How different countries warn their people before disaster strikes
The systems that give people time to survive a disaster
Few days ago, a catastrophic flash flood tore through the Nepal-China border region, triggered by a glacier collapse and landslide high in the mountains. Hundreds of people were killed, entire villages were buried, and thousands more were left missing as rescue teams worked through dangerous, still-unstable terrain.
There are no adequate words for a loss on that scale. My thoughts are with everyone in Nepal affected by this disaster, the families waiting for news, the communities rebuilding, and the rescue workers risking their own safety to help others. If you are reading this from one of the affected areas, I am deeply sorry for what your country is going through.
A disaster like this leaves a painful question hanging over it: could more warning have saved more lives? It is a hard question, and not always a fair one, some events, like a sudden glacier collapse, give almost no notice at all. But it is also the question that has driven decades of work around the world to build systems that give people those precious minutes, hours, or days before catastrophe arrives.
This piece is a look at that work. Not at any single tragedy, but at how different countries, facing different threats, have built systems to warn their people and reduce the loss of life. It is written in the hope that understanding what works might, in some small way, contribute to fewer families facing what so many in Nepal are facing now.
What a warning system is really racing against
When a disaster hits, the difference between a tragedy and a near-miss often comes down to a single thing: how much warning people got, and whether that warning reached them in time.
A few seconds before an earthquake. A few days before a flood. That window is everything. It is the time to take cover, to move to higher ground, to evacuate a village.
Different countries, facing different threats, have built very different systems to buy their people that time. Some rely on dense sensor networks, some on satellites, some on massive physical barriers, and increasingly, many on AI. Here is a tour of how some of the most disaster-prepared countries in the world actually do it, and what the rest of us can learn from them.
Japan
No country has invested more in early warning than Japan, and for good reason. It sits on the meeting point of four tectonic plates and experiences roughly 1,500 earthquakes a year, along with tsunamis, typhoons and volcanic eruptions.
Japan’s answer is one of the most advanced warning setups on the planet, and it works on a clever piece of physics.
The earthquake early warning system exploits the fact that earthquakes send out two kinds of waves. A fast, weak “P-wave” arrives first, followed by the slow, destructive “S-wave” that does the real damage. A national network of over 4,000 seismometers detects the harmless first wave, and computers instantly calculate where the quake is and how strong the dangerous wave will be.
That gap between the two waves is the warning. It is often only seconds, but seconds are enough to:
- Automatically stop bullet trains before they derail
- Halt elevators at the nearest floor so people are not trapped
- Shut down factory equipment and gas lines
- Give people a moment to get under a desk or away from windows
- Alert surgeons mid-operation
J-Alert, launched in 2007, is the delivery system that ties it together. It is a satellite-based network that pushes warnings, for earthquakes, tsunamis, volcanic eruptions, severe weather and more, straight to sirens, TV, radio and every mobile phone in the affected area, within seconds.
The honest lesson from Japan is not only about technology, though. After the devastating 2011 tsunami, which killed close to 20,000 people despite a working warning system, investigators found that many died because they delayed evacuating, trusted seawalls, or returned home too soon. A warning only works if people act on it.
The Netherlands
The Dutch took a completely different philosophy. Much of the country sits below sea level, so instead of just warning people about floods, they set out to physically stop the sea from ever getting in.
The result is the Delta Works, widely considered one of the engineering wonders of the modern world.
After a catastrophic North Sea flood in 1953 killed over 1,800 people, the Netherlands spent the next four decades building an enormous network of dams, dikes, sluices, locks and storm surge barriers. In doing so, they shortened a jagged 700-kilometer coastline into a straight 80-kilometer wall of defense against the sea.
The most striking part is how automated it has become. Some of the giant barriers, like the one protecting Rotterdam, close themselves automatically when sensors predict the sea will rise past a dangerous level, no human decision required, though a government team watches around the clock just in case.
But the Dutch did not stop at concrete. They also built the brains behind the barriers:
- Delft-FEWS, an open flood-forecasting software developed by the Dutch research institute Deltares, combines live weather and water data to predict flooding in near real time. It has since been adopted in more than 50 countries.
- A national water center issues official flood alerts, informed by these models, so authorities can prepare days ahead.
The Dutch lesson is that warning and prevention work best together. You forecast what you can, and you build defenses for what you cannot afford to get wrong.
