California. The Golden State. Home to almost 40 million people and the world’s fourth-largest economy, it is a global capital of technological innovation and the epicentre of the entertainment industry. But beneath the glitz and glamour, prosperity and prestige, the state is a natural disaster hotspot.
From earthquakes to wildfires, Californians are accustomed to battling the forces of nature. In 2025 alone, wildfires ravaged more than 500,000 acres of land statewide, killing 31 people and causing hundreds of billions of dollars worth of damage.
Wildfires – and how to prevent, mitigate and fight them – have dominated California’s disaster planning in recent years. And with good reason: the state has been in a constant state of drought for 25 years, only becoming drought-free for the first time this January. The last five years in particular have seen some of the worst fire seasons in living memory, with millions of acres scorched.
But California is built on extremes at both ends. Alongside vast wealth sits deep poverty; alongside years of drought and fire come episodes of torrential rain and devastating floods.
Only in 2023, hundreds of thousands were left without power as storms and flooding battered the state. Highways collapsed, trains derailed, and thousands were evacuated from their homes.
That sounds bad enough, but according to the latest research, episodes like this pale in comparison to what’s to come.
Scientists and officials are now preparing for not one threatening storm, but a 30-day maelstrom of megastorms unlike anything seen in the state for almost 200 years. Such a scenario was always possible, but rising global temperatures are making it more likely – and far more destructive.
“It was always a when, not if,” says Dr Daniel Swain, a climate scientist at UCLA, who co-authored the study warning of the coming storm. “Before global warming, that ‘when’ might have been centuries away. Now it’s quite likely to be within my own lifetime.”
This storm system, dubbed ‘ARkStorm 2.0’, could strike this year or in 60 years – no one knows for sure. Whenever it does, it is likely to be one of the most costly disasters in global history. The only question is whether California can prepare in time.
What is an ARkStorm?
Short for ‘Atmospheric River 1,000-year Storm’, the ARkStorms are hypothetical – yet very plausible – megastorm scenarios. Those familiar with weather on the US West Coast may already know atmospheric rivers (ARs) well.
These vast corridors of water vapour flow through the sky, transporting moisture from the tropics towards the poles. When they collide with land and mountain ranges, that vapour condenses into prolonged, heavy rainfall or snowfall.
One well-known example is the so-called Pineapple Express, a strong AR that funnels warm, moisture-laden air from near Hawaii towards the west coast of North America.
Though often strong rainfall events, ARs are important to water supplies in the region, contributing up to 50 per cent of California’s annual precipitation.
However, when these systems stall, or multiple hit in quick succession, they can dump extraordinary amounts of rain and snow in a short time, dramatically raising the risk of landslides, river flooding and reservoir overflow.
In 2010, a team of interdisciplinary scientists led by the US Geological Survey developed the first ARkStorm scenario: a sequence of super-strong atmospheric rivers striking California just days apart.
Far from being a speculative worst-case, the scenario was designed to mirror real historical events, including devastating storms in 1861-62 that lasted 45 days and turned hundreds of miles of the Central Valley into an inland sea.
Today, if an ARkStorm hit, it would be the most costly geophysical disaster of all time. In such an event, damages are expected to exceed $1tn, with flood depths in some areas up to 6m (20ft). Millions of people would need to be evacuated, as water invaded populated areas from San Francisco to Los Angeles.
In 2022, researchers revisited the ARkStorm scenario to factor in one – pretty important – factor that was left out of the original analysis: our old friend climate change. Suffice to say that if the first scenario was troubling, ARkStorm 2.0 was truly mind melting.
The mother of all ARkStorms
Climate change impacts rainfall events, and in particular extreme rainfall events, in two main ways.
First, warmer air can hold more water vapour. For every degree Celsius of warming, the atmosphere’s moisture-holding capacity increases by around 7 per cent, giving storms more fuel to work with.
Swain and his colleagues call this the “expanding atmospheric sponge effect”.
He says: “You can essentially just imagine an ordinary kitchen sponge that grows progressively larger with warming. It has the capacity to absorb much more water, so you can soak up much more water.”
This is why, under a warming climate, average precipitation will largely stay the same, and droughts and wildfires will still be commonplace because more water is locked up in the atmosphere, rather than on the ground.
“The air is demanding more water more rapidly from the landscape and drying it out faster,” Swain continues. “But of course, if you have a very large sponge that’s completely saturated and you wring it out, it yields more water. And that's analogous to the more intense precipitation.”
In ARkStorm 2.0 simulations, atmospheric rivers transport around a quarter more moisture in future climates than they do today, driving much heavier cumulative rainfall.
