You’re out of free articles.
Log in
To continue reading, log in to your account.
Create a Free Account
To unlock more free articles, please create a free account.
Sign In or Create an Account.
By continuing, you agree to the Terms of Service and acknowledge our Privacy Policy
Welcome to Heatmap
Thank you for registering with Heatmap. Climate change is one of the greatest challenges of our lives, a force reshaping our economy, our politics, and our culture. We hope to be your trusted, friendly, and insightful guide to that transformation. Please enjoy your free articles. You can check your profile here .
subscribe to get Unlimited access
Offer for a Heatmap News Unlimited Access subscription; please note that your subscription will renew automatically unless you cancel prior to renewal. Cancellation takes effect at the end of your current billing period. We will let you know in advance of any price changes. Taxes may apply. Offer terms are subject to change.
Subscribe to get unlimited Access
Hey, you are out of free articles but you are only a few clicks away from full access. Subscribe below and take advantage of our introductory offer.
subscribe to get Unlimited access
Offer for a Heatmap News Unlimited Access subscription; please note that your subscription will renew automatically unless you cancel prior to renewal. Cancellation takes effect at the end of your current billing period. We will let you know in advance of any price changes. Taxes may apply. Offer terms are subject to change.
Create Your Account
Please Enter Your Password
Forgot your password?
Please enter the email address you use for your account so we can send you a link to reset your password:

Canadian wildfire smoke is returning to the United States this week, triggering air quality alerts around the country. But when I open up my weather app and check the weather conditions in some of the cities that the smoke will hit — New York, Pittsburgh, Chicago, Nashville — the word “smoke” doesn’t appear anywhere. Even the air quality alerts don’t mention it.
Smoke exists in a weird place, weather-wise. Our vocabulary for it is entirely divorced from the usual ways we talk about the outside world; our partly cloudies and rains and snows exist alongside temperatures and wind speeds and dew points that, put together, arm us with a crisp picture of the weather before we ever step outside. Describing smoke, on the other hand, sort of depends upon whom you ask.
The National Weather Service recognizes smoke as a type of weather event, but the agency rarely talks about it that way. The NWS’s observations of those smoky June days in Chicago only mention “haze,” a catch-all term that is generally used in the context of transportation (if it’s hazy, visibility is low). Apple Weather and Accuweather don’t have icons for smoke, but Weather Underground does. For the most part, smoke makes itself known through exactly one metric: the Air Quality Index, which was first created to measure something else entirely and is so separated from the weather that it doesn’t even appear on the NWS’s forecasts.
Experts told me this is by design. Air quality and weather exist on separate, if parallel, tracks: Weather data from the NWS turns into the forecasts we see from TV meteorologists and in the weather apps on our phones. The air quality forecasts turn into the number we see in the EPA’s AirNow app.
“Air quality forecasting is a little bit different from weather forecasting,” said Amy Huff, an atmospheric chemist at the National Oceanic and Atmospheric Administration who used to be an air quality forecaster herself. While the NWS issues weather forecasts, Huff told me that air quality forecasts aren’t conducted at the national level, but by state, local, and tribal environmental agencies. Each of those agencies has different pollutants they’re looking out for, and different thresholds at which they’ll send out air quality alerts. “The process is going to vary, because not everyone is requesting the same thing.”
For the most part, this has worked fine. Local sources of pollution historically have the most impact on air quality, and local environmental agencies know which pollutants are the most relevant to their area. California’s environmental protection agency, for example, forecasts for a wide range of pollutants due to a history of serious air quality issues. Maryland, on the other hand, just forecasts for ozone and PM 2.5 particles (which are present in wildfire smoke).
Get one great climate story in your inbox every day:
Where air quality and weather overlap is through the alerts system. If the local agencies think pollution is going to be bad enough to cause harm, they tell the NWS, which will send out the alert through its system — which is how we see those alerts in our phone’s weather apps.
