AI and the weather forecast

Summary

Podcast Summary: "AI and the Weather Forecast"

Click Here/1A Cyber Monday Special – December 30, 2025
Host: Dina Temple-Raston (Click Here) with Jen White (1A)
Guests: John Morales (veteran meteorologist), Amy McGovern (AI Institute, University of Oklahoma), Paris Perdicaris (University of Pennsylvania/Microsoft), Dharna Noor (The Guardian), Jeff Masters (Yale Climate Connections, former NOAA)

---

Overview

This special episode, a collaboration between Click Here and NPR’s 1A (as part of their "Cyber Monday” series), explores how artificial intelligence (AI) is dramatically altering weather forecasting. It dives into the promising advances made possible by AI, the vital role of government infrastructure like NOAA, and how proposed policy changes—particularly widespread funding and staffing cuts during the second Trump administration—are threatening the entire system. Through expert interviews and compelling personal stories, the episode paints a picture of a field at a crossroads, where technological promise collides with political and economic realities.

---

Key Discussion Points and Insights

The Promise of AI in Weather Forecasting

• Breakthroughs in Speed and Accuracy:
  • AI allows for rapid data processing—forecasts that previously took hours now take seconds.
  • AI models, using historical data, can spot patterns and make predictions with increasing precision.
  • “A four day forecast today is as accurate as a one day forecast was three decades ago.” — Jen White [02:16]
• How AI Works in Forecasting:
  • Traditional models rely on mathematical and physics equations; AI models learn from massive historical datasets, akin to how an experienced sailor predicts storms from subtle cues.
  • “AI models excel by learning patterns in actual data rather than trying to solve complicated systems of mathematical or physics equations...” — Paris Perdicaris [04:01]

Notable Quote

"AI can spot patterns that humans can't see and turn a day's worth of data into a forecast in seconds."
— Jen White [01:19]

The Critical Foundation: NOAA and Global Cooperation

• NOAA (National Oceanic and Atmospheric Administration) acts as the backbone of both American and global weather forecasting through its network of satellites, weather balloons, and data-sharing agreements.
  • Twice daily balloon launches across the world collect vital real-time atmospheric data.
  • International cooperation is described as “an act of global unity” in meteorology.
• "There's something kind of romantic about it, isn't it? That all these people come out at the same time all over the world and release these Mini Cooper–sized balloons out into the atmosphere."
  — Dina Temple-Raston [19:22]

Notable Quote

"In meteorology, there are no political boundaries... everybody is releasing their own set of weather bodies."
— John Morales [19:10]

The Limits of AI: Climate Change and ‘Black Swan’ Events

• Both AI and traditional models rely on historical data, but “the past is not predicting the future in the way it used to,” especially amidst climate change.
  • AI struggles with “black swan” events—rare or unprecedented storms exacerbated by climate instability.
  • Example: Rapidly intensifying hurricanes and "once-in-a-century" floods now happen every few years.
• “AI is terrible at black swan events.” — Dina Temple-Raston [03:24]
• "How do you predict the extremes if the extreme has never happened before for anything, for statistics in general, how do you predict something that's completely out of your distribution?"
  — Amy McGovern [22:04]

Political and Economic Threats: Federal Cuts, Project 2025, and Fallout

• The second Trump administration, following the Heritage Foundation’s Project 2025, is dramatically cutting NOAA’s budget (proposed 26-27% cut by 2026) and staffing (14-17% staff lost in the past year).
  • Many weather service offices are now unable to staff overnight shifts; balloon launches and crucial data collection are diminished by about 17%.
  • Cuts have meant forecast “degradation,” staff burnout, and, in some communities, less accurate or late severe weather warnings.
• Private sector companies lack NOAA’s infrastructure (satellites, balloon network, trained staff) and rely on its data for their own forecasts.
  • “Private companies just can’t step in where NOAA used to be.” — Dina Temple-Raston [07:27]

Notable Moment

Veteran TV meteorologist John Morales, usually calm and trusted, publicly warned viewers:

“I am here to tell you that I am not sure I can do that this year.”
[27:55]
(about making reliable hurricane predictions after the cuts)

Real-World Impacts: From Local to Global

• Communities at Risk:
  • Delays and inaccuracies in forecasts—as in recent Alaskan storms—can have devastating real-world effects, from flooded villages to aviation hazards.
  • Aviation safety has been particularly compromised, with FAA weather centers dramatically understaffed; this creates risks reminiscent of past fatal air crashes due to poor weather data.

