Circulation characteristics of autumn rainstorm days anomaly and construction of its prediction model in Hainan

By using daily precipitation at 18 meteorological automatic stations in Hainan, NCEP/NCAR monthly reanalysis data with a reso-lution of 2.5 °×2.5 °, NOAA SST and historical data of CFSv2 model, the corresponding atmospheric circulation characteristics of Hainan au-tumn rainstorm anomaly and its relationship to SST are analyzed for the period from 1982 to 2013. The circulation factors closely related to au-tumn rainstorm days, which are predicted well by CFSv2, and SST are extracted as predictors to build a predictive model of autumn rainstorm days. Results show the following. (1) Autumn rainstorm days in Hainan are closely related to atmospheric circulation. In more autumn rainstormyears, there is prevailing easterly wind near Hainan. Atmospheric pressure in the tropical western Pacific to south China sea is lower, the tropi-cal vortex systems tend to be more active, and there is a southeast wind anomaly in the region, which provides abundant moisture. The zonalwind shear in western Pacific is weaker, a warm core structure easily forms, corresponding to the occurrence and development of typhoon. On the other hand, the SST forcing effect is significant. In the context of the tropic central and eastern Pacific SST anomaly, atmospheric circula-tion and tropical convection activities are affected, resulting in autumn rainfall anomaly. (2) The autumn rainstorm days are closely associated with the lower level height field, sea level pressure, low level wind and zonal wind shear in the tropical Pacific. CFSv2 has good prediction skillon the high-impact areas of such circulation fields. (3) Predictive model of autumn rainstorm days is built by using optimal subset regression based on effective model information. The results of cross-validation and independent samples test demonstrate that the predictive method and model have good prediction skill, which can provide an effective reference for short-term climate prediction of autumn rainstorm days.