FIREWX OSSE for Wildfire Remote Instrumentation System Design and Impacts to Fire Behavior Model Forecasting

Rapid detection of wildfires is critically important for timely decision-making, whether for mitigation response to impending explosive wildfire spread to prevent destruction, and/or monitoring firefront advancement into nonthreatening regimes. In a perfect world, high spatial and temporal resolution satellite data would be available to the fire-fighting community. Currently available active midwave infrared (MWIR) thermal wildfire monitoring is an either/or proposition. Low earth orbit (LEO) and geostationary (GEO) satellites range in spatial resolution from ~375 m to 2 km and temporal revisits from several hours to as little as every 30 seconds. The need to improve on that performance has driven the National Oceanic and Atmospheric Administration (NOAA) to require that the next generation of GEO satellites, the Geostationary eXtended Observations (GeoXO) series, improve the fire detection resolution to 1 km from 2km, while maintaining high temporal frequency. Beyond GeoXO, however, a purpose-built geostationary fire imager is being proposed specifically to detect initiation and location of wildfires with a five-fold improvement in temporal resolution compared to five minute GEO standard while achieving ~300 m ground sample distance at 40°N. Additionally, a number of small-sat LEO constellations designed for wildfire detection and monitoring are currently in development and/or being proposed by private and public entities, with nadir ground sample distance of 80 m or better, and sub-daily revisits, with decreasing imagery latency as the number of satellites within the constellation grows. In order to assess relative performance of these concepts, an Observing System Simulation Experiment (OSSE) has been developed to simulate in-band 3.9 μm infrared wildfire signals at orbital altitude. Fires are generated in the wildfire-atmosphere coupled model WRF-FIRE, where the heat release is a function of dry and moist fuel load combustion and fuel distribution is spatially defined by fuel map data. Radiative emissions from burning and not-burning regions are simulated through off-line radiative transfer codes using WRF output atmospheric profiles and satellite orbital and instrumentation specifications. Simulation of infrared imagery from an emerging fire provides guidance to optimize infrared satellite instrumentation design for rapid and highly sensitive wildfire detection in a low-count environment. Additionally, the simulations can highlight strengths of LEO and GEO systems working individually and in concert, to aid in wildfire detection algorithm development and creation of actionable data products, such as fire spread model forecasting. The performance of fire spread model forecasting by ingesting fire location data has been an active area of study ^1,2,4,5. In this work, an OSSE has been developed for the initial 3 hours of the 2018 Camp Fire. Localization of fire perimeters and spots from the simulated imagery, along with error analysis, guides the input of new information in WRF-FIRE integration steps to assess forecasting skill. An experiment contrasting data latency periods shows a forecasting advantage for assimilation of detected spot fires occurring an hour after the initial ignition into WRF-FIRE, over WRF-FIRE without timely knowledge of the spot fires. In a four member ensemble run, the arrival of fire fronts at the Paradise, California wild-land urban interface can be anticipated 24.5 ± 9.5 minutes earlier by WRF-FIRE with assimilation of the spot fires than the arrival roughly at 16:36 UTC documented by weather radar data ³. Without knowledge of the spots, the firefront simulated by WRF-FIRE is still a few kilometers from reaching the WUI by the end of the run at 17:00 UTC. Next steps include deriving fire perimeters from the OSSE data for both GEO and LEO designs, to assimilate into WRF-FIRE to analyze fire spread skill, both in the spatial distribution and its dependence on satellite data latency for the perimeter information.

References

1. Kale, MP et al., 2024,”Operational Forest-Fire Spread Forecasting Using the WRF-SFIRE Model”, Remote Sens., 16, 2480.
2. Kochanski, AK et al, 2023, “Analysis of methods for assimilating fire perimeters into a coupled fire-atmosphere model”, Front. For. Glob. Change, 6, 1203578.
3. Lareau, NP et al, 2022, “Tracking wildfires with weather radars”, JGR Atmos, 127, e2021JD036158.
4. Szasdi-Bardales, F et al, 2024. “Integrating dynamic wildland fire position input with a community fire spread simulation: A case study of the 2018 Camp Fire”, Fire Safety Journal, 143, p.104076
5. Turney, FA et al, 2023, “Sensitivity of burned area and fire radiative power predictions to containment efforts, fuel density, and fuel moisture using WRF‐fire”, JGR Atmos, 128, e2023JD038873.