Playing with Fire in a Struggling Agriculture: Uncovering the Environmental Cost through Spatial Analysis

The datasets consist of vectors in .shp format with attributes of the surface affected by wildfire or burned area. All the data have resulted from the Deep Learning U-Net approach based on Sentinel 2 MSI images. The satellite images are publicly available on the Copernicus Data Space Ecosystem website https://dataspace.copernicus.eu. The analysis covers the period from January to April 2024, identified as the early spring of 2024. The supporting datasets consist of: Polygon vector data for the wildfire resulted from the DL pixel classification on Sentinel 2 images: Fire_S2__ro_poly.shp Point vector data for the wildfire - centroid of the above polygons with the surface affected: Wildfire_ro.shp Polygon vector data delineating the burned area resulted from DL U-Net pixel classification on Sentinel 2: Burned_area.shp Point vector data for the burnt area - centroid of the polygons with the surface affected: Burned_area_point_centroid.shp Metadata: Spatial Reference: Stereo 70; EPSG 3844 Spatial Resolution: 10 meters Attribute field: Surface (hectares) Source: Pixel classification on Sentinel 2 MSI; © 2025 The authors Geometry: vector polygon and vector point We hereby confirm that all vector data regarding the wildfire and burned area are coming from our analysis based on Sentinel 2 MSI covering the time interval January to April 2024 for Romania. © 2025 The authors