Enhancing Wildfire and Smoke Forecasting by Integrating Fire Observations: A Comparative Analysis of Methods for Integrating Infrared and Satellite Data Into a Coupled Fire‐Atmosphere Model

Abstract Accurate forecasts of fire spread and smoke impacts using coupled fire‐atmosphere models require advanced methods for fire initialization. This paper proposes and tests three different methods for integrating fire observations. The methods reconstruct the initial fire evolution needed for the atmosphere spin‐up at the beginning of the simulation and identify actively burning fire regions. One method is based solely on IR perimeters, while the other two leverage IR perimeters and satellite detections. One uses only the most recent satellite data, while the other contextualized method uses two consecutive satellite detections to identify which sections of the fire perimeter experienced significant growth and which were inactive. This integration method addresses the problem of inaccurate identification of actively burning regions, thus correcting the previously seen overestimated growth in real‐time forecasts. This problem is fundamental when forecasting multiday fire incidents, which benefit from updating the state of the fire at the beginning of each forecast. The methods were tested within the WRFx fire forecasting system. The entire real‐time forecast for the 2021 Caldor Fire was rerun using the three methods and compared against observations. The analysis of the fire growth, as well as surface PM 2.5 concentrations and smoke heights, indicate that the contextual method utilizing both IR perimeters and satellite detections offers significant improvements when compared to the other methods for all variables analyzed, with the most considerable improvement in forecast skill seen in forecasts with 24–48 hr lead time.