Uncertainty Datasets for the Landsat Burned Area Product Across the Southeastern United States

Characterizing spatial uncertainty in satellite-derived burned area products is essential to accurately represent fire activity and manage fire-prone ecosystems. However, most product evaluations rely on summary statistics, where their use to adjust burned area estimates is unclear. This dataset provides a comprehensive characterization of the spatial and temporal uncertainty in the Landsat Burned Area (LBA) product across the southeastern United States (1984-2024), a region with a high abundance of small, low severity fires. Landscape and temporal drivers were used to spatially project error rates and adjust estimates of burned area, accounting for all and unrelated error (i.e., missing or falsely mapped fires). This approach produced spatial uncertainty surfaces and estimates of error-adjusted burned area that reflect variability in LBA accuracy. The resulting data enhance our understanding of fire mapping in the southeastern US and enables more robust and informed use of the LBA products in fire management, emissions modeling, and prescribed fire planning.