2012 年 68 巻 4 号 p. I_439-I_444
The discrepancy of bias-corrected monthly temperature and precipitation data simulated by multiple General Circulation Models (GCMs) are compared among the bias-correction methods. This study classifies previous methods into 2 groups. The methods in one group consider about the statistical characteristics of the change of GCM outputs from historical simulation to future simulation, while the methods in the other group do not. The difference of the bias-corrected data is not large in the validation using the historical GCM outputs and observed data. The difference in the future GCM outputs is, however, large among the two methods. The variance of the increment of bias corrected-data from historical period to future period among GCM outputs are also compared between the two methods. Our proposed method, which is one of the methods that consider the characteristics of the change, is better than other methods because the variance of our method is smaller than that of other methods.