Second, warming raises the snowline. In the past, much of the Sierra Nevada mountain range’s winter precipitation fell as snow, temporarily locking water away in mountain snowpack and releasing it gradually during spring melt.
In a warmer climate, more precipitation falls as rain, and rain that lands on existing snow can rapidly melt it, sending large volumes of water downhill all at once.
A bigger sponge and higher snowline is what Swain aptly calls the “double whammy effect.”
“You get increased precipitation intensity, but also a large shift from snow to rain. What once would have fallen as snow and not contributed immediately to flood risk is now falling as rain and turning straight into runoff.”
Together, these effects drive dramatic increases in peak flows in rivers and waterways. In parts of the southern Sierra Nevada, ARkStorm 2.0 modelling suggests runoff could be 200-400 per cent higher than in historical megastorm equivalents.
From rare disaster to rising risk
Historically, floods on the scale of California’s Great Flood of 1861-62 were thought to occur perhaps once every century or two. That event turned the Central Valley into an inland sea nearly 300 miles long, drowning farms, towns and transport routes, and forcing the state capital to temporarily relocate from Sacramento to San Francisco.
But ARkStorm 2.0 modelling suggests that climate change has already shifted those odds. Compared with conditions in the early 20th century, the likelihood of a megaflood-capable storm sequence has more than doubled.
If emissions are not curbed, by 2060, catastrophic events that previously occurred once every two centuries are set to occur three times per century, the study suggested.
“Flooding has been an underestimated risk historically relative to more visible natural disasters like wildfires and earthquakes,” Swain says. “But it’s now a risk which is increasing over time, which is going to compound the consequences of that historical underestimation if we don’t do things differently.”
The Central Valley is the danger zone
If ARkStorm 2.0 were to strike, flooding would not be evenly distributed across the state. The greatest risks converge on the Central Valley, where runoff from the Sierra Nevada funnels into the Sacramento and San Joaquin river systems before spreading across some of the most heavily engineered floodplains in the world.
This vast low-lying region is both California’s agricultural heartland and a critical artery of its water supply system. It is criss-crossed by hundreds of miles of levees, canals, pumping stations and reservoirs designed to protect cities, farmland and transport corridors from seasonal floods.
But those defences were largely built for a climate in which winter precipitation arrived in shorter bursts and snowmelt dominated flows in spring.
“The San Joaquin Valley in particular really stands out to me as being of great concern,” Swain says. “It’s the breadbasket region, but it’s also got cities home to several million people, and it’s where much of the state’s water infrastructure that supplies Southern California is located… But it’s also where we expect the largest increases in flood severity and magnitude in a warming climate.”
In ARkStorm 2.0 simulations, runoff in key Sierra Nevada drainage basins often exceeds anything seen in historical megastorms, with little overlap between past and future extremes.
In practical terms, that means flood protection infrastructure may face water levels well beyond those it was designed to withstand – and for far longer than engineers ever anticipated.
Mapping a disaster that has never happened
To turn storm simulations into maps of real-world flooding, researchers need detailed inundation modelling – complex analyses that simulate how water would move across landscapes, interact with levees and spill into towns and infrastructure.
Those studies are now underway as part of the ARkStorm 2.0 programme, using national-scale flood models developed by Oak Ridge National Laboratory.
The aim is to generate high-resolution flood maps that state and local agencies can use for planning and emergency response, and according to a spokesperson for the lab, the first results from this work may be available in the autumn.
“That’s one of the positive outcomes for the state,” says Michael Anderson, California’s state climatologist at the Department of Water Resources. “We’ll have a tool that lets us anticipate where flooding concerns may be, and start figuring out what to do about it.”
Until those maps are complete, however, emergency planners lack a full picture of exactly which neighbourhoods, roads and facilities would be underwater in a true megaflood scenario.
In the meantime, officials in some areas have begun to prepare through large-scale emergency exercises – similar to war games but in this case, the enemy is water.
One such exercise took place back in 2013 and 2014 in response to the ARkStorm 1.0 simulations. Researchers at the Desert Research Institute (DRI) downscaled the scenario to the Lake Tahoe region, bringing together more than 300 people to take part in discussions and tabletop drills, testing how they would respond in emergency situations to threats such as landslides, road closures and prolonged power outages.
“Just getting people in the same room, talking to each other was important,” says Dr Christine Albano, an associate research professor at the DRI. “There are an infinite number of ways an ARkStorm could play out, but these exercises help people know who to call and give people a tangible sense of what could happen.”
According to Albano, that first exercise has already yielded benefits when flooding occurred in 2017. She says it improved preparedness and responses, including helping agencies pre-position equipment, staff, fuel and supplies to ensure they could continue operating during power outages.