Notable Quote

“We are going to be tracking hurricanes like it’s 1999. We are on a time machine to the previous century. Except this isn’t a party and people could die.”
— John Morales [29:21]

• Economic Stakes:
  • Weather forecasts save billions: improvements in hurricane tracking alone have saved $5 billion per major hurricane, far exceeding NOAA’s annual budget.
  • Further cuts are deemed “the exact wrong time” by forecasters, as climate change makes weather more volatile.

Broken Public Infrastructure vs. Private Innovation

• Privatization advocates believe tech companies can fill the gap; experts agree private AI models can improve processing, but not replace data collection.
• “They [private companies] might be able to build more efficient models, but they need the underlying information from NOAA in order to do that.” — Dina Temple-Raston [10:00]
• Citizen scientists and grassroots efforts have stepped in, but “it isn’t nearly enough to replace what’s been lost.” [25:23]

Why Human Meteorologists Still Matter

• Human interpretation is vital to spot model weaknesses and understand local factors; AI lacks this “judgment” and context.
  • Example: Algorithmic forecasts in New Hampshire disregard extreme wind data, endangering outdoor enthusiasts.
  • “Humans can understand what the limitations of models are. Certain models have a particular sort of flaw where you see it over and over. And the human can say, well, you know, there's these two scenarios... I don't believe this model because in the past we've seen it has trouble in this sort of situation.” — Jeff Masters [41:07]

---

Memorable Quotes and Timestamps

• Data & AI in Weather:

  • “A typical weather prediction model can give you a 24-hour forecast, but it takes about an hour to give you that new forecast. The AI model can give you two in a second or two.”
    — Amy McGovern [20:33]

• On the fragility of the system:

  • “You're attacking the basic infrastructure to do science, something that's not easily recovered from.”
    — Jeff Masters [33:55]

• On NOAA’s irreplaceability:

  • “The handoff to private companies, it can't really happen without NOAA's infrastructure. All those tech companies building AI systems for weather, their models don't work without good data.”
    — Dina Temple-Raston [25:23]

• On what’s at stake:

  • “The forecast doesn’t have to be just about storms. It can be about hope.”
    — Dina Temple-Raston [30:45]

---

Timestamps for Key Segments

• [00:02] – Episode setup; importance of AI in weather and the backdrop of federal cuts
• [01:19] – What’s changed with AI forecasts (Jen White & Dena Temple-Raston)
• [04:01] – Interview: Paris Perdicaris explains how AI forecasts work
• [05:44] – Dharna Noor on impacts of NOAA disinvestment for daily life and industry
• [07:27] – Dena Temple-Raston on what happens if NOAA is dismantled
• [09:46] – Can private firms replace NOAA? (Short answer: no)
• [13:48] – Produced Click Here report (John Morales’ story, historic overview)
• [17:37] – Global weather balloon launches, international cooperation
• [20:33] – Amy McGovern on AI’s capabilities and limits
• [23:13] – John Morales discovers Project 2025’s plan to "dismantle" NOAA
• [27:55] – John Morales’ viral, emotional broadcast warning to the public
• [29:21] – “Tracking hurricanes like it’s 1999” but with greater stakes
• [33:55] – Jeff Masters: What makes this wave of cuts uniquely damaging
• [35:27] – Real impacts: Alaska forecast bust, balloon data loss
• [38:58] – AI’s strengths and limits in weather modeling (Jeff Masters)
• [41:07] – Why human expertise is still crucial
• [43:00] – Staffing cuts & risks to aviation weather safety
• [44:38] – Return on investment for weather infrastructure; urgent need to fund science

---

Conclusion

The episode concludes by sounding an alarm: at the very moment AI breakthroughs could revolutionize forecasting, the essential backbone of data—NOAA—is under serious threat. The convergence of unstable public infrastructure, unchecked technological optimism, and escalating climate extremes creates risks for every aspect of modern life. Yet, the narrative also holds out hope: with smart investment, human expertise, and resilient infrastructure, the transformative potential of AI can indeed be realized. But only if society chooses to support—and fund—the weather systems on which we all rely.

