This dataset contains six periods of ice lake catalog data in the central Himalayan region in 1990, 2000, 2010, 2015, 2020, and 2022. The storage format is ESRI Shapefile, which can be used to characterize the spatio-temporal evolution characteristics of ice lakes in the study area. The ice lake boundary in 1990, 2000, 2010, 2015, and 2020 is based on Landsat remote sensing images, relying on the Google Earth Engine (GEE) platform, and automatically extracted using the Improved Normalized Difference Water Index (ENDWI) coupled with the Otsu dynamic threshold method; the ice lake data in 2022 is extracted and generated based on Sentinel-2 images. Data projection uses Albers equal area projection (Asia_North_Albers_Equal_Area_Conic, parameters Central_Meridian: 95.000000;Standard_Parallel_1: 15.00000000;Standard_Parallel_2: 65.00000000;Latitude_Of_Origin: 30.00000000)。This catalog refers to the RGI Glacier Catalog. In 1990, 2000, 2010, 2010, and 2015/2020, it is divided into three types: ice lakes, glacier contact lakes, and non-glacier contact lakes; in 2022, ice lakes will be divided into four types: ice lakes, glacier contact lakes, non-glacier contact lakes, and non-glacier supply lakes. The extraction results of all time phases are subjected to manual visual interpretation and vector editing to ensure data accuracy. This multi-time series glacial lake catalog can provide basic geographical data support for regional glacier change monitoring, glacial lake burst flood (GLOF) disaster risk assessment, and cryosphere related research.
Landsat Landsat image data from the U.S. Geological Survey website http://earthexplorer.usgs.gov/).
Sentinel 2 is downloaded from the Copernicus website https://dataspace.copernicus.eu/.
RGI7 data is downloaded from www. glims. org.
Relying on the Google Earth Engine (GEE) platform, the improved normalized Difference water index (ENDWI) coupled with Otsu dynamic threshold method is used to automatically extract it.
The extraction error of ice lake data sets has a significant scale effect. The relative uncertainty of the area of small ice lakes is higher, and the relative error decreases as the area of ice lakes increases. The ice lake vector boundary plane accuracy can reach ±0.5 pixels (±15m and ±5m). The percentage error of the relative area of the ice lake boundary corrected by manual visual interpretation is between 0% and 22%. The local cross-validation results with Zhang et al.(2023) ice lake dataset showed (R2=0.99) that the errors were concentrated around 0, and there was no obvious systematic deviation. Mountain shadows, lake ice and snow cover, and differences in image spatial resolution are the main sources of error. 10 m resolution Sentinel 2 Compared with 30m Landsat images, the image can effectively reduce the uncertainty of glacial lake boundary extraction.
| # | number | name | type |
| 1 | 2025YFE0211400 | Technical System Construction for AI-ready Transformation of Scientific Data in Cold and Arid Regions | National key R & D plan |
| 2 | 22ZD6FA005 | Rapid degradation of the cryosphere and its impact on regional sustainable development | Gansu science and technology research project |
| 3 | CSFSE-ZZ-2402 | Study on Mechanism, Early Warning, Prevention and Control of Snow-ice Compound Chain-induced Disasters in Alpine Regions | other |
| 4 | 300108266601 | Study on Dynamic Monitoring and Chain Evolution Mechanism of High-altitude Ice-rock Disaster Chain Based on Multi-source Collaborative Observation | other |
This work is licensed under
CC BY 4.0 (Creative Commons Attribution 4.0 International License).
CSTR:11738.11.ncdc.db7810.2026
10.12072/ncdc.db7810.2026
