A global small island remote sensing classification dataset based on synthesis from 2020 to 2024

This study is based on the GEE platform, combined with Sentinel-2 satellite data, using image normalization technology to enhance the recognition of smaller islands, and using support vector machine classification algorithm to extract ecological information from global islands. Using global island vector data provided by the United States Geological Survey, the Institute for Environment and Natural Resources, and the United Nations Environment Programme's World Conservation Monitoring Center as island boundaries, a total of 269391 small islands were selected for ecosystem classification. The dataset contains 1599 blocks and a color mapping table. The data size is 16.1GB. Island ecosystems are divided into five categories: water bodies, vegetation, urban areas, shallow reefs, and bare land. However, in this dataset, the water bodies were subjected to null value processing, resulting in only four classifications presented in the dataset. The TIF images extracted from this dataset can be viewed, read, and statistically analyzed using remote sensing software such as ENVI, Arc GIS, QGIS, etc.