A flash flood can transform a river within minutes. This is what communities near the Nepal-Tibet border witnessed on Wednesday.
A massive flash flood ripped through villages along the Bhote Koshi–Trishuli river system carrying mud, debris and fast-flowing water. It has claimed the lives of at least 469 people and left hundreds more missing, including 39 Australians. Preliminary investigations suggest the flood was caused by a major collapse of ice and rock high in the Himalayas.
During such disasters, every extra minute is valuable time to reach higher ground. So how can we ensure communities are prepared in the face of such rapidly developing hazards?
Our research suggests early warning systems may help.
How do early warning systems work?
Early warning systems use various tools, data and communication channels to detect natural disasters before they strike.
These systems do much more than send a text message after a flood or fire has started. They are designed to identify a potential hazard early, track it as it develops, and forecast where it may go. The most effective early warning systems communicate this information quickly, enabling the public and emergency responders to act.
These systems have saved lives. In the Swiss village of Blatten, monitoring enabled authorities to evacuate residents before a massive rock-ice avalanche swept through the region in 2025. In Iceland, monitoring and established emergency procedures allowed more than 800 people to escape sudden glacial floods during the 2010 Eyjafjallajökull eruption.
Tracking where a disaster travels and how it develops is crucial. In the case of floods, river-monitoring stations at strategic intervals can measure water level and discharge. Cameras and other sensors can provide information about flow velocity and debris. Combined with real-time flood modelling, these observations can indicate when a flood may reach downstream communities, how severe it could be, and which areas may be inundated.
This information also helps emergency responders. If authorities know one settlement may have 20 minutes before a flood arrives while another has 50 minutes, they can prioritise evacuations, position rescue teams, and identify roads and bridges that may soon become inaccessible.
In Nepal, emergency warnings were sent out but were not enough to save local villages.
The cascading disaster was triggered by the collapse of a massive glacier in Nepal’s Langtang national park at about 8:40am. Nepal’s Flood Forecasting Division soon received information that a major flood had entered the Bhote Koshi river system from Tibet.
Within minutes, more than 600,000 mobile warning messages were sent to the public. By then, the Syabrubesi monitoring station in northern Nepal had already stopped transmitting. As the flood moved downstream, authorities continued to track it and issue warnings.
Nepalese authorities have known about the risk of such a disaster occurring for years. A 2020 assessment by the International Centre for Integrated Mountain Development and the United Nations Development Programme identified 47 potentially dangerous glacial lakes across the region’s main river basins. This was based on data from satellite images and other remote-sensing technologies used to monitor glacier activity. High-risk areas can then be monitored using cameras, lake-level gauges, weather stations, and sensors measuring ground or ice movement.
The Nepali government has rolled out similar technology at Imja Lake, a fast-growing glacial lake in the Himalayas. Authorities installed sensors and automated sirens in 2016, but these have since fallen into disrepair.
The catastrophic Nepal flood reveals the limitations of early warning systems.
For communities close to the source, flood warnings simply came too late. This is why warning systems should be automated wherever possible. When sensors detect dangerous conditions, information should immediately reach forecasting and emergency authorities. Location-based mobile alerts must then be sent to people in threatened areas.
But mobile networks and electricity can fail, so phones cannot be the only solution. Sirens, radio and trained local task forces provide essential backup and ensure people know where to go and which evacuation routes to follow.
The monitoring network itself must be resilient. In Nepal, several weather stations stopped transmitting or were swept away during the recent floods. Monitoring infrastructure must be designed to keep functioning even if some instruments fail. This may involve establishing more strategically spaced weather stations, communication pathways, and backup power systems.
Early warning systems cannot prevent natural disasters or fully protect vulnerable communities. They must be used alongside hazard mapping, which identifies areas that are more vulnerable to a particular disaster. Resilient infrastructure and established evacuation routes are also vital to prepare for natural disasters.
