Attribution of landscape fires to warfare vs peacetime drivers

Russia’s full-scale war against Ukraine has led to a significant increase in the number of landscape fires, a major part of which is localized in the combat zone. Such fires cause economic and ecosystem losses in Ukraine. They also lead to additional greenhouse gas (GHG) emissions that affect the global climate system and can therefore be monetized and claimed as reparations from the aggressor country, based on the cost of GHG emissions. This poses the task of attribution by distinguishing fires caused by military actions from those caused by natural factors or ordinary human activity typical of peacetime.

This study aims to develop and test a methodology for attributing landscape fires using the Fire Weather Index (FWI) – a fire hazard indicator that considers meteorological conditions, fuel moisture, and wind. “Attribution” in the context of this methodology is defined as the fraction of landscape fires caused by military actions relative to their total area.

Two approaches were considered for attribution: a historical analogue and spatial comparison.

The historical approach was rejected due to significant differences in weather conditions, land use, agricultural practices, legislative requirements, and the lack of detailed fire mapping in Ukraine before 2022.

Instead, a spatial comparison method was applied, based on the assumption that for the same land-cover type under the same weather conditions (FWI) during the same season, fire areas should be proportional across the entire territory of Ukraine in the absence of war.

The attribution methodology uses a geographic information system and distinguishes:
“buffer zone” (direct war impact) – a cumulative 30-kilometer buffer on both sides of the moving frontline in 2022; and
“controlled zone” – territory controlled by the Government of Ukraine, without ground hostilities and outside the buffer zone.

Initial data included shapefiles of (1) fire polygons (derived from Sentinel, MODIS, and VIIRS); (2) FWI raster (from the Copernicus Emergency Management Service for the European Forest Fire Information System, EFFIS); and (3) land-cover raster data (coniferous and deciduous forests, croplands, other).

First, areas under four land-cover types were selected within the controlled territory, approximately equal in size to the same types in buffer zone, and daily FWI classes were assigned to each fire according to its geolocation. The next step involved calculating the fractions of area under fire for each FWI class and land-cover type in both zones for every calendar season. As a result of comparing these relationships, a table of attribution coefficients (in percent) was obtained, demonstrating that most fires in the buffer zone can be attributed to the war for all land-cover types and seasons over the three-year period.

Summary. The developed approach allows estimation of the fraction of fires that can be directly attributed to Russian aggression, taking into account spatial and meteorological conditions without access to ground observations in combat zones, occupied territories, or mined areas. The methodology was used to calculate additional emissions in the “Climate Damage Caused by Russia’s War in Ukraine” reports (Lennard de Klerk et al., 2025) and is intended to serve as a basis for claimed reparations.