Seasonal Forecast of Asian Dust over South Korea in Spring Season

This study aims to develop statistical models for the seasonal prediction of Asian dust in South Korea. The predictand is the average occurrence frequency of Asian dust in the spring season(March to May). Monthly data of Asian dust days at 28 sites in South Korea and four climate factors(rainfall amount, snow cover, surface air temperature and wind speed) in source regions of Asian dust during 1979 to 2007 are used. Three source regions that affect the Korean Peninsula are considered. Two statistical models, a multiple regression and a three-grade logistic regression model, are applied to generate the quantitative forecast and the ternary forecast(normal, below normal, above normal). Predictors are selected by the correlation analysis and stepwise regression method.