Table1_Extended-Range Forecasting of PM2.5 Based on the S2S: A Case Study in Shanghai, China.DOCX

Air pollution has become one of the most challenging problems in China, especially in economically developed and densely populated regions such as Shanghai. In this study, the long short-term memory (LSTM) model is introduced for the application in extended-range forecasting of PM2.5 in Shanghai by incorporating three members of the Subseasonal-to-Seasonal Prediction project (S2S) forecasting, moderate-resolution imaging spectroradiometer (MODIS) aerosol optical depth (AOD) data, and large-scale circulation factors derived from ERA-5 reanalysis. Therefore, an accurate ∼40-day PM2.5 prediction model over Shanghai was developed, providing new insights for air pollution extended-range forecasting. The new model not only exhibited much better accuracy but also captured the pollution process more closely than traditional methods, such as multiple regression (MLR). The prediction root-mean-square errors (RMSEs) based on the China Meteorological Administration (CMA), the U.K. model, and the European Centre for Medium-Range Weather Forecasts (ECMWF) were 24.84, 24.35, and 22.27 μg m−3, respectively, and their Heidke Skill Scores (HSSs) were between 0.1 and 0.5. As a result, the S2S-LSTM model for extension period pollution prediction with higher accuracy developed in this study could further burst the hot spots of pollution extended-range prediction research. However, limitations of the prediction model are still in existence, especially in dealing with only a single site instead of a two-dimensional prediction, which requires further investigation in future studies.