Including Small Fires in Global Historical Burned Area Products: Promising Results from a Landsat-Based Product

Abstract

State-of-the-art historical global burned area (BA) products largely rely on MODIS data, offering long temporal coverage but limited spatial resolution. As a result, small fires and complex landscapes remain underrepresented in global fire history reconstructions. By contrast, Landsat provides the only continuous satellite record extending back to the 1980s, with substantially finer resolution. However, its use at a global scale has long been hindered by infrequent revisit times, cloud contamination, massive data volumes, and processing demands. We compared MODIS FireCCI51 with the only existing Landsat-based global product, GABAM, in a mountainous region characterized by frequent, small-scale fires. GABAM detected a higher number of burn scars, including small events, with higher Producer’s Accuracy (0.68 vs. 0.08) and similar User’s Accuracy (0.85 vs. 0.83). These results emphasize the value of Landsat for reconstructing past fire regimes in complex landscapes. Crucially, recent advances in cloud computing, data cubes, and processing pipelines now remove many of the former barriers to exploiting the Landsat archive globally. A more systematic integration of Landsat data into MODIS-based routines may help produce more complete and accurate databases of historical fire activity, ultimately enabling improved understanding of long-term global fire dynamics.