“The people who pay attention to the air quality on a day-to-day basis historically are the sensitive groups. So if you're a parent of an asthmatic child or you're a senior citizen, you have COPD, you are much more susceptible to the impacts of air quality,” Huff said. “But for the general public, air quality is not really something that historically people are aware of, until there's an event like this.”
For a long time, this made a lot of sense. When air quality monitoring and standards were first set up across the country, regulators were reacting to industrial air pollution that has since reduced dramatically. The fact that air quality is usually good enough in most parts of America to go unremarked-on (outside of wildfire-prone states in the west, at least) means that the regulations worked. This is also why people who live on the West Coast are more familiar with the risks of wildfire smoke: It’s a common enough regional phenomenon that locals know how to talk about and deal with it.
What we’re seeing with the Canadian wildfires is more complicated. Measuring and forecasting pollution from localized sources, including wildfires, is relatively easy. But the Canadian wildfire smoke is getting caught up in the same low pressure systems that usually bring rain around to the Midwest and East Coast — in essence creating a smoke storm. Understanding what’s happening inside those storms is difficult.
“Satellites can detect fires, or we get human reports, so we know where they are location-wise and how much smoke is coming out,” said Shobha Kondragunta, who leads the Aerosols and Atmospheric Composition team at NOAA. “But these fires are injecting smoke into the atmosphere, and these satellites don’t provide the vertical structure of the smoke plume.”
In other words: Smoke is easy to see from above, but it’s hard to tell just how high or low in the air the smoke is sitting. Forecasters can use models to try and predict the smoke’s verticality, but they’re not always accurate, in part because fires themselves are so unpredictable.
“You already have the complexity of predicting the weather, but then you have to add on top of that the difficulty of predicting how a fire is going to behave,” Huff told me. “There’s a lot of things that depends on, like the type of fuel that's burning and what the atmosphere is like around the fire. So it gets complicated quickly trying to predict all these things.”
What they can tell with a fairly good amount of certainty is where that smoke will go as it drifts into weather systems that usually pick up more benign passengers, like the water that eventually turns into rainstorms. So we know when smoke is coming, but we don’t know whether it will be low enough to trigger an air quality alert. It’s like seeing the approach of storm clouds without knowing how much rain will fall.
But conditions can change quickly, and the air quality forecasting system isn’t set up to respond as quickly as weather forecasts are. Many environmental agencies don’t have full-time air quality forecasters, so forecasts can sometimes be delayed simply because there’s nobody around to issue a forecast. Air quality alerts can also trigger automatic operational changes like changes to public transit service and optional telework, and those changes take time to implement. To give agencies and companies time to respond, Kondragunta and Huff told me, some regions mandate that air quality forecasts can’t be updated for 24 or even 48 hours after they’re issued.
When smoke does turn hazardous, it’s usually up to the media to communicate the problem — a system that Huff thinks worked quite well during the June smoke events. “It seemed like people were getting the message and were changing their behavior to protect themselves,” she said. And as more Canadian wildfire smoke has made its way to the United States this summer, local environmental agencies seem to be issuing alerts earlier, as we’ve seen this week.
I can’t help wondering, though, if it’s time to make room for a bit of uncertainty in how we talk about smoke, and to let the word replace “haze” when we know it’s coming — even if we don’t quite know how it will affect air quality. Environmental agencies would still be able to take their time forecasting air quality, but at least people would know that smoke was coming even if an alert is delayed. That would allow them to take precautions like packing a mask just in case the air quality does turn bad, just as they might take an umbrella with them if the forecast called for rain.
As my colleague Robinson Meyer has written, the Canadian wildfire smoke could keep coming until October. And while the American wildfire season has been relatively quiet so far, a rash of Southwest heat waves means it could soon pick up. The future is hazy; our weather forecasts don’t have to be.
Read more about the wildfire smoke:
Log in
To continue reading, log in to your account.
Create a Free Account
To unlock more free articles, please create a free account.
The data center boom is everywhere you look in U.S. economic and emissions data.
This is an edition of Heatmap Daily, an evening review of the day’s news written by our executive editor. Sign up for it here.
