A novel statistical-dynamical method for a seasonal forecast of particular matter in South Korea

Abstract

Societal concerns about air quality in East Asia are still growing despite country-level efforts to reduce air pollution emissions. In coping with this growing concern, the government and the public demand a longer-lead forecast of air quality to ensure sufficient response time until society prepares for countermeasures such as a temporary reduction of specific emission sources. Here we propose a novel method that produces skillful seasonal forecasting of wintertime (December to February) PM10 concentration over South Korea. The method is based on the idea that climate condition and air quality have co-variability in the seasonal time scales and that the state-of-art seasonal prediction model will benefit air quality forecasting. More specifically, a linear regression model is constructed to link observed winter PM10 concentration and climate variables where the predicted climate variables were furnished from NCEP CFSv2 forecast initialized during autumn. In this case, climate variables were selected as predictors of the model because they are not only physically related to air quality but also 'predictable' in CFS hindcast. Through analysis of retrospec-tive forecasts of 20 winters for the period 2001-2020, we found this model shows statistically significant skill for the seasonal forecast of wintertime PM10 concentration.