Sentinel-2 MultiSpectral Instrument（MSI）データを用いた2025年大船渡市大規模林野火災の森林焼損度評価

抄録

The spatial distribution of burned areas and severity in Ofunato City, Japan, was estimated, using 2025 Sentinel-2 MultiSpectral Instrument (MSI) surface reflectance data. The Normalized Burn Ratio (NBR) was derived from MSI data acquired in the pre-fire (15 February 2025) and the post-fire (13 May 2025) periods. Snow-covered areas were excluded from the pre-fire NBR. The difference between pre- and post-fire NBR was calculated to obtain the severity level (dNBR). Damage severity was visually interpreted from field survey photographs, and the agreement between the dNBR-derived and the field-based severity level was quantified. High- and moderate-high-severity areas based on dNBR corresponded to field-based damage severity ranging from extreme to moderate-low. The good correspondence of these two dNBR severity levels is attributed to the fact that, in evergreen conifers, canopy loss and extensive defoliation enabled satellite observation of the ground surface. In contrast, dNBR severity levels of moderate-low, low and no damage were underestimated. This is attributed to the presence of abundant green foliage in the canopy of evergreen conifers at these severe levels, which limits the ability to observe the ground surface from satellite.