It isn’t exactly a new thought, but I’ve been struck recently by how many trends in America’s economic and environmental data are fundamentally about the data center boom and the return of electricity demand:
First, the Energy Information Administration reported this week that U.S. emissions grew by more than 2% last year, driven by surging electricity demand and an increase in coal-fired generation. What caused that higher power demand? New factories and data centers — as well as record summertime cooling demand.
Second, many of the new factories driving that higher power demand are themselves producing goods that are … let’s say … data center-adjacent. There are the enormous new semiconductor fabs, of course. But Ford and General Motors have also set up new production lines (or repurposed old ones) to manufacture grid-scale batteries to meet power demand.
Third, take a look at the recent U.S. spending on private non-residential construction — in other words, everything American companies are building that is not houses, condos, or apartments.
The construction industry’s spent almost $60 billion on data centers over the past year, which is more than it spent on all other office buildings combined (and more than it spent building warehouses, too). Just a handful of categories — data centers, power plants, electricity infrastructure, and certain kinds of electronics manufacturing — now make up a third of all U.S. private non-residential construction investment. They’ve never made up such a large share of construction spending since data collection began in 2014.
As The New York Times recently noted, the American economy is unusually dependent on the American stock market right now — and the stock market is unusually dependent on artificial intelligence. This week, investors started to balk at the enormous spending hyperscalers are planning to keep building out the AI boom; Alphabet’s shares dropped 8% this week after it boosted its planned 2026 capital expenditure and signaled 2027 will be even bigger. If the data center boom started to slow down in earnest, then more than just that budget will change.
Speaking of which, my colleague Emily Pontecorvo wrote earlier this week about how many businesses are struggling to even estimate their carbon emissions from artificial intelligence. The carbon accounting startup Watershed recently unveiled a new formula to help companies get a sense of their AI-related emissions.
But even that formula is still limited by the amount of data hyperscalers publish — and they don’t publish that much. Google, for instance, is the only AI company that has (laudably) provided estimates of its emissions on a per-prompt basis. Yet no company has published its per-token emissions, or how emissions sync up with particular models or regions.
So Emily asked Google: Why aren’t you — or any other model provider — disclosing this kind of data yet?
The tech company didn’t get back to us until after we’d published Emily’s story. But its response was interesting enough that I wanted to quote some of it here.
The problem is “industry consensus,” Cooper Elsworth, a Google spokesperson, told us. “There is currently very little consensus on how to comprehensively and fairly measure the serving environmental impact of generative AI (such as text generation),” he wrote. “Without standardized, ‘apples-to-apples’ frameworks, it is difficult to compare different providers accurately.”
That’s partly because energy use — and emissions data — can vary from site to site and depend on “custom-built hardware, software compilers, and advanced inference techniques.” And he claimed Google doesn’t always have the measurement hardware in place to provide such specific estimates: “Providing precise, repeatable data requires highly advanced measurement infrastructure,” he said. “For example, software-based energy monitoring tools often suffer from sampling biases. For our study, we had to step away from top-down averages and directly measure actual energy at the physical power supply unit (PSU) level across our deployed fleet. Not all providers have the telemetry or data sets required to benchmark their operations at this level of granularity.”
Read Emily’s story to understand the other reasons why estimating — or even “guesstimating” — AI-related carbon emissions is so challenging.
A conversation with Emma Uridge of the Kansas Health Institute.
This week’s conversation is with Emma Uridge, analyst with the Kansas Health Institute. Uridge spent copious hours analyzing state and local laws on data center development to best understand how policymakers are responding to the potential environmental public health impacts of large AI infrastructure, including power and water. The report, which came out this week, also goes in depth into those health impacts. I reached out to her to discuss what she sees as must-watch territory for our readers on this emerging policy arena.
Our conversation was lightly edited for clarity.
What is actually being done on policy when it comes to data centers — beyond moratoria of course?
