Initial learning from the flash flood in the Rasuwa district of Nepal

On 26 August 2026, a flash flood on the Bhotekoshi river swept through the Rasuwa district of Nepal, killing more than 1,355 people and destroying everything in its path (Neupane, 2026). 

Preliminary assessments by the National Disaster Risk Reduction and Management Authority suggest that the devastating flood was caused by an avalanche in the Nepal–China border area, which initially blocked the Lhende river (Pokharel et al, 2026), with water backing up before it was suddenly released. The existing early warning system was designed to warn of monsoon floods and was overwhelmed by the speed, volume and force of the flash flood. The flood swept away many of the automatic flood-monitoring stations installed in the upper reaches of the Bhotekoshi river before they could transmit data to trigger warnings (Pokharel et al, 2026). While the Flood Forecasting Division records indicated that 679,295 early-warning SMS messages were disseminated through Nepal Telecom and Ncell to populations in identified downstream risk areas (Government of Nepal NDRRMA, 2026), it was found that the warning did not reach the people in Rasuwa in time. A rapid damage and needs assessment conducted by the Government of Nepal estimated the total disaster effects at US$ 2.69 billion, while the total recovery and reconstruction needs are estimated at US$ 4.77 billion (Government of Nepal NDRRMA, 2026).

The event reinforces the importance of adapting MHEWS to address emerging hazards and those exacerbated by the impacts of climate change. For example, the risk of glacial lake outburst flood (GLOF) in Nepal was highlighted in the 2025 GSR, which noted that there were only two functioning EWS in the glacial lakes region of Nepal, despite the country hosting 21 of the 47 dangerous glacial lakes in the region (UNDRR and WMO, 2025). In the same report, supraglacial lakes and permafrost thaws that increase the instability of mountain slopes were noted as emerging drivers of massive GLOF. However, this event was not a GLOF; rather, it was a large-scale collapse of rock and ice, which turned into a devastating debris flow and flood (Sahana et al, 2026).

While search and rescue work continues amid fears of further flood events and detailed technical assessments will follow, much can already be learned from the event. In particular, the Rasuwa flash flood highlights the challenge of implementing warning systems for cascading and compounding hazards. The tragic event highlights the importance of:

• Developing risk knowledge, especially identifying priority hazards at the local level, for example, risk assessment to identify high-risk glacial lakes.
• Establishing appropriate and sufficient monitoring and observation systems. Existing warning systems were tailored for monsoon floods rather than flash floods. The flood moved so quickly that monitoring stations inside Nepal were destroyed before they could transmit a warning. The community-level river gauges that gradually measured river behaviour were also too slow for a flood travelling at an average speed of 193 km per hour (Alexander, 2026). Seismic sensors that detect large mass movements would have picked up the avalanche/landslide, which was initially thought to be an earthquake (Haritashya, 2026). Cameras and other forms of remote sensing are essential tools for monitoring conditions in remote locations. Indeed, there are already plans to rebuild an EWS for landslides using seismic sensors, cameras and a satellite link (Singh, 2026).
• Designing systems that are robust and will still work even if infrastructure (especially power and communication systems) is damaged and have sufficient redundancy in case of failures.
• Automated warning systems. In events like these, every minute counts, so sensors that automatically trigger warnings will save more lives.
• Dissemination across multiple channels. Sirens and loudspeakers can trigger immediate action by communities at risk while messaging through radio, and CB has greater reach and can provide more details, including how to respond (Haritashya, 2026).
• Trusted local, informal networks. While informal networks can be overlooked in formal plans, for this event, lives were saved downstream as a result of warning messages sent by families and friends. For example, a headmaster evacuated his school minutes before the flood destroyed the school building as a result of a relative upstream calling the school accountant to warn of the impending flood (Alexander, 2026).
• Knowing how to act and being able to do so. When a siren demands immediate action, people must already know how to respond, for example by relocating to higher ground. Yet mobility is essential and some people, especially the very young, elderly and people living with disability, who may struggle to evacuate in time, need assistance. Initial reports suggested that many elderly people and children were unaccounted for (Giri, 2026).
• Considering disaster risk as a fundamental data set for planning development, especially of critical infrastructure. The flood damaged 55 km of roads, 37 motorable bridges, 68 suspension bridges and 13 hydropower projects (Giri, 2026) demonstrating the importance of incorporating risk assessment into long-term infrastructure planning before decisions are made (Ghimire, 2026).
• Transboundary collaboration. Early warning in transboundary river basins depends on international cooperation. Monitoring can take place in one country while communities at risk are across the border. Rapid information-sharing and trusted communication between neighbouring countries are an essential part of the warning system (Costa, 2026). At the South Asian Parliamentary Dialogue on Environment, Climate and Sustainable Development held in Bhutan in early September, participants proposed “a regional disaster-preparedness and data-sharing mechanism covering shared early-warning systems, glacial lake and flood monitoring across Himalayan basins, and the exchange of climate-risk data across borders” (The Kathmandu Post, 2026).