With the planet heating up and nature growing more deadly and destructive, new tools to help people gauge their climate risks are landing faster than DVD-sized hail in Texas. But these models still need fine-tuning. For now, a simpler approach might be to just throw a dart at any map of the U.S. Wherever it lands, there’s your climate risk.

Nearly the entire U.S. population — 99% of the country — has been subject to at least one National Weather Service extreme-weather alert since May, according to the Union of Concerned Scientists, a nonprofit advocacy group. And it’s only August! We’re still just halfway through the “danger season” of May to October, as scientists described it to the Los Angeles Times.

Many of these people have suffered through multiple disasters at once, an unwelcome phenomenon known as “compound events.” People in Houston, for example, last month had to deal with the aftermath of Hurricane Beryl, including widespread power failures, in the middle of a crushing heat wave.

As of this writing, more than 58 million Americans are under alerts for extreme weather, including California wildfires, brutal heat in Texas and the Southwest and flooding from Alaska to the Northeast. Not all of these were necessarily caused by climate change. But an atmosphere warmed by greenhouse gasses makes such phenomena more frequent and powerful. For example, at least 60% of the heat alerts people have received this season were clearly made likelier by climate change, according to the UCS, citing the research group Climate Central.

Warmer air even makes it rain more by soaking up moisture like a sponge and dumping it in buckets. For every degree Celsius the atmosphere warms, it can hold 7% more water vapor. Deadly flash floods in the Northeast over the weekend, when 3 or more inches of rain fell per hour in some spots, weren’t delivered by Hurricane Ernesto, which at that point was far out to sea. Instead, a meteorological “traffic jam” slowed down thunderstorms hauling warm water up from the Gulf of Mexico, according to commercial forecaster AccuWeather Inc.

Scientists called the floods “thousand-year” events, meaning the chances of them happening in any given year were one in 1,000. Last weekend, we got two of them in one night.

The combination of increasingly chaotic weather and population growth in some of the riskiest places in the country means that the number of expensive weather disasters has ballooned in recent years. So far in 2024, there have been 19 separate events that have inflicted at least $1 billion in damage, according to the National Oceanic and Atmospheric Administration. Adjusted for inflation, this is already the fourth-busiest year on record. And as attentive readers will remember, it’s only August!

These things add up. Billion-dollar disasters alone have done $50 billion in damage so far this year, according to the NOAA. That tally doesn’t include recent or sub-billion-dollar catastrophes such as those Northeastern floods or this spring’s Texas wildfires, the worst in the state’s history. Globally, natural disasters caused $120 billion in damage in just the first six months of 2024, only about half of which was insured, according to Swiss reinsurer Munich Re. The bulk of that pain was felt in the U.S.

Little wonder investors have been pouring money into climate-risk models, as Bloomberg Green detailed recently. Insurers, politicians and homebuyers understandably want granular predictions about where the next disasters will strike in the coming decades. There’s even a blossoming industry in betting on these risks that has echoes of the financial crisis — including The Big Short co-star David Burt sifting through model data to estimate potential mortgage losses and Nouriel “Dr. Doom” Roubini building an exchange-traded fund that will include supposedly climate-friendly REITs. Roubini’s team will use artificial intelligence to comb ZIP-code-level model data to score REITs on climate resiliency, according to the Financial Times.

The catch here is that the models might occasionally have some trouble with, how do you say, modeling. A recent study from researchers at the University of California at Irvine pitted their model against that of First Street Technology Inc., the best-known provider (and one I and the U.S. government consult often) to gauge climate risks in parts of Los Angeles. The two agreed only 24% of the time. In comments to Bloomberg Green, First Street backed its own product as having superior results. But a third model from CoreLogic Inc. agreed with Irvine or First Street results only about half the time. And the number of these models just keeps growing.

It might be reassuring if we could get them to give more, or at least find the One Model to rule them all. But the fact that heat waves or routine thunderstorms increasingly have the capacity to turn biblical on any given Sunday, and in any part of the country, means that, practically speaking, models will only ever help us so much. Insurers and investors could lose mere profits if they don’t take a more holistic approach to risk assessment. Policymakers and homeowners have much more at stake, including lives, homes and livelihoods. They need to be ready to handle the gamut of weather disasters, no matter what a risk score tells them.

Models and the cool maps they produce may offer the appearance of granular precision, but any modeler will tell you their results can’t promise that level of foresight. There are margins of error. Don’t let your house fall into them.

Mark Gongloff is an editor and columnist covering climate change.