Weather satellites for wildfire tracking: Mapping burned area in near-real time using geostationary satellite data

America’s operational geostationary weather satellite mission, GOES-R, delivers multispectral images of the United States every five minutes. Together, these highly frequent images form a detailed time-series of surface reflectance which can be used to monitor wildfires in near-real time. Conventionally, polar-orbiting satellites like Landsat and MODIS are used for mapping burned areas at high spatial resolution. Yet reliance on these high-resolution images often comes at the cost of time. Days or weeks may pass before polar-orbiting satellites get a clear view of the burn scar unobstructed by clouds or smoke. During these delays, the unstable post-fire zones are vulnerable to flash floods and debris flows, threats which forecasters can only predict with burn severity images. This study explores whether GOES-R, despite its low spatial resolution, could provide burned area maps that would improve the timeliness of post-fire risk warnings. The GOES-R near-infrared and shortwave infrared channels are used to compute the Normalized Burn Ratio (NBR) spectral index, the industry standard for monitoring burn severity. In this study, I modeled GOES-R time-series for 132 wildfire events with harmonic functions to estimate midday NBR before and after each fire. GOES-R measured midday NBR as consistently as MODIS. The NBR values were correlated strongly (r = 0.90), though GOES-R overestimated MODIS NBR by a factor of 1.4. A further advantage of the dense time-series of GOES-R imagery is its capability to investigate sub-diurnal variations in reflectance caused by the changing solar angle throughout the day. The daily NBR amplitudes significantly increased due to wildfire, meaning that post-fire areas exhibited a wider range in reflectance throughout the day than unburned areas. This increased surface anisotropy may result from heterogenous burned canopy texture. In summary, geostationary satellites show potential for near-real time monitoring of land surface disturbance, with burned-area mapping at the forefront of their capabilities.