Caucasus Glacier Lakes Database
About the dataset behind the interactive glacier lakes map.
About the dataset
The database presents the first systematic, multi-temporal inventory of glacial lakes across the Greater Caucasus, compiled from Sentinel-2 (2018, 2025) and Landsat (2000) satellite imagery, Soviet topographic maps, field bathymetric surveys, and published historical records. A total of 1,384 lakes above 2000 m a.s.l. and with surface areas of at least 500 m² were identified. Of these, 983 were classified as extraglacial, water bodies of glacial origin in currently glacier-free valleys, while the remaining 411 lakes (supraglacial, proglacial, and periglacial) were subjected to a structured GLOF hazard assessment. Periglacial lakes dominate the inventoried types (90%), whereas proglacial lakes, though fewest in number, represent the most dynamic and hazard-relevant category. The total area of glacial lakes (excluding extraglacial lakes) is 5.05 km² ±0.79 km². The area of extraglacial lakes is 14.37 km² ±1.96 km². Sixteen new lakes have appeared since 2018, concentrated predominantly in the Central Caucasus. The spatial distribution reflects regional climatic and topographic gradients: the Western Caucasus holds the largest number and total area of lakes, while the Central Caucasus contains the highest proportion of glacier-proximal, higher-hazard lakes.
Hazard assessment was carried out using a six-criterion scoring system based on dam material, distance to glacier, lake area, presence of surface outflow, lake age, and shore-collapse hazard. Of the 411 assessed lakes, 19% were classified as high-hazard, 53% as moderate-hazard, and 28% as low-hazard. Basins of the Central Caucasus, including the Malka, Baksan, Chegem, Cherek, Urukh, and Ardon river systems, show the highest concentrations of moderate- and high-hazard lakes, consistent with the region's intensive contemporary glaciation and history of GLOF events such as the 2017 Bashkara and 2025 Maly Azau outbursts. Lake elevation was found to be statistically unrelated to hazard class, while lake area showed a significant positive relationship with hazard score.
The dataset provides an essential baseline for ongoing environmental monitoring, natural hazard management, and risk-reduction planning in the Caucasus region, particularly for the protection of downstream communities and the rapidly growing mountain tourism infrastructure. Future research directions include repeated satellite-based monitoring to track lake evolution, field-based bathymetric surveys for volume estimation, and numerical modelling of potential GLOF scenarios for the highest-hazard lakes.
The dataset is available from PANGAEA (https://doi.pangaea.de/10.1594/PANGAEA.993983, https://doi.pangaea.de/10.1594/PANGAEA.993979, https://doi.pangaea.de/10.1594/PANGAEA.993970, Kidyaeva et al., 2026).
The research was supported by the Russian Science Foundation, grant no. 25-77-10049.
Project team:
Ekaterina Pavlyukevich – project leader, PhD, expert in modelling and monitoring of water discharge in high-mountain rivers of the Caucasus, glacial lakes, and glacial lake outburst floods.
Vera Kidyaeva – PhD, expert in hydrology of mountain lakes and modelling of hazardous hydrological phenomena: floods, debris flows, and glacial lake outburst floods.
Viktoriia Iudina – PhD, expert in the study and modelling of hazardous hydrological phenomena: floods, debris flows, and glacial lake outburst floods.
Alena Averina – expert in modelling and monitoring of high-mountain river discharge.
Afanasiy Gubanov – expert in mountain glaciology, including glacier evolution, glacier mass balance, avalanche nourishment, glacier geophysics, and moraine cover.
Taisiia Postnikova – PhD, expert in glacier modelling.
Inna Krylenko – PhD, expert in mathematical modelling of hydrological processes in river basins and channels, integration of hydrometeorological cycle models, and application of remote sensing methods for the study of water bodies.
Artem Gorbarenko – expert in machine learning methods and the application of neural networks to semantic segmentation, object detection, and time series forecasting in geophysical process modelling.