Indigenous knowledge strengthens early warnings for glacial lake outburst floods

Glacial lakes high in the Himalayas can appear remote and silent, yet when their unstable natural dams fail, they can unleash floods capable of racing hundreds of kilometres downstream. New research from Newcastle University suggests that protecting communities from these glacial lake outburst floods, or GLOFs, will require more than sophisticated monitoring equipment. In Bhutan’s isolated Lunana region, residents often recognised danger through sounds, vibrations and messages from relatives before official warning systems reached them. The findings reveal a critical gap between detecting a disaster and enabling people to escape it safely.

GLOFs occur as glaciers retreat and meltwater accumulates in lakes at or near their termini. These lakes may be held back by loose deposits of rock, sediment and ice rather than by stable bedrock dams. A sudden collapse, an avalanche into the lake, rapid melting or another disturbance can displace enormous volumes of water, producing a fast-moving flood wave. Because the water can gather debris and erode river channels as it travels, the resulting flood may destroy homes, bridges, farmland and hydropower infrastructure across a remarkably large area—sometimes as far as 400 kilometres from the source lake.

The Newcastle University study examined how people in Lunana experienced GLOF warnings and evacuations during events in 2019 and 2023. The research found that awareness of flood risk was generally high, partly because of sustained communication and outreach by the Bhutanese government. However, awareness did not always translate into the ability to respond quickly. Residents described warning and evacuation processes that could become confused or disorganised, particularly in a remote mountain environment where roads, telecommunications and escape routes may be limited or disrupted.

For many residents, the first indication that a flood might be approaching did not come from a formal early warning system. Instead, people reported hearing unusually loud river sounds or feeling vibrations in the ground. Such environmental cues are especially important in valleys where water levels can change rapidly and where the distance between a hazardous lake and downstream settlements makes every minute valuable. These observations also demonstrate that local knowledge can function as a form of real-time environmental monitoring, complementing instruments that may fail, lose power or be unable to communicate during an emergency.

Mobile phones played another decisive role. Calls from friends and relatives were among the most common ways residents received initial alerts and subsequent updates, while social media channels also helped circulate information. These informal networks can move warnings rapidly through communities, but they are vulnerable to power cuts, damaged infrastructure, poor reception and the circulation of incomplete or contradictory information. The researchers therefore argue that warning systems should not rely on a single communication channel. Instead, official alerts should be reinforced through several independent and resilient routes, including mobile networks, radio, community leaders, environmental signals and locally trusted messengers.

The study also identified problems in the final stage of disaster response: evacuation itself. Most participants said they sought additional information from relatives or friends before leaving, suggesting that people often needed confirmation before trusting or acting on an alert. Some residents recalled that departures felt chaotic and that they were not prepared with a “grab bag” containing essential supplies and valuable possessions. In a GLOF, hesitation can be dangerous because flood waves may move faster than people can organise transport, gather family members or retrieve necessities from their homes.

Sonam Rinzin, the Newcastle University doctoral researcher who led the study, said the findings show that early warning technology alone cannot protect mountain communities. The most effective approach, he argued, combines formal monitoring with local knowledge, diverse communication networks and regular evacuation drills. Practice can help residents identify safe routes, decide when to leave, support children and older people, and prepare emergency supplies before a crisis begins. It can also expose weaknesses in official plans, such as unclear responsibilities, inaccessible shelters or communication systems dependent on a single power source.

The research forms part of a broader collaboration involving Newcastle University, the Royal Government of Bhutan and Druk Holding and Investments. The team is developing a network of low-cost sensors designed to monitor conditions associated with GLOF hazards and their potential triggers. These instruments may provide data on changes in water levels, ground movement or other signals linked to instability. The researchers are also using ensemble modelling, which compares multiple simulations rather than relying on one predicted outcome, to estimate realistic ranges of flood behaviour and improve the design of downstream warning systems.

A central element of the programme is the website Himalayan Hazards, which is intended to deliver sensor data directly to communities exposed to mountain hazards. Researchers say that data will be most valuable when it is translated into clear, actionable information: whether residents should remain alert, prepare to move or evacuate immediately. Professor Rachel Carr of Newcastle University said warning systems must be designed around the realities of how people receive, understand and respond to alerts. The lessons from Lunana may extend well beyond Bhutan, as warming temperatures continue to destabilise high-altitude environments across the Himalayas and other mountain regions. In the race to reduce GLOF fatalities, the decisive technology may not be the most advanced sensor, but a system that communities trust, understand and are ready to use.

Web References: Newcastle University; Himalayan Hazards — www.himalayanhazards.org

References: International Journal of Disaster Risk Reduction, DOI: 10.1016/j.ijdrr.2026.106211