Evaluating multi-satellite VIIRS active fire observations as a burned area proxy across North America

Wildfire activity fundamentally regulates terrestrial carbon dynamics and atmospheric emissions, necessitating timely and spatially complete mapping of burned area. Traditional reflectance-based burned area products provide robust spatial records but often have multi-week to multi-month processing latencies that limit near-real-time applications. Active-fire detections provide low-latency observations, but their use as a burned area proxy is constrained by temporal sampling gaps and commission errors associated with sub-pixel fire detection. This study represents one of the first applications to use the full three-satellite Visible Infrared Imaging Radiometer Suite (VIIRS) constellation, comprising S-NPP, NOAA-20, and NOAA-21, to address these limitations. Here, we evaluate how the spatial performance of a cumulative VIIRS active-fire approach varies with temporal sampling, latitude, and fire regime. We developed a VIIRS Cumulative Active Fire (VCAF) approach using the 375-m VIIRS active-fire record across North America from 2012 to 2024, and evaluated its spatial agreement with visually interpreted reference labels and the North American Fire History (NAFH) dataset, using MCD64A1 as a benchmark burned area product. Results from the first full year of the three-satellite configuration (2024) indicate that VCAF had higher spatial agreement than MCD64A1 in parts of the high-latitude domain (>50°N), with particularly high patch-level recall for small reference burned patches (<100 ha) while maintaining high recall for large patches (>1000 ha). The observed performance varied substantially with latitude and fire regime. In high-latitude boreal regions, the greater potential sampling frequency of the VIIRS constellation was associated with lower omission errors and greater spatial completeness in some regions, although environmental conditions such as persistent cloud cover remained important limitations. In contrast, VCAF systematically overestimated burned area in fragmented lower-latitude fire regimes. These results indicate that multi-satellite VIIRS active-fire observations can provide a potentially useful burned area proxy in selected high-latitude fire regimes and highlight the importance of accounting for temporal sampling and fire-regime characteristics when interpreting active-fire-based burned area estimates.