---

Useful for listeners unfamiliar with the episode, this summary captures the main themes, expertise of guests, and the current precarious state of weather forecasting in the United States.

Transcript

[A]

From recorded future news and prx, this is click here. Hey, it's Dena. The Click Here team is taking a short break to get ahead on reporting for next year. And when we come back in January, we've got a surprise waiting for you. Click Here will soon be heard on public radio stations nationwide. We more on that soon. For now, we want to revisit a series we do with 1A, the daily news show from NPR and WAMU. Twice a month, I get together with 1A host Jen White for something we call Cyber Monday. We talk tech, play a story from Click Here and open the phones to hear what listeners think. This time we talked about how artificial intelligence is changing the way we predict the weather. It's spotting storms sooner, issuing warnings faster, and giving forecasters new tools to save lives. But that same technology is colliding with something bigger, politics, budgets, and the fragile systems that make forecasting possible in the first place. Here's host Jen White.

[B]

This is 1A. I'm Jen White. Forecasting the weather has always been a complex blend of art and science. Now artificial intelligence is joining the mix. It can spot patterns that humans can't see and turn a day's worth of data into a forecast in seconds. But AI can only learn from what has already happened. As climate change fueled by greenhouse gas emissions leads to more severe and frequent extreme weather events, AI predictions are still limited. At the same time, the National Weather Service is being hit with severe funding and staffing cuts under the second Trump administration. Those cuts make it harder for scientists to make accurate forecasts and issue timely alerts during storms. So what is, what does it all mean for you from the morning commute to hurricane season? Joining us now for the latest installment of our Cyber Monday series is Dena Temple Rastin. She's the host of the Click Here podcast from Recorded Future News and prx. Dina, always great to talk to you.

[C]

Thanks so much.

[B]

So a four day forecast today is as accurate as a one day forecast was three decades ago. If Vantis and AR are supposed to make longer forecasts even more accurate, what's actually happening behind the scenes?

[C]

There's a lot of math happening behind the scenes and a lot of data. I mean, one of the things that I learned that I didn't know before we reported this out is that twice a day all around the world, these meteorologists go out and simultaneously, an hour before noon and an hour before midnight, Greenwich Mean Time, they release these weather balloons out into the atmosphere. And everyone shares that data. And between that and Doppler radar And computers crunching all the numbers, you're getting a much better flavor forecast. And those balloons are getting real time data that you can't get anywhere else. They're getting air pressure real time, humidity real time. And that's the reason why all of this is so much more accurate than it was, you know, when I was a kid.

[B]

What are some challenges in using AI to predict the weather, especially as we think about extreme weather events that are happening more frequently and intensely than they used to?

[C]

Yeah, we think of AI as being really good at crunching numbers and to find patterns really quickly, but they're finding patterns from the past, and the past is not predicting the future in the way it used to. That's what the really big problem is here, that these events that we're having now, you know, with these floods in Texas, et cetera, et cetera, are black swan events. And AI is terrible at black swan events.

[B]

Well, recently for Click here, you spoke with Paris Perdicaris. He's a mechanical engineering professor at the University of Pennsylvania, and Microsoft tapped him to build a better forecast using AI. Here's how it works.

[D]

AI models excel by learning patterns in actual data rather than trying to solve complicated systems of mathematical or physics equations like the traditional tools are doing. So similar to how an experienced sailor can look at the clouds and feel the wind and then is able to predict a storm, rather than basically learning from experience, not by solving equations like how traditional prediction tools work.