The United States
The US faces almost every kind of disaster, hurricanes, tornadoes, wildfires, floods, across an enormous and varied landmass. Its approach leans heavily on reaching people through devices they already carry.
Get Joshclarkyy’s stories in your inbox
Join Medium for free to get updates from this writer.
Wireless Emergency Alerts (WEA) are the backbone. These are the alerts that make your phone buzz with that distinct, jarring tone, for tornado warnings, flash floods, or an evacuation order. They are broadcast to all compatible phones in a targeted geographic area, so you get them even as a visitor who never signed up for anything.
Behind those alerts sits a web of specialized agencies:
- The National Weather Service issues watches and warnings for storms, floods and tornadoes.
- Tornado warning systems combine Doppler radar, trained spotters and outdoor sirens in high-risk regions.
- A dedicated tsunami warning system covers the Pacific and other coastlines.
- The broader Integrated Public Alert and Warning System (IPAWS) lets authorities push a single alert out across phones, TV and radio at once.
The US model shows the strength, and the limit, of the phone-first approach. It reaches enormous numbers of people instantly, but it depends on cellular coverage and on people not dismissing the alerts as noise.
The United Kingdom
The UK came to a national mobile-alert system relatively recently, but built a clean version of it.
The UK Emergency Alert System, rolled out in 2023, can send a loud, distinctive alert to every compatible phone in a defined area, even overriding silent mode, for life-threatening events like severe flooding or wildfires.
It works through cell broadcast, the same core technology behind alerts in Japan and the US. Rather than texting individual numbers, it broadcasts a message to every phone connected to the mobile masts in a target zone. That means it does not need to know your number, works even when networks are congested, and reaches everyone in the area at once.
Alongside it, the UK runs a long-standing flood warning service through the Environment Agency, which monitors rivers and coasts and issues tiered alerts, flood alert, flood warning, and severe flood warning, to registered residents and businesses.
The UK’s approach is a useful example for countries building a system now: you do not have to invent the technology from scratch. The cell-broadcast model is proven, and it can be stood up nationally in a way that reaches almost everyone.
The global layer: AI warnings that cross borders
The newest and arguably most important development is not tied to any one wealthy country. It is aimed squarely at the places that have historically had the least protection.
Google Flood Hub is an AI-powered flood forecasting system that predicts river floods up to seven days in advance. What makes it remarkable is its reach and its price: it is free, and it now covers areas home to over two billion people across more than 150 countries.
The technology behind it matters, because it solves a specific problem of poorer regions. Accurate flood forecasting traditionally needs a dense network of river gauges that many countries simply do not have. Google’s models use machine learning, satellite imagery and weather data to generate reliable forecasts even for rivers with little or no local monitoring, something that was not possible before.
The forecasts show up where people already look, on Google Search, Google Maps and Android notifications, and aid organizations like the Red Cross build them into their own warning efforts. In some pilots, families have received cash aid days before a flood arrived, giving them time to protect their homes or move.
This points at where early warning is heading: AI models that can protect people anywhere, not just in countries rich enough to build a national sensor grid.
What actually makes a warning system work
Looking across all of these, the same lessons keep surfacing. It turns out the technology is only part of the story.
A warning system saves lives when it gets four things right:
- Detection. Sensing the threat early, whether through seismometers, river gauges, radar or AI models.
- Speed. Getting from detection to a warning in seconds or minutes, not hours. Automation matters here, because humans are too slow in a crisis.
- Reach. Actually getting the alert to every person in danger, which is why cell broadcast, sirens and multiple channels all matter. One channel is never enough.
- Trust and action. People have to understand the warning and act on it. Japan’s 2011 experience is the hard proof that the best alert in the world fails if people hesitate.
That last point is the one technology cannot fully solve. A system can deliver a perfect warning to a phone in someone’s hand, but it cannot make them move. Education, drills and public trust are as vital as any sensor.
The bigger picture
The encouraging truth is that early warning works. Countries that invest in it, in the sensors, the delivery systems and the public education around them, measurably lose fewer people to the same disasters that once killed thousands.
The even more encouraging shift is that this protection is no longer only for wealthy nations. AI-driven systems like Flood Hub are extending days of warning to places that never had any, and the underlying models keep getting better and cheaper.
Disasters themselves are becoming more frequent and more intense, so none of this is a solved problem. But the direction is clear. The gap between a disaster striking and people knowing it is coming is shrinking, and every hour of warning that technology can buy is measured, ultimately, in lives.
