Particulate Matter Prediction and Shapley Value Interpretation in Korea through a Deep Learning Model

This study collected and analyzed data to predict particulate matter (PM) concentrations in Korea at regular intervals. Automated synoptic observation system data, real-time atmospheric from AirKorea, Geostationary Multipurpose Satellite – 2A were used. We also used deep learning, which is useful for PM predictions. The learning model a neural network (NN) of with diameter less than 2.5 μm (PM2.5) 10 (PM10). To illustrate the results NN model, we calculated Shapley value using eXplanable Artificial Intelligence (XAI) SHapley Additive exPlanations (SHAP) library. difference analysis according aerosols was explained. analyze contribution features each grid, SHAP values normalized. normalized clustered represented visually. PM2.5 PM10 classified into four clusters. next day's predictions both heavily influenced by weather variables western region, air quality more influential inland region. Unlike PM2.5, prediction southern region affected greater degree wind.