ECC Forecasting Of Drought Index Over Southeastern China In Summer

On the basis of the fact that there exists significant negative correlation between temperature and precipitation over the southeastern China in summer, a drought index is appropriately constituted. The article takes summer drought index of 117 stations in the southeastern China from 1951 to 2004 as the predictand, which are counted out with monthly mean precipitation and temperature of 160 stations in China collected by Beijing Meteorological Center, and then analyzes its spatial and temporal distribution by EOF. Choosing monthly mean sea-surface temperature, 500hPa height in the northern hemisphere and surface temperature in Eurasian continent from NCEP/NCAR reanalysis as predictors, we predict summer drought index over southeast based on single predictor BP method, and finally gather five forecasts as a comprehensive model, getting the following results: (1)According to significant negative correlation between temperature and precipitation over the southeastern China in summer, a drought index can reflect the climate anomaly. In view of the locality and complexity, the drought index, instead of precipitation, is more advantageous to establish a reliable forecast method; (2)Analyzing the drought index by EOF method , four spatial patterns showing summer drought/flooding distribution in southeastern China are obtained: consistent pattern in the middle and lower reaches of the Yangtze River valley; negative correlation between north and south; negative-correlation anomaly distribution between the north of the middle and lower reaches of the Yangtze River valley and other areas; saddle structure. The leading four spatial patterns explain about 19.12%, 15.95%, 11.59%, 6.51% of the covariance respectively. The corresponding time series to the first spatial pattern has slightly rising tendency, shows that the variability tendency of drought index in the middle and lower reaches of the Yangtze River valley drops slightly, namely the past several dozens years to the leaning waterlogged change; (3)Taking surface temperature (the tropical Indian Ocean, the tropical Pacific and the north Pacific), 500hPa height in the northern hemisphere and surface temperature in Eurasian continent in winter (December, next January, next February) as predictors to predict drought index can provide certain useful information. Among them, surface temperature in Eurasian continent has highest skill, then the tropical Pacific surface temperature, 500hPa height in the northern hemisphere and the tropical Indian Ocean, the skill of the north Pacific is lowest of all; (4)Through analyzing forecast skill origin of single predictor, it is found that different variables recognize different forcing and give skillful forecasts over different regions of the southeastern China, and ensemble forecast makes up single member's insufficiency; (5)ECC forecast can collect the forecast information provided by several predictors, and give more overall and fully basis compared to the single factor CCA forecast. Besides, the superensemble mean, whether spatial correlation or temporal correlation between predicted and observed fields, has better effect and stabler skill than the equally weighted ensemble mean; (6)By the independent sample examination, it is confirmed that this model has certain actual forecast ability.