Forest disturbance detection by using remote sensing and artificial intelligence in Africa

The dataset arises from the "Forest Disturbance Detection Using Remote Sensing and Artificial Intelligence in Africa" (EO4Forest) project, a collaboration financed by the European Space Agency involving Wrocław University of Environmental and Life Sciences in Poland and Lagos State University in Nigeria. Aimed at forest monitoring in the Ogun and Lagos states, it incorporates detailed maps for 2015, 2019, 2022, and 2023, with a 20-meter per pixel resolution to ensure precise land cover representation. A legend file elucidating established land classes accompanies the m

The data includes an extensive set of training and validation samples, which are crucial for the accurate use and evaluation of the Random Forest classification technique used to analyze the data. The data segments also offer maps of land cover and forest gain and loss, with a focus on recent updates for specific subareas in 2022 and 2023, which can be found in the NTR_ForestUpdate folder.

Derived from Sentinel-2 and Landsat-8 satellite imagery, the dataset uses the Random Forest algorithm to determine land cover types, highlighting its importance for environmental studies, particularly those related to deforestation, reforestation and afforestation. It also includes maps of forest change to indicate areas of forest loss and gain, shedding light on the dynamics of Nigeria's forest cover along with the geographical boundaries of the study area.

The methodology behind the dataset, including data acquisition, processing, classification and analysis, is detailed in Python scripts available in a companion GitHub repository. This approach ensures transparency and reproducibility, providing users with access to both the processed data and the methodologies generating those results, thus offering a robust framework for advanced forest monitoring.