[B]

The AI system he built is called Aurora, and it learns from weather data collected from NOAA and the National Weather Service.

[D]

So we have a very large database of information, maybe Starting from the 60s or 70s up until now, with almost hourly resolution of how the atmosphere has evolved through those decades. So this is the historical data that AI tools are trying to tap into and and extract patterns and understand to make future forecasts.

[B]

That was Paris Perdicaris. He's a mechanical engineering professor at the University of Pennsylvania. Dana, what role does NOAA play in providing the information forecasters need to make accurate weather predictions?

[C]

It basically is the spine for all of this, the backbone of how to make a forecast. It's not just the balloons that are going up, but it's all from Europe. They're all the meteorologists there. They're all coming together to show weather patterns so that now, in the old days, you know, when you crunched a number and you were a meteorologist, maybe it would take you hours to put that together. Now it just takes seconds. And that's what AI is really helping with.

[B]

Well, in April, we spoke with the Guardian's climate reporter Darna Noor about how proposed budget cuts to NOAA could affect us all.

[E]

It might be harder for farmers to know how productive their crops are going to be. It might be harder to know if you're somebody who flies frequently. It might be harder to know whether or not you're going to be able to have a safe takeoff and landing. Even things like global trade could be affected. It might be more difficult to know, for instance, whether shipments of certain materials are going to be impacted by extreme weather. And crucially, this also impacts some of the sectors that have been broadly supportive of the Trump administration. If you, for instance, are working in offshore oil drilling in Louisiana, you're probably gonna be affected if we have less accurate information about the rising seas and about coastal degradation or, frankly, if you need to know about hurricane patterns. And so I think, you know, really, it's all facets of American life that will be affected here.

[B]

That was Dharna Noor, climate reporter at the Guardian, and you can find that [e-mail] since the start of the year, NOAA's National Weather Service has lost 14% of its staff, or around 600 employees. Many weather service offices around the country are facing severe staffing shortages. Some don't have the staff to operate overnight, leaving communities without timely updates on weather events. That can develop quickly. And as the government shutdown continues, those who are working aren't getting paid. The agency announced earlier this year it plans to rehire hundreds of empty positions, but has made little progress. Now. Data Project 2025. This is a policy blueprint from the conservative think tank the Heritage foundation, explicitly calls for dismantling noaa. So far, the Trump administration has implemented, or is in the process of implementing, about half of Project 2025's proposals. If NOAA were dismantled, what would that mean for weather forecasting?

[C]

Oh, I think it would be debilitating for weather forecasting. You know, there's sort of this idea in the Trump administration that private sector can do everything better and that all they need to do is take weather and give it to Google or the Aurora program or, you know, whatever it is. But the truth of the matter is these private sector companies need no as information in order to make these forecasts. They don't have the satellites that are out there. They don't have the weather balloon infrastructure. And According to Project 20, NOAA is slated to have a 26% cut in its budget in the 2026 budget. So this would be just devastating. And I Think even the Trump administration is starting to understand this, as are members of Congress, because some of this they've tried to pull back because they realize that private companies just can't step in where Noah used to be.

[B]

We're talking to Dina Temple Rastan. She's the host of the Click Here podcast from Recorded Future News and prx. Dina, how, if at all, can meteorologists develop accurate forecasts without NOAA's publicly available weather data? And not just meteorologists here in the US but globally?

[C]

Well, they can't. And I think that's what started us on this story, is that we were talking to this kind of famous meteorologist in Florida who's been very good at predicting whether hurricanes were gonna take a turn or whether they were gonna actually hit Florida. And he started noticing a degradation in what he was getting in. And he went on the air and actually said, look, I'm gonna be making predictions that are like 1990 type predictions because I'm losing the information I need to do my job well.

[B]

And as someone who's relying on those forecasts, that must be terrifying, especially if a severe weather event is expected.