Research must focus on what interventions are most effective and in what regions. In the wake of the recent Nepal floods, several important questions remain, including how closely river-monitoring stations should be spaced, and how much warning time different communities need to evacuate safely.
Communities may have only minutes before a disaster strikes. An effective early warning system is about making every one of those minutes count.
Prashidha Khatiwada does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.
At around 8:45am on August 26, Karbir Gaire, a customs and information officer at Nepal’s Rasuwa Customs Office, suddenly felt what seemed like an earthquake. He went outside, looked towards China and saw a huge flood accompanied by dust. He shouted, “The flood is coming, run!”
Gaire and others immediately ran uphill. About 10 seconds later, the muddy water swept away the structures around him.
Gaire later told local media outlet Ratopati those 10 seconds were decisive: people who recognised the danger and ran towards higher ground survived. Others who delayed or ran elsewhere were swept away, or went missing.
His experience captures a potentially life-saving truth: a warning only saves lives if it produces immediate action.
For Australia’s Nepalese community, including me and my family, this disaster is deeply personal. Families and friends across Australia are waiting anxiously for news of loved ones.
But beyond the immediate devastation, the disaster raises an important question: could lives be saved if people were trained to act immediately when a flood warning arrives?
I believe the answer is yes.
Disaster in Nepal
The devastating August 26 flood originated in the high Himalayas near the Nepal-China border and sent a massive surge of water through the Bhote Koshi and Trishuli-Narayani river systems.
Hundreds of people have died, and many remain missing in Nepal and Tibet. At least 39 Australians have been reported missing.
It is important not to portray this disaster simply as a failure of warning. Nepal has an established flood forecasting and early-warning system, including hydrological monitoring and mobile phone alerts.
The preliminary technical report from Nepal’s Flood Forecasting Division, as reported by the news media, provides an important timeline. Authorities received information at about 9am that a flood had entered the Bhote Koshi from the Tibetan side. Between 9:15 and 9:16, some 600,000 early-warning SMS messages were sent to vulnerable communities in Rasuwa, Nuwakot, Dhading and Chitwan.
The division says the timely dissemination of the warning helped many people reach safety.
The lesson, therefore, is not that early warning is useless. Rather, early warning must be connected to early action. For example, people need to understand what “evacuate” means.
A resident walks along a street covered in mud following flash floods and a landslide disaster in Devighat, in Nepal’s Nuwakot district on August 26, 2026.Prakash Mathema/Getty Images
How do people react in emergencies?
My research on emergency evacuation behaviour provides relevant insights. In one study, we surveyed 1,134 train passengers in Melbourne about how they would respond during an emergency evacuation. We found passengers were more likely to be reactive than proactive. For example, they were more likely to wait for instructions from station staff than to immediately move towards an exit themselves. They were also likely to follow what other passengers were doing.
In a second study, we surveyed 950 passengers, 500 at Melbourne Airport and 450 at Qingdao International Airport in China, about how they would behave during an emergency. Importantly, we found passengers’ perceptions of how they would cope were related to their perceived ability to evacuate safely. This suggests that knowing what to do – such as following evacuation procedures and leaving belongings behind – can influence how confident people feel.
Although we examined transport emergencies rather than floods, the lessons are relevant: people should not have to work out what to do for the first time when disaster is already unfolding.
They should already know when to leave, where to go and how to reach safety. This shows the public needs to be prepared before emergencies occur.
We need to prepare the public for floods with a simple, rehearsed response. When an extreme flood warning is issued, people in vulnerable areas should already know their local evacuation procedure. They should be familiar with how to leave the river area and move to higher ground. They should not wait to see the water, return for possessions or attempt to save a vehicle.
Technology is necessary, but not enough
Nepal needs continued investment in flood and rainfall monitoring, mobile alerts and community warning systems.
But the physical monitoring infrastructure is vulnerable: four automatic hydrological monitoring stations were reportedly swept away in Nepal. Warnings should still be possible through alternative sensors, satellites, cameras, sirens, radio, public-address systems and trained local networks.
