multi-timescale components of different factors.Furthermore, based on the IMFs of SST anomaly in the east of southern Indian in winter, the prediction model of the summer precipitation is constructed by the multivariate linear regression method. The trends of the 6 independent sample tests are accord with that of the observation. This method provides an idea in the regional climate prediction based on the multi-timescale features of predictant and predictor.