[C]

And if you're in Florida, imagine how often those severe weather events are expected. And when he's telling them, hey, batten down the hatches, he needs to have that information. And he got quite emotional. This is what turned us onto the story, is that he got quite emotional on the air talking about how he didn't think he could do his job in the way people had come to expect him to do it over the last 30 years. And it's all because of these cuts.

[B]

So you said private companies aren't positioned to take over all the work NOAA does, but do they have a role to play here in helping to meet critical gaps in weather data?

[A]

I'm not sure.

[C]

Data as much as processing, I think that is more of what's going on. So they might be able to build more efficient models, but they need the underlying information from NOAA in order to do that. And they're not going to pay tens of billions of dollars to put satellites up in the air or create the infrastructure of literally thousands of meteorologists around the world releasing balloons twice a day. It's just not going to happen. It's a fundamental misunderstanding of how all this works.

[B]

So we've got of AI technology and the role it can play in weather forecasting. But then we also have, at least for now, disinvestment at the federal level in our weather forecasting capabilities. So in your reporting, what have you learned about the delicate relationship between scientific advancement and federal investment.

[C]

Well, I mean, this has been a problem not just across weather, but all kinds of research. We're seeing it now during the Trump administration. The problem is they're doing this exactly at the time when weather is as unpredictable as it's ever been because of climate change. So if you had patterns that you could expect were coming down the pike, that would be one thing. But you know, you have a flood in Texas, you have, you know, torrential rains you weren't expecting in Louisiana. So this is more or less the exact wrong time to be deciding that you're going to pass this over to the private sector.

[B]

And what are the forecasters you spoke to saying about the funding cuts and layoffs at NOAA and what it means for their work longer term?

[C]

They think it'll make it really hard for them to do their job. And, you know, I always thought until we reported out this story, I had the old fashioned idea that forecasters were wrong half the time. They're actually right 85% of the time, only wrong 15% of the time. Without NOAA, it's going to be hard to come up with those kinds of numbers.

[B]

That's Dena Temple Rastin. She's the host of the Click Here podcast from Recorded Future News and prx. Coming up, we hear from a veteran meteorolog about why he believes the country's forecast system is in danger. And you can add your voice to the conversation. Email [e-mail] Dina, thanks for speaking with us.

[C]

You're welcome.

[B]

We'll be right back.

[F]

Looking for more of the cybersecurity and intelligence coverage you get on Click Here, then check out our sister publication the Record from Recorded Future News. You'll get breaking cyber news from reporters in New York, Washington, London and Kyiv, among others. And you'll see for yourself why it attracts hundreds of thousands of page views every month. Just go to the Record Media.

[A]

You're listening to Click Here.

[C]

I'm Dina Templerest.

[A]

Today on Click Here, we bring you an episode from our friends at 1A, a daily talk show from NPR and WAMU. They had me on to talk about our reporting on AI and weather forecasting. Here's host Jen White.

[B]

We're discussing the future of weather forecasting. That's as advancements in AI collide with sweeping layoffs and funding cuts at noaa. Their data serve as the backbone of the nation's weather service. It's part of our Cyber Monday series and ongoing partnership with the Click podcast from Recorded future news and PRX in the latest episode. Click Here. Host Dina Temple Raston explores how decisions in Washington are affecting weather forecasting around the country.

[A]

Something remarkable is happening in weather. Artificial intelligence is quietly transforming the forecasts. We all rely on spotting storms sooner, issuing warnings faster. It promises to save lives, to blunt economic fallout, maybe even help us adapt to climate change. But all that progress hit a speed bump recently, a big one. The problem isn't the science or the machines. The real storm is elsewhere. And it broke in the last place you'd expect a local weather report.

[G]

As you've grown accustomed to my presentations over my 34 years in South Florida.

[A]

Meteorologist John Morales went on the air this past June with less of a weather report and more of a warning that something fundamental about the skies we all share is changing.

[G]

What we're starting to see is that the quality of the forecasts is becoming degraded.

[A]

John has been the voice Floridians trust when the clouds start to churn for three decades now. But he didn't start tracking hurricanes to save lives. He kind of stumbled into it by way of another passion. Baseball.

