2013 年 69 巻 4 号 p. I_391-I_396
This study investigates the predictability of Thailand more-than-normal rainfall in 2011 from pre-monsoon to rainy season, which induced severe floods and immense damages, by incorporating seasonal prediction conducted by a general circulation model (GCM) and downscaling based on Singular Vector Decomposition (SVD) analysis. Our coupled atmosphere and ocean GCM has ability to predict tropical sea surface temperature (SST) in seasonal timescale. Taking this advantage, we derived statistical relationship between a large-scale tropical SST field and regional rainfall over the Indochina Peninsula using SVD analysis. Based on this transform function, SSTs predicted by the GCM were translated into the predictand of regional rainfall. Our final prediction products well captured the more-than-normal rainfall during the pre-monsoon and the late-monsoon seasons in Thailand 2011, while the rainfall in the early-monsoon season was underestimated due to less seasonal predictability of SST in the Indian Ocean.