This paper describes efforts to develop aerosol data assimilation capabilities for NOAA’s regional air quality modeling system by assimilating PM_2.5 observations within the Joint Effort for Data Assimilation Integration framework. Results from the September 2020 Western U.S. wildfires show that assimilating AirNow and/or PurpleAir PM_2.5 data reduces mean absolute errors of 1–24 h PM_2.5 forecasts over the continental United States on average relative to a control without data assimilation.