[G]

I was a Santurce crabbers fan. The Canrejeros de Santurce. This is the team where Roberto Clemente played and many other greats.

[A]

As a kid in Puerto Rico, he'd sprawl on the floor, pencil in hand, writing out the box scores from the evening paper. And once every last stat was tallied, he'd turn the page from balls and strikes to storms.

[G]

You know, in the 70s mainly, you know, before Internet, you would get these hurricane tracking charts that would either be inserts in the local newspaper or you could go to the hardware store and grab one of them.

[A]

Not the typical hobby for a young kid, but. But John loved it. He'd sketch out storms, building his own weather reports by hand.

[G]

I would take down all kinds of information. What the barometric pressure was, what the trends were, what direction was it moving, what was the expectation?

[A]

And he took that obsession to Cornell, where he studied meteorology and even began entering weather forecasting contests.

[G]

I won a couple of times.

[H]

A.

[G]

Couple of semesters, I actually came out victorious.

[A]

From Cornell, he made his way to the National Weather Service and eventually to Florida. Meteorologist John Morales is joining us now to break it down for us.

[G]

John, hi, Good afternoon to both of you. And indeed, the advisory is in. This is still a formidable hurricane, now ranked as the second strongest.

[C]

So what do you think people understand least about being a meteorologist?

[G]

I think most people have a preconception that we're wrong half the time. And that's really pretty far from the truth. These days, we're right about 85% of the time and wrong 15% of the time. That's a pretty good batting average.

[A]

It took a lot to get to that kind of accuracy. The science of weather forecasting. It started, well, not that differently than what John used to do in his living room. All the math done by hand. Then there were these gradual advances. Computers, satellites, radars, even this delightfully retro thing that surprised me. Something that's critical to weather science, but sounds more like it belongs at a kid's birthday party. Balloons.

[G]

Well, it's not a party balloon. Okay. I would say it's the size of a very small compact, like a Mini Cooper or a Fiat 500. Right.

[A]

If you've never seen one of these weather balloons launched, picture this. An overcast morning in the parking lot of a squat brick weather station. A technician in a crisp white polo, National Weather Service emblem stitched over the pocket steps outside, cradling what looks like a bundle of limp latex. In a few minutes, it will swell to the size of the Mini Cooper John talked about. After that, we launch the balloon. It gets to about 100,000ft, takes 90 minutes to pop. As it ascends, it beams down all kinds of weather data. Temperature, humidity, wind speed. Every single day, scientists all over the world walk out into the elements to release these giant gleaming white orbs into the sky.

[G]

The two times the balloons are released are an hour before, before noon and an hour before midnight.

[A]

Simultaneous launches synchronized across hundreds of stations around the globe, performing not just an act of science, but also something that feels rare right now. An act of global unity.

[G]

In meteorology, there are no political boundaries. So whether it's Russia or Saudi Arabia or South Africa, everybody is releasing their own set of weather bodies.

[C]

There's something kind of romantic about it, isn't it? That all these people come out at the same time all over the world and release these Mini Cooper sized balloons.

[A]

Out into the atmosphere.

[G]

The mere fact that there is this level of cooperation internationally, right? The fact that they know that this is going to lead to better results for everybody. Well, I mean, that's the type of collaboration that we need to tackle many other challenges that we have going on internationally right now.

[A]

All of this technology, the radar, the satellites, those balloons that all got supercharged in 1970, that's when the federal government established the national oceanic and Atmospheric Administration, or noaa. NOAA became a hub for weather research and innovation, not just in The United States. But all around the world, thanks in large part to NOAA, a four day forecast today is as reliable as a one day forecast was 30 years ago. And right now we're in the middle of another huge leap forward in forecasting. Thanks to artificial intelligence.

[I]

A typical weather numerical weather prediction model can give you a 24 hour forecast, but it takes about an hour to give you that new forecast. The AI model can give you two in a second or two.

