A STUDY OF SST-FORCED VARIABILITY AND POTENTIAL PREDICTABILITY OF SEASONAL MEAN FIELDS USING THE JMA GLOBAL-MODEL

An ensemble climate simulation experiment has been conducted using the Japan Meteorological Agency (JMA) global model to study the SST-force d atmospheric variability and potential predictability of seasonal mea n fields. The ensemble consists of three model integrations each for t he same 34-year period. The three integrations use the same observed S ST for the period 1955-1988 but start from different atmospheric initi al states. The variance ratios of the SST-forced variability to the to tal variability of the seasonal mean fields are computed. The ratios a re considered to represent ''potential predictability'' (possible maxi mum predictability when SST is perfectly predicted). The variance rati os of pressure fields are generally high (50-90 %) in the tropics but low (less than 30 %) in the extratropics, suggesting that the potentia l predictability of the seasonal mean fields is high in the tropics bu t low in the extratropics. The variance ratios of precipitation take a wide range of values within the tropics from 74 % for N.E. Brazil rai nfall to 31 % for Indian summer monsoon rainfall, indicating large reg ional differences in the potential predictability of seasonal mean rai nfall within the tropics. The variance ratio of globally-averaged glob al mean land surface air temperature is high (66 %) but the ratio is l ow (less than 30 %) for the seasonal mean local surface air temperatur e over most land area. This suggests that the potential predictability of local land surface air temperature is low.