Impact of climate dataset variability on spatiotemporal modelling of PM2.5 concentration using deep neural network models

Meteorological parameters have been used as features in spatiotemporal PM 2.5 models owing to their influence on the formation, concentration, and dispersion of PM 2.5 particles. This study investigated the effectiveness of various meteorological datasets, resampled using different methods, in predicting PM 2.5 concentrations with optimized deep neural network (DNN) models. The results showed that a cubic-spline-resampled European Centre for Medium-Range Weather Forecasts (ECMWF) dataset, as a meteorological feature in the DNN models, produced the best results in predicting PM 2.5 values with MAE and RMSE values of 3.597 and 5.363 μg/m 3 , respectively. Day of the year, the aerosol optical depth, and temperature were the features with the most significant influence in predicting PM 2.5 concentrations based on their SHAP values for each DNN models. The distribution and correlations between resampled values of each resampling method and dataset combination were discussed. This study also discusses future exploration for improving the spatiotemporal PM 2.5 -concentration modelling process. • Different meteorological datasets were used to model PM2.5 . • Meteorological datasets were resampled using different methods. • Daily PM2.5 concentrations were predicted using DNN models. • Spatial and temporal prediction errors were presented. • Key features were determined using SHAP values.