This dataset contains the annual average and three years average PM2.5 (particulate matter) concentration during 2000-2015 for the area of the continental United States. All the datasets are in 1km of spatial resolution, and in ESRI Img format. The unit is ug/m^3. This method is generated using a machine learning-based geostatistical methods incorporating meteorological, anthropogenic and geographic factors. Please refer the following paper for the details:

@article{Liu2018,
  title={Improve ground-level PM 2.5 concentration mapping using a random forests-based geostatistical approach},
  author={Liu, Ying and Cao, Guofeng and Zhao, Naizhuo and Mulligan, Kevin and Ye, Xinyue},
  journal={Environmental Pollution},
  volume={235},
  pages={272--282},
  year={2018},
  publisher={Elsevier}
}