[A]

This is Amy McCovert. She's a computer science and meteorology professor at the University of Oklahoma. She also leads the National Science Foundation's AI Institute. They focus on making AI weather systems reliable, which has become a big job because lots of companies are trying to build AI weather models these days. Even big tech has jumped in. Companies like Microsoft and Google are already predicting storm tracks four or five days before landfall, which is important because when it comes to weather, speed matters. And while the introduction of AI to all of this seemed very promising, Amy's quick to say it's not perfect. For example, AI is pretty good at spotting big storms, but not as good at hyperlocal forecasts. Also, AI doesn't understand the physics of weather or the calculus that governs storms.

[I]

Things like you can't negatively rain, negative rain.

[A]

It's hard to know how to plan for that kind of forecast, which should.

[I]

Be obvious, like if it's either raining or it's not, it's not sucking rain out of the atmosphere. But an AI model doesn't know that. Right?

[A]

But actually the biggest problem is that AI learns from the past and the past doesn't look much like the present anymore.

[I]

How do you predict the extremes if the extreme has never happened before for anything, for statistics in general, how do you predict something that's completely out of your distribution?

[A]

Rising sea surface temperatures are turning modest disturbances almost overnight into category three, four, or even five hurricanes once in a century. Floods are coming every few years. Tonight, the Big Apple is a big mess.

[C]

And tens of millions of Americans are.

[A]

Experiencing torrential rain and record flooding.

[C]

Heavy downpours are expected to continue.

[A]

But the biggest threat to all of this AI progress, it has nothing to do with physics or even algorithms. It's something that John Morales stumbled on last year as he was preparing to speak at a weather conference.

[G]

There was a summer community meeting coming up for the American Meteorological Society, and they had asked me to be a panelist on it. And part of what we were going to be discussing was potential threats to America's weather enterprise. And in preparation for this panel I learned a lot about Project 2025.

[A]

Project 2025, the blueprint for the Trump administration's second term. As John read through it, he ran into this section that stopped him cold.

[G]

Project 2025 explicitly says that NOAA is to be well dismantled because the excuse was that NOAA is a climate Alarmist Agency.

[A]

Project 2025 calls for the gutting of NOAA, the National Weather Service, the satellites, the research centers, the entire system that weather forecasting relies on. As John sat through the conference, he listened as dozens of panels touted the glories of AI, but said nothing about looming threat to Noah. And he kept thinking to himself, all the promise of AI, none of it can be fulfilled without good data, data that NOAA provides. And then, just months after the conference, John's worst fears came true.

[B]

All right, we're getting some breaking news.

[A]

That we've been following over the last hour with a Trump administration official confirming that about 5% of the national oceanic and Atmospheric Administration are being let go. The proposed cut to NOAA's budget is 27% in 2026, which will compromise the agency's weather stations and research centers and its aging satellites. Nearly a third of NOAA's Hurricane Hunters have been grounded already. Pilots who fly straight into the eye of a storm to measure its intensity, and those car sized weather balloons, they've been reduced by 20%.

[G]

I've seen some of the maps of where these gaps are. So for example, you look at the Northern Plains and the Intermountain west, and there's huge gaps there with no data whatsoever.

[A]

As with many of the Trump administration's cuts, they seem to think that private companies can do it better. Meteorology Professor Amy McGovern isn't convinced which.

[I]

Company is going to put in several hundred million dollars per radar to cover the United States in radar data.

[A]

All of that, all of that data.

[I]

Is all collected inside noaa and all.

[A]

Of it is used. And here's the irony. The handoff to private companies, it can't really happen without NOAA's infrastructure. All those tech companies building AI systems for weather, their models don't work without good data. It's the definition of shooting yourself in the foot and then wondering why you can't run the race. Some lawmakers are starting to push back, drafting legislation to try to restore some of NOAA's critical positions. And outside Washington, grassroots efforts are trying to fill the void. Public agencies are partnering with private companies. Even citizen scientists are chipping in where NOAA's resources have been cut. It's hopeful and inspiring to see people step up to save something critical. But it isn't nearly enough to replace what's been lost