Fine-resolution gridded landscape fuel data in Ukraine

In this study, we present a map of 84 landscape fuel classes for Ukraine, parameterized according to the widely used Fuel Characteristic Classification System (FCCS). The developed fuel map has a resolution of approximately 30 meters and is relevant for 2021. Two-step methodology for the classification and mapping of landscape fuel was applied. Initially, general fuel types were defined using data on land use/land cover, canopy height, and forest/shrub percent cover. These fuel types were then divided into the final number of classes. To this end, we processed catalogues with descriptions of biotopes and species diversity for each ecoregion of Ukraine, and involved Digital Elevation Model data, hydrological basins data and geospatial information on settlements. FCCS includes a large amount of input parameters, enabling the calculation of fire potential and a number of important fire behaviour parameters. Unfortunately, there are no sufficient measurements to parameterize all of input characteristics for the created fuel classes, or fuelbeds according to the FCCS. However, most parameters were managed to reproduce using various information sources, including field surveys, catalogues of biotopes, ecological literature, digital photo series etc. Other parameters (mainly surface woody fuels and duff) were extracted from the corresponding fuelbeds existing in the Fuel Fire Tools (FFT, fire management application that integrates several modules, including FCCS) database. General description and climate type specification along with previously defined parameters were matched to select most appropriate fuelbeds.

To validate the developed fuel data, above-ground biomass (AGB) and total available fuel loading were calculated through FFT software and compared to the ESA CCI (European Space Agency, Climate Change Initiative) global forest AGB dataset for 2021 and 300-meter global fuel map for 2015 developed by Pettinari and Chuvieco. In overall, the created fuelbeds were found to underestimate mean living woody biomass for the sampled pixels (53.95 t/ha against 67.46 t/ha). At the same time, correlation coefficients are equal to 0.89 and 0.86 for Pearson and Spearman correlation, respectively. Upon closer examination, biomass in shrubs, tree scrub and young forests was underpredicted to a greater extent, while better accordance was achieved for mature forests, particularly for open ones. The created fuel dataset also showed a good agreement with the global fuel map for both above-ground biomass and fuel loading.

The gridded landscape fuel data in such a high resolution, developed in this study, are extremely important for a wide range of scientific and applied tasks, including fire management, evaluation of emission rates and modelling of smoke effects from wildfires. It should be noted that this data can be reclassified to ensure compatibility with the FireEUrisk fuel map for Europe.