Glacial lakes, as characteristic hydrological and geomorphological units formed by glacial ablation, have exhibited rapid expansion trends in alpine and polar regions worldwide in recent years. The systematic inventorying of glacial lakes documenting their spatial locations and geometric parameters, holds significant scientific value for understanding glacier evolution, responses to climate change, freshwater resource assessment, and early warning of glacial lake outburst floods (GLOF) hazards. To address persistent challenges such as ongoing debates over the delineation of potential glacial lake distribution areas and the heavy reliance on manual visual interpretation in remote sensing -based glacial lake extraction, this study developed an innovative approach that integrates mountain units with glacier boundaries to define potential distribution areas of glacial lakes. Based on Sentinel-2 MSI imagery and the Attention DeepLab V3+ deep learning model, we achieved high-precision, fully automated extraction and systematic inventorying of glacial lakes across the entire Tibetan Plateau. The results indicate that there were 29,089 glacial lakes with an area ≥0.005 km², covering a total area of 2020.65 km²on the Qinghai–Tibet Plateau in 2023. These lakes are primarily distributed in the southeastern part of the plateau, including the Nyainqêntanglha Mountains, the Eastern Himalayas, and the Hengduan Mountains. In terms of size structure, small glacial lakes (0.01-0.05 km²) account for the largest proportion numerically (62.66%) yet contribute only 20.39% to the total area. In contrast, large glacial lakes (>1 km²), though rare in number (0.68%), but contribute 24.22% of the total lake area. Regarding lake types, non-glacier-contact lakes dominate (73.88%), followed by moraine-dammed lakes (24.27%), while supraglacial lakes represent the smallest proportion (1.85%). The area uncertainty caused by the spatial resolution of the remote sensing imagery is estimated at 134.88 km², accounting for 6.68% of the total glacial lake area. Overall, this dataset not only accurately reflects the current status of glacial lake resources on the Qinghai–Tibet Plateau, but also provides essential data support for key applications such as GLOF risk assessment and regional hydropower development planning.
Tian et al. (Sun,) studied this question.