Statistical Seasonal Forecasting of Winter and Spring PM_2.5 Concentrations Over the Korean Peninsula

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Climate Science

Background:

• Year-to-year variations in Korean fine particulate matter (PM2.5) concentrations are influenced by climate variability.
• Long-term PM2.5 data (2005-2019) for Korea were compiled using national and Seoul government observations.

Purpose of the Study:

• To develop a predictive model for winter and spring PM2.5 concentrations in Korea.
• To achieve 1-3 month lead time forecasts for seasonal PM2.5 levels.
• To identify key climate variables influencing PM2.5 variations.

Main Methods:

• A multiple linear regression model was developed using slowly varying boundary conditions.
• Forward selection stepwise regression identified significant predictors: sea surface temperature (SST), soil moisture, and 2-m air temperature.
• The model was validated using historical PM2.5 observational data.

Main Results:

• The wintertime (DJF) PM2.5 prediction model, using SST and soil moisture, achieved a 0.69 correlation with observations.
• The springtime (MAM) model, incorporating North Pacific SST and East Asian air temperature, showed improved accuracy with a 0.75 correlation.
• A linear relationship was found between seasonal mean PM2.5 and the frequency of high PM2.5 concentration days.

Conclusions:

• The developed multiple linear regression model effectively predicts seasonal PM2.5 concentrations in Korea.
• Climate variables, particularly SST and soil moisture, are crucial for forecasting PM2.5.
• The model's findings can aid in air quality management and public health advisories.