A reference Landsat burned-area dataset over Mediterranean Algeria (1984–2025) with an assessment of historical and new-generation products

Algeria, the country with the largest mean annual and cumulative burned area (BA) on the southern rim of the Mediterranean Basin, experiences recurrent fires with substantial ecological and socioeconomic impacts. Yet, like many countries, it lacks a long-term, spatially explicit dataset of fire perimeters. Forest-agency records provide valuable fire information but are incomplete and contain discrepancies in BA location and extent, limiting fire-regime characterization. Available satellite-derived BA products offer systematic mapping at broad spatial and temporal scales and serve as the best available surrogate. Here, following internationally standardized protocols for BA dataset generation and validation, we present ALGEBA (ALGEria Burned Area), a 30-m BA reconstruction for fire-prone Mediterranean Algeria, derived from the Landsat archive for 1984–2025. Spatial validation against 10-m Sentinel-2 reference data for 2017 and 2021 yielded Dice coefficients > 90% and absolute relative biases < 3.5%, while median temporal reporting accuracy was 28 days against MODIS active-fire detections (2001–2011) and 24 days against VIIRS detections (2012–2025). Comparisons with forest-agency records and seven historical and new-generation BA products highlight ALGEBA as the most consistent and temporally continuous reconstruction of fire activity in Mediterranean Algeria to date, offering a robust basis for fire-regime analysis and Mediterranean-scale comparisons. Forest-agency records accounted for 70.6% of ALGEBA's cumulative BA, with strong interannual agreement. FireCCI51 overestimated BA twofold relative to ALGEBA, contrary to its widely reported global and regional underestimation bias. Used as a regional benchmark, ALGEBA revealed close cross-product consistency between VNP64A1 and MCD64A1 across the MODIS-to-new-generation transition, whereas Sentinel-3 products diverge from FireCCI51.