High-resolution (3km) forecasting of smoke and visibility for the US by ingesting the VIIRS and MODIS FRP data into HRRR-Smoke during August 2018

The High-Resolution Rapid Refresh with Smoke (HRRR-Smoke) model is based on the HRRR numerical weather prediction modeling system. The model simulates primary aerosols (smoke) from wildland fires in real time on 3km resolution grid over the entire CONUS domain. HRRR-Smoke estimates biomass burning emissions and simulates fire plume rise in an inline mode by using the fire radiative power (FRP) data from the VIIRS (onboard S-NPP and NOAA-20) and MODIS (Terra and Aqua) satellite instruments using as input data. The model includes the impact of smoke on radiation and visibility diagnostics.

The HRRR-Smoke model was used in forecasting of smoke from the numerous wildfires burning in the western US during August 2018. The model was initialized every hour by assimilating the latest meteorological datasets and ingesting the satellite FRP data. The RAP-Smoke model (13.5km resolution), which covers the entire North America, provided boundary conditions of smoke concentrations to HRRR-Smoke. Thus, HRRR-Smoke was able to capture the transport of smoke from Canadian wildfires to the CONUS domain. During summer 2018, two field campaigns (WE-CAN-C130 and BBFLUX) took place to conduct in-situ and remote sensing measurements by using aircrafts and mobile labs. The HRRR-Smoke forecasts are extensively evaluated by using the surface PM2.5 and Aerodyne mobile lab's speciated aerosol concentration measurements. The simulated smoke fields within the entire atmospheric column are compared against the VIIRS AOD data. The visibility and meteorological forecasts are validated against hourly METAR observations obtained by hundreds of the weather stations.

Several sensitivity simulations of HRRR-Smoke are performed to study the impact of the heavy smoke episodes on weather forecasting during August, 2018. The analysis of the model results revealed that the inclusion of smoke extinction can significantly improve both visibility and air temperature forecasts. In this presentation we also discuss August 19-20, 2018 episode, when very high (200-300 ug/m3) concentrations of PM2.5 were detected in Spokane, WA and surrounding areas. We present the fine-scale HRRR-Smoke simulations and their verification for this episode, and discuss how the modeling tool can be used to analyze the transport of smoke over complex terrain.