As part of the EO AFRICA R&D MAPSPADES project (https://www.eoafrica-rd.org/research/research-projects-2023-2024), this dataset describes 877 high-resolution images downloaded from Google Earth Pro and classified according to the degree of desertification observed at each location. The locations are spread over the steppic region of Algeria, which covers more than 270,000 km2 (about 11% of the territory) and stretches more than 1200 km from the east to the west of the country. This region is particularly affected by land degradation and loss of indigenous plant biodiversity due to many factors, including human activities and climatic variations.

The purpose of this dataset is to provide "ground truth" data for training and validation of various machine learning or statistical models that study desertification or vegetation conditions from Earth observation data. Each scene represents an area of 30m by 30m and is documented with its coordinates, the date of image acquisition, and the observed desertification class (extremely severe, severe, moderate, light, none). These labels were obtained by averaging desertification scores ranging from 0 to 10, manually assigned by 3 people, and then classifying them with equal intervals.