A dataset of annual surface water extent in the Xijiang River Basin from 1986 to 2023

The Xijiang River Basin, located in southwestern China, is an important part of the Pearl River watershed. Water resources in the basin are of great value to regional ecological security and socio-economic development. This study takes the Xijiang River Basin as the study area and is based on the Landsat satellite data and GlobeLand30 land cover data. Traditional methods based on single water index threshold are prone to feature interference and often yield poor accuracy, so a fusion extraction method with multi-water index threshold is proposed. The new method has obvious advantages in removing the influence of feature interference. This method is used to construct a dataset of annual surface water extent in the Xijiang River Basin from 1986 to 2023 based on Landsat with a spatial resolution of 30 meters, including 38 periods of surface water extent vector data. The dataset is compared and verified with Google Earth high-resolution images sample points by random sampling, indicating an overall accuracy of 93.19%, and a Kappa coefficient of 0.8639. Further comparison with the European Union Joint Research Center (JRC) surface water extent data and in-situ measured surface water sample points showed that the overall accuracy relative to the JRC annual data exceeded 92%, with the Kappa coefficients above 0.84. All measured points are all within the 30-meter buffer range of the extracted surface water boundaries, demonstrating strong consistency. This dataset provides long-term time series data for water resource management and ecological protection in the Xijiang River Basin, and offers a scientific basis for assessing the impacts of climate change on basin ecology and supporting the sustainable use of regional water resources.