Nepal already has experience with community-based flood early-warning systems that combine monitoring, communication and community preparedness. In these systems, rising floodwaters can be detected in near-real time and warning information is transmitted to trained local caretakers, who relay it to vulnerable downstream communities and relevant authorities through channels such as SMS, mobile phones and sirens.
Education determines whether the warning becomes action.
Recovery will take years
The immediate priority remains rescuing those who can be found, locating the missing, treating the injured, and providing food and medical assistance.
But recovery will extend far beyond the immediate rescue operation.
Homes, roads, bridges, schools and livelihoods will need to be restored, while survivors may need long-term psychological and social support.
Reconstruction should therefore aim not simply to rebuild what was lost, but to build back safer, with stronger infrastructure, resilient communication networks, better evacuation routes and better-prepared communities.
The tragedy of Nepal’s floods reminds us we cannot prevent every extreme event but we can improve people’s chances of survival. An effective early-warning system is measured not only by how many warnings it sends but also by whether people understand them and act in time. As one survivor’s experience shows, sometimes just 10 seconds can mean the difference between life and death.
For those who lost their lives, and for the families still waiting for loved ones, my deepest thoughts and prayers are with you. The most meaningful tribute to those lost is to learn from this tragedy and build a safer, better-prepared future.
Nirajan Shiwakoti receives funding from the Australian government and other research funding organisations for research on transport, mobility and emergency evacuation. He is of Nepali origin and has a personal connection to Nepal, where he has family and community ties. He has no financial or commercial interests in the technologies, organisations or infrastructure discussed in this article.
On August 26 a glacier collapsed in the Langtang region of Nepal near the China border, triggering an ice and rock avalanche. This apparently entered the Lhende Khola, a tributary of the Bhote Koshi river, where it either caused or contributed to a temporary blockage. When this burst, a catastrophic flash flood travelled almost 100 kilometres downstream.
A glacier collapses
There’s nothing to stop the avalanche down the Trishuli river valley
Did warm conditions play a role?
There is increasing interest in the role of unusually warm conditions and glacial permafrost instability. Warming can increase meltwater penetration into the crevasses of glaciers and weaken ice-rock bonds. British Antarctic Survey researchers reported unusually high ground temperatures two days before the collapse.
Levan Tielidze has received funding from the Australian Research Council.
Digital Storytelling Team does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.
Woolston Eyes in Warrington, a dumping ground for canal dredgings, has become best place in UK for black-necked grebe
A dumping ground for dredgings from the Manchester Ship Canal has become the best place in Britain for the rare and beautiful black-necked grebe, hosting 64% of Britain’s breeding population.
Since 2021, Woolston Eyes in Cheshire has been restored, extended and turned into a mosaic of freshwater pools, reedbeds, willow scrub and rough grassland, managed by the RSPB and Woolston Eyes Conservation Group.
For five years, dangerously overpowered e-bikes have been legally imported to Australia.
Many are effectively electric motorbikes, able to reach 50km/h – or even faster if modified. Riders – often young men – can get hold of one without needing a motorbike license. Injuries and deaths have soared.
Authorities have tried to police the issue on the streets. But the real problem starts at the nation’s borders and e-bike retailers. This month, the federal government announced plans for a national safety standard for e-bikes, with e-scooters to follow.
It’s an important, if overdue, step. But the standard will only work if it clearly distinguishes safe, useful and popular pedal-assisted electric bicycles (pedelecs) from overpowered and easily-modified devices sold under the same label of “e-bike” or for “off road” use only.
How did we get here?
Pedelecs – pedal-assisted e-bikes – are legal across Australia. These e-bikes can have motors of up to 250 watts, with the electric assist capping out at 25km/h. These e-bikes have become very popular as a way to get around cheaply and sustainably.
