"SAREdgeNet-Derived 2022 Tibetan Plateau Lake Extent (Shapefile Format)"

"This dataset consists of 2022 Tibetan Plateau lake extent data generated by the SAREdgeNet lightweight deep learning algorithm, presented in Shapefile format (a commonly used geospatial vector format). It is specifically designed to provide high-quality foundational data for remote sensing water body monitoring, climate change research, and algorithm performance validation.The core data of this dataset is derived from 288 Sentinel-1 satellite SAR images, integrated with 30m-resolution STRM Digital Elevation Model (DEM) data. It covers the entire Tibetan Plateau, spanning approximately 460,000 km\u00b2\u2014a region where lake area accounts for over 50% of China\u2019s total lake area and serves as a sensitive zone for global climate change. The data has undergone standardized preprocessing, including radiometric calibration, speckle noise filtering, and accurate registration. Additionally, it integrates SAR scattering features, topographic features, and geometric characteristics of lakes, effectively avoiding misclassification of dunes, shadows, and other non-water features as lakes, thus ensuring the accuracy of lake boundaries.The dataset includes the complete extents of various lakes on the Tibetan Plateau, ranging from small lakes to large ones such as Qinghai Lake. The annotation benchmark is based on manually refined NDVI-based water body data, with an overall extraction accuracy exceeding 96%. As vector data in Shapefile format, it can be directly used in mainstream GIS software (e.g., ArcGIS, QGIS) to support multiple operations such as lake area statistics, boundary analysis, and spatiotemporal change comparison, without the need for additional format conversion.This dataset provides reliable support for both researchers (for lake extraction model training and algorithm generalization validation) and relevant authorities (for water resource surveys and ecological protection assessments). Currently, the dataset is available upon request by contacting the corresponding author (email: [e-mail]). The source code of the supporting SAREdgeNet algorithm has been open-sourced (https:\/\/github.com\/Abolewaer\/SAREdgeNet), facilitating users to reproduce results or conduct secondary development."