So first I’d like to just talk about the point of moratoria. It’s helpful to talk about how these policies emerge in the first place. One area where moratoria are helpful is when a data center is proposed but the county has no approach for how they’d like to potentially regulate them. That’s temporary, most of the time. It lets local governments conduct research on the various impacts and also negotiate community benefits, ones that can mitigate any potential negative impacts — like Lancaster Pennsylvania, which instituted a community benefit agreement that maximized the potential benefits of development while mitigating what large data centers can do. That agreement looked at capping municipal water use at 20,000 gallons per day and requiring 100% clean energy. It had financial penalties for non-compliance. The company also committed $20 million to their local economic development and clean energy fund. There are ways to negotiate with developers.
We also see amendments to existing zoning. Data center proposals are increasingly popping up in rural areas, many of which are unzoned, so there’s no way a county can negotiate unless there’s a moratorium in place.
Other policy solutions include different performance standards or requiring on-site renewable energy, like what Jefferson County, Missouri, looked at. Also setback requirements, mandatory noise buffers, ending by-right zoning.
Where are local governments getting ideas for regulating data centers?
A lot of the technical information comes from developers. That can in cases be seen as a biased source of information. I wouldn’t say there’s a dedicated group providing assistance to local governments when a project is proposed — which is a similar story to wind industry development, where we have only a handful of consultants who provide technical advice. It can be really helpful to get a multi-disciplinary approach to hearing information. It can be helpful to have the utility commission, public health folks, those in academia, as well as the developer.
As of right now, especially in rural areas, local governments have a hard task of balancing pushback while getting the most accurate, evidence-based, neutral information to make decisions. That balance can be contentious.
What is the federal government doing on data center policy? How is the Trump administration approaching it?
A few things there. In the early days, the drive was for AI expansion and to be competitive with foreign adversaries. Now due to the amount of public pushback in red and blue localities and a more cautious approach.
I’m not seeing a lot of actual policy movement at this time.
I know the EPA is looking at the chemicals used in cooling data centers because when that water is cycled through the system, some of it is discharged into the water system, so they’re looking at the Toxic Substances and Control Act for monitoring that.
How much of an impact does this minimal federal role have on industry behavior?
Y’know, this isn’t specific to data centers. This is true for all kinds of large-scale development: there’s a need to require some sort of federal monitoring and regulation.
That’s where I see an emerging role for public health. At the federal level, there could be policy movement towards requiring some sort of environmental monitoring at data centers to make sure they’re operating responsibility. Looking at specific water use relative to water availability and what happens when there’s a time of severe, persistent drought. With air quality too — we’ve seen areas where the grid isn’t as reliable so their diesel generators are kicking on more and affecting air quality for residents.
We’re just not seeing all of that right now. We need corporate disclosure.
What do you see as the most important public health impacts from data center development?
It varies by localities. The most discussed obviously is water usage. One thing I’d note about my conversations with folks enthusiastic around emerging tech is, there are still questions that need to be asked about the capacity of localities to support a data center. Like a small town in Kansas may only be using 40% of their water for their utility needs. If a data center came online, how much of that water goes to the data center?
One area underexplored within the public health discipline is energy poverty and energy security. The ability of a household to meet the needs of everything energy provides in our lives. It’s known we have an aging electric grid but we’re not talking enough about large-scale blackouts when the grid is not sufficient to support some of these new data centers.
Plus more of the week’s big development fights.
1. Laramie County, Wyoming — Meta is fighting the fine it received in the Cheyenne data center water pollution controversy, and the conflict between the tech giant and the city’s small board of public utilities is continuing to spill out into the public.
2. Niagara County, New York — This county just rejected a solar project’s highway work permits in a show of retaliation against the state’s Office of Renewable Energy Siting.
3. Barron County, Wisconsin — The anti-solar protest is the new campaign stop in deep red Wisconsin.
4. Chesapeake, Virginia — A large battery storage project on the Virginia coastline is on the rocks amidst rampant local opposition.
5. Lewis County, West Virginia — West Virginia is now a key battleground in the fight over transmission, as a line spanning all of West Virginia and Maryland — and cutting through Data Center Alley in Virginia — causes compounding consternation.