In 2021, a previous Australian government removed e-bikes from its definition of a road vehicle. This led to an unintended consequence: overpowered e-bikes could be imported as long as they were used on “private property”.
Imports of these high-powered vehicles surged. Speeds of over 50km/h are common. Some riders modify their vehicles to go even faster. It’s not legal for these overpowered vehicles to be used on public roads and paths. But the law is often ignored, especially by adolescents.
Over the past five years, states and territories have responded with their own laws and inquiries. But these laws are largely reactive, as this tier of government has no power over import rules.
Consumers have been left to navigate a confusing market. Many online and brick-and-mortar outlets sell overpowered bikes and modification kits without enough information about safety and local road rules. Police, schools and employers have been left trying to identify dangerous vehicles after they are already in use.
Overpowered or modified e-bikes aren’t meant to be used on public roads – but many end up there.Travis Morrison/Getty
Risks are real – but not identical
Public discussion of e-bikes has been dominated by images of battery fires and young people riding at high speeds and doing wheelies. These are legitimate reasons for concern.
Between 2016 and 2020, seven people died in collisions involving e-bikes and another seven on other electric micromobility devices such as scooters. Deaths appear to be increasing. Eight people died in e-bike crashes in Queensland and New South Wales in 2025. Injuries are on the rise too.
The problem is, these figures do not distinguish between overpowered or modified vehicles and safer, road-legal e-bikes, which have risks similar to conventional bikes.
The batteries in e-bikes can occasionally catch fire. But the biggest risk comes from high-capacity, low-quality, unbranded or incompatible lithium-ion batteries – particularly those used in overpowered e-bikes, according to official advice. Damaged batteries, dodgy replacement chargers and poor conversions also add risks.
In short, it’s not fair to equate a safe, compliant e-bike using a tested battery and compatible charger with an overpowered electric motorbike modified for greater speed or a cheap, poorly-made device.
What should the new standard include?
Product safety standards lay out the requirements a product must meet at a minimum before it can legally be sold in Australia.
The proposed new standard must clearly define e-bikes as pedal-powered vehicles ridden like a bike. If a vehicle performs like an electric motorbike with a throttle and no need to pedal, it must not be marketed as a bicycle.
The standard must cover the complete electrical system – not simply the battery cells.
Independent testing, certification, supplier traceability and workable recall processes should be required.
Certified replacement batteries and chargers must be part of the standard, to allow riders to safely replace batteries and chargers.
Online marketplaces based overseas must not be allowed to sell unsafe devices to Australian consumers. Importers and platforms should not be able to shift all responsibility to buyers.
The standard should specify that speed and power controls must be tamper-resistant and ban aftermarket modification kits. Bike frames, brakes and tyres must be suitable for the vehicle’s weight, cargo rating and performance.
Authorities may need a separate classification for higher-powered devices used for deliveries and require them to meet the safety, registration and licensing rules applied to motorcycles.
Clear point-of-sale information is equally important. Buyers should be told the device’s power and assisted-speed limits, where it can legally be used, the local road rules, and how to charge and store it safely.
New standards will not prevent every crash or fire. We will still need enforcement, consumer education and safe cycle paths. But over time, new standards will weed out the most dangerous products.
Higher-powered delivery e-bikes may need their own classification and set of rules.George Chan/Getty
E-bikes are important
Ongoing controversy around overpowered e-bikes has overshadowed the very real promise of e-bikes.
This simple technology is already cutting demand for oil. Riders use e-bikes to replace short car journeys, get to train stations, manage hills and longer distances, carry children and groceries and continue cycling as they age or lose some mobility.
Better regulation is urgently needed. But policymakers must make sure the promise of this technology isn’t undermined by a crackdown on illegal models.
Success will mean dangerous devices become harder to buy – and safe e-bikes become easier to trust.
Richard J. Buning has received funding from the Australian Urban Research Infrastructure Network, Queensland's Department of Transport and Main Roads and the iMOVE Australia Cooperative Research Centre.
