Global glacial lakes have expanded markedly in recent decades, accompanied by many catastrophic glacial lake outburst floods. However, our understanding of glacial lake evolution and connections with global warming and glacier thinning remains insufficient due to fragmented studies and a lack of standardized mapping. Here we evaluated the glacial lake changes and their drivers, taking care to accurately map glacial lakes while excluding non-glacier fed lakes in the same geographic area. In 2020, a total of 71,508 glacial lakes were identified worldwide, covering an area of 21,770.9 $$\pm$$± 544.2 km². The total volume of 1280.6 $$\pm$$± 354.1 km³ is equivalent to a sea-level rise of 3.45 mm. Since the 1990s, glacial lake numbers, areas, and volumes increased by 54%, 11%, and 9%, respectively. The spatial evolution patterns align well with resolved large-scale, decadal glacier mass changes. Glacier-fed lakes extensively expanded due to glacier retreat and melting, while ice-dammed lakes were primarily affected by thinning ice dams, resulting in heterogeneous changes and underestimated outburst potential. Under the sustained glacier degradation scenario, the results of this study have implications for further evaluation of glacial lake water resources and outburst risks on a global scale.

Glacial lakes in the Himalayas are widely distributed. Since 1900, more than 100 glacial lake outburst floods (GLOFs) have originated in the region, causing approximately 7000 deaths and considerable economic losses. Identifying potentially dangerous glacial lakes (PDGLs) is considered the first step in assessing GLOF risks. In this study, a more thorough inventory of PDGLs was presented that included numerous small-sized glacial lakes (<0.1 km²) that were generally neglected in the Himalayas for decades. Moreover, the PDGL evaluation system was improved in response to several deficiencies, such as the selection of assessment factors, which are sometimes arbitrary without a solid scientific basis. We designed an optimality experiment to select the best combination of assessment factors from 57 factors to identify PDGLs. Based on the experiments on both drained and non-drained glacial lakes in the Sunkoshi Basin, eastern Himalayas, five assessment factors were determined to be the best combination: the mean slope of the parent glacier, the potential for mass movement into the lake, the mean slope of moraine dams, the watershed area, and the lake perimeter, corresponding to the GLOF triggers for ice avalanches, rockfalls and landslides, dam instability, heavy precipitation or other liquid inflows, and lake characteristics, respectively. We then applied the best combination of assessment factors to the 1650 glacial lakes with an area greater than 0.02 km² in the Himalayas. We identified 207 glacial lakes as very high-hazard and 345 as high-hazard. It is noteworthy that in various GLOF susceptibility evaluation scenarios with different assessment factors, weighting schemes, and classification approaches, similar results for glacial lakes with high outburst potential have been obtained. The results provided here can be used as benchmark data to assess the GLOF risks for local communities.

The Tibetan Plateau and its surroundings have recently experienced several catastrophic glacier-related disasters. It is of great scientific and practical significance to establish ground-based early warning systems (EWSs) to understand the processes and mechanisms of glacial disasters and warn against potential threats to downstream settlements and infrastructure. However, there are few sophisticated EWSs on the Tibetan Plateau. With the support of the Second Tibetan Plateau Scientific Expedition and Research Program (STPSER), an EWS was developed and implemented in the Sedongpu Valley, southeastern Tibetan Plateau, where repeated river blockages have occurred due to ice/rock collapse debris flow. The EWS collected datasets of optical/thermal videos/photos, geophone waveforms, water levels, and meteorological variables in this sparsely populated zone. It has successfully warned against three ice-rock collapse–debris flow–river blockage chain events, and seven small-scale ice-rock collapse–debris flow events. Meanwhile, it was found that the low-cost geophone can effectively indicate the occurrence and magnitude of ice/rock collapses by local thresholds, and water level observation is an efficient way to warn of river blockages. Our observations showed that there were no immediate meteorological triggers for the ice-rock collapses and associated debris flows. Several factors, such as the volume and location of the collapses and the percentage of ice content involved, influence the velocities of debris flows and the magnitude of river blockages. There are still two possible glaciers in the study area that are at risk of ice collapse. It is worth monitoring their dynamic changes using high-resolution satellite data and the ground-based EWS to safeguard the surrounding hydrological projects and infrastructure in this transboundary region.

A glacial lake outburst flood (GLOF) is a typical glacier-related hazard in high mountain regions. In recent decades, glacial lakes in the Himalayas have expanded rapidly due to climate warming and glacial retreat. Some of these lakes are unstable, and may suddenly burst under different triggering factors, thus draining large amounts of water and impacting downstream social and economic development. Glacial lakes in the Poiqu River basin, Central Himalayas, have attracted great attention since GLOFs originating there could have a transboundary impact on both China and Nepal, as occurred during the Cirenmaco GLOF in 1981 and the Gongbatongshaco GLOF in 2016. Based on previous studies of this basin, we selected seven very high-risk moraine-dammed lakes (Gangxico, Galongco, Jialongco, Cirenmaco, Taraco, Beihu, and Cawuqudenco) to simulate GLOF propagation at different drainage percentage scenarios (i.e., 25%, 50%, 75%, and 100%), and to conduct hazard assessment. The results show that, when any glacial lake is drained completely or partly, most of the floods will enter Nepal after raging in China, and will continue to cause damage. In summary, 57.5 km of roads, 754 buildings, 3.3 km² of farmland, and 25 bridges are at risk of damage due to GLOFs. The potentially inundated area within the Chinese part of the Poiqu River basin exceeds 45 km². Due to the destructive impacts of GLOFs on downstream areas, appropriate and effective measures should be implemented to adapt to GLOF risk. We finally present a paradigm for conducting hazard assessment and risk management. It uses only freely available data and thus is easy to apply.