The FireSpread_MedEU dataset contains high resolution wildfire propgation data for 103 wildfire events in the Mediterranean area and Europe between 2017 and 2023. Each event is represented by varying amounts of fire propagation steps (at least two), depending on its final size and duration. Using the individual burned areas of all propagation steps of a wildfire, spatiotemporal fire progression can be reconstructed over the course of mutliple days in high accuracy. 

For the 103 wildfire events, a total of 320 individual propagation steps were reconstructed from Planet Lab Inc. satellite images. Planet offers daily updated high resolution data covering four bands in the Blue, Green, Red and Near-Infrared (NIR) region. Using a semi-automated approach in Python, we extracted initial burned areas from these images based on NIR thresholding. Afterwards, we manually refined every burned area in QGIS by optically comparing the retrieved burned area with their respective satellite image displayed in different band combinations. This way, we were able to draw highly accurate borders for every available fire day. In some cases, smoke or clouds prevented a clear view on the exact boundary, leading to potential offsets of real and drawn burned area border. This is indicated in the additional features of the dataset. Depending on the availability of clear sky satellite data, the duration between the propagation steps varies. In the best case scenario, fire spread is updated daily (63% of all propagation steps). 

FireSpread_MedEU is provided in the shapefile format (EPSG:3035). It contains burned area geometries of each propagation step, from which the spatial and temporal evolution of a wildfire event can be reconstructed (in accordance with the accuracy limits of the underlying satellite data). The dataset contains additional information regarding day and time of satellite image processing, smoke or cloud obstruction, size of the burned area, an ID that links the fire to the EFFIS burned area database, and a subjective quality assessment by the authors. The latter is based on the level of cloud and smoke obstruction in the image and can be used to filter the burned areas according the users needs. A detailed explanation of the features is provided in the pdf that accompanies the zipped shapefile that contains the data. 

In the updated version, one additional fire and the last available pre-fire dates of every wildfire event were added to the dataset. Additionally, missing days in-between fire propagation steps are are also included by providing the reason for their absence in the Info column of the dataset. In the updated version, every entry with a burned area also contains the fractions of land cover classes that were burned during each fire day. These classes are based on the United Nations Food and Agriculture Organization’s (UN FAO) Land Cover Classification System (LCCS) and the respective data was downloaded from the Copernicus Climate Data Store (https://cds.climate.copernicus.eu/datasets/satellite-land-cover?tab=overview). For every entry, the respective year was used.

This work was performed during the TREEADS project, which has received funding from the European Union’s Horizon 2020 research & innovation programme under grant agreement No 101036926. Content reflects only the authors’ view and European Commission is not responsible for any use that may be made of the information it contains. Planet Labs Inc. data provided by the European Space Agency (ESA). Several burned areas were refined with Planet data provided by the Federal Agency for Cartography and Geodesy.