Processed Multisource PM₂.₅ Prediction Dataset Using Satellite AOD and Meteorological Data for Maryland Monitoring Stations

This dataset contains processed multisource environmental observations developed for machine-learning-based PM₂.₅ prediction in Maryland, USA. The dataset integrates ground-based PM₂.₅ and meteorological observations with NASA satellite-derived Aerosol Optical Depth (AOD) data for the Howard, Padonia, and Beltsville monitoring stations. Two temporal resolutions are provided: 60-minute and 30-minute datasets. The data were temporally synchronized, cleaned, and prepared for machine-learning analysis. The 60-minute datasets use hourly observations, while the 30-minute datasets incorporate temporally interpolated meteorological variables and matched AOD observations. The dataset is provided to support reproducibility and further research on satellite-assisted PM₂.₅ prediction and multisource spatiotemporal environmental data analysis.