The DRI is now using the ARkStorm 2.0 scenario to run planning exercises with emergency services and utilities, and has developed an online tool that allows local officials to explore how prolonged flooding could affect their communities.
But Albano says they currently don’t have funding to run more tabletop exercises for ARkStorm 2.0, despite the scale of the risk.
Read more:
- A 100 ft 'mega tsunami’ could hit the US at any moment. And that’s only the beginning
- A new mega-earthquake hotspot could be forming beneath the Atlantic
- Why sinking cities may now be a bigger climate crisis than rising seas
Exercises like those run by the DRI are important in revealing weak points in the system – but they can’t always fix them. In California, that responsibility falls largely to the Central Valley Flood Protection Plan (CVFPP), a long-running state initiative aimed at upgrading levees, restoring floodplains and reducing the risk of catastrophic river flooding.
Following the devastation in New Orleans from Hurricane Katrina in 2005, California launched an investigation into its own flood management system to find out where its own weak points might be.
“That assessment made us realise, ‘Wow, we really let the system slide for a bit,’” Anderson says. “It needs lots of improvements.”
In 2012, the state formally adopted the CVFPP, with projects ranging from reinforcing embankments to setting levees back from riverbanks, all designed to allow floodwaters more room to spread safely.
The plan was updated in 2022 and is expected to cost up to $30bn over the next 30 years, which may sound like a lot but is a drop in the ocean compared to the damages that would be inflicted by an ARkStorm-type event.
Still, progress is slow. Even when funding is approved, projects can take a decade or more from planning to completion – a sobering timeline when the next megastorm could, in theory, arrive any winter.
“A lot of good progress has been made,” Anderson says. “What we’re trying to do is raise the bar so the system performs to a higher standard even if you can’t completely protect it against the worst-case scenario.”
California has already had a taste of how cascading infrastructure failure can unfold. In 2017, damage to spillways designed to release surplus water at Oroville Dam – the state’s second largest reservoir – forced the evacuation of nearly 200,000 people downstream, amid fears that uncontrolled erosion could lead to catastrophic flooding.
“That specific issue has been fixed,” Swain says. “But the question is, where are the other weaknesses? We patched that one, but there are other auxiliary structures at dams that are seriously at risk in an event like this.”
In an ARkStorm-scale event, Swain adds, it may not be dams that fail first, but the long, ageing levee networks that protect cities, transport routes and industrial zones across the Central Valley.
Ultimately, to avoid the worst possible disaster when the big one hits, parkland and farmland may have to be sacrificed and flooded to protect urban areas.
“Systems to do that aren’t really in place yet, but they could be in a decade or so if we decide to prioritise them,” Swain says. “Given that a flood of great magnitude is in some form inevitable, we do still have some control over how bad the effects are and also which places are actually going to see the worst flooding.”
Are we ready?
Taken together, California’s preparations for ARkStorm-scale flooding form a patchwork of improved forecasting, emergency training and gradual infrastructure upgrades. Each helps. None is trivial to implement.
But many scientists and planners argue that the pace of adaptation remains dangerously out of sync with the pace of risk.
“There’s a lot more that needs to be done. We are not prepared,” Albano says. “Our reservoir systems and flood infrastructure were not built for storms that are warmer, longer-lasting and dominated by rain rather than snow.”
Proper protection also depends on having detailed inundation models to show exactly how floodwaters would spread – work that is now underway, but still incomplete.
And even when the science is clear, Swain says, turning it into concrete action can be sluggish. A lack of sustained political buy-in and long-term funding has made progress “painstakingly slow,” he says.
By the US Geological Survey’s own admission, California’s flood protection system is not designed for an ARkStorm-scale event – and it may never be able to eliminate all the risks such a storm would bring.
Yet when it comes to flood disasters, authorities do have one major advantage: foresight. Unlike earthquakes and even to some extent wildfires, extreme storms can often be seen forming days or even weeks in advance, allowing time to prepare.
Each year, millions of people across the state participate in the Great ShakeOut exercise, designed to help prepare in the event of a major earthquake.
Similar catastrophe planning approaches could be used to prepare for flood events, if there is the political will and public awareness to get behind it.
Ultimately, no one knows when an ARkStorm will strike. But the science is clear that it will. “Statistically, it could happen this year, or it might not happen for 70 years, if we get lucky,” Swain says.
In that sense, ARkStorm 2.0 is less a distant threat than a glimpse of a very real future – one in which California must confront not only scarcity of water, but too much of it, arriving all at once. Whether the state can strengthen its defences before that future arrives remains an open question.
Read more:
