Modeled air data are predictions, or estimates, of the levels of PM2.5 and ozone in the air. These estimated predictions are applied to areas that do not have air quality monitors and fill in time gaps when monitors may not be recording data. EPA provided the modeled air data for PM2.5 and ozone by statistically combining air monitoring data from the Air Quality System (AQS) Database with data from EPA's Community Multiscale Air Quality (CMAQ) model, using a statistical method called Hierarchical Bayesian modeling. The CDC used the results from EPA's Hierarchical Bayesian modeling to derive the modeled PM2.5 and ozone estimates for each county in the United States. Those county estimates are available on the air quality data query page.