Preliminary Test Of Summer Rainfall Prediction In China Based On Land Surface Thermal Factors

There are many influencing factors of summer precipitation in China; precipitation prediction is the emphasis of meteorological services and the long-term challenges for our scientists. In the paper, the underlying surface thermal abnormality is the entry point, based on the correlation analysis between April soil temperature (ST) over the Eurasian continent, the global sea surface temperature (SST) in the previous winter and summer precipitation in Eastern China, STs and SSTs over key regions are selected as the predictors of the summer precipitation prediction. Based on the observations during the period1961—1990, the Barnett-Preisendorfer canonical correlation analysis (BP-CCA) and the ensemble canonical correlation analysis (ECC) are used to build statistical prediction models of the summer precipitation in Eastern China by taking ST, SST or SST combined with ST as the predictors, respectively. Independent samples-validation is performed by using the observations during1991—2010. The main conclusions are summarized as follows:(1) According to the correlation between EOF time series of summer precipitation in Eastern China and soil temperature over the Eurasian continent and global sea surface temperature, selecting three areas of soil temperature on the Eurasian continent and five areas of sea surface temperature as predictors of CCA, there is certain predictive skill of soil temperature and sea surface temperature factors for summer precipitation.(2) There is certain predictive skill of soil temperature and sea surface temperature predictors for summer precipitation by building the prediction models of individual predictors. The predictive skill of summer precipitation is not the same by taking ST, SST or SST combined with ST as predictors; the predictive skill of the same predictor is different on the different regions. Compared with three ST and five SST predictors, the predictive effect of B, S2and S3region is better. The region of better predictive skill is the Yangtze River basin by taking ST, the Northeast China by taking SST, both the Yangtze River basin and the Northeast China by combining ST with SST.(3) In independent samples-validation, results show that the model by taking SST combined with ST as predictor exhibits higher predictive skill than that only taking one of them, implying that consideration of the ST improves the performance of the predictive model. The predictive model based on the land surface thermal factors has some skills in predicting the summer precipitation in Eastern China, taking the combination of SST and ST show better performance in summer precipitation prediction. There is less effective skills of SST in the Yangtze River areas, the reasons of the phenomenon are more complex, the possible reason is decadal weakening of the relationship between SST and summer precipitation. In the northeast, the skills are more high by taking SST than ST, it shows that the ST containing weaker prediction signal, however, ST still provide the necessary reference for prediction of summer rainfall.(4) Establishing prediction model by observations of the1961—1990and predicting summer rainfall of1991—2010in Eastern China, evaluate the prediction results. In the Yangtze River region, the score is66.9and anomaly correlation coefficient is0.09, indicating that the prediction model has predictive capability. ECC can collect the information of multiple predictors, it provide more predictive basis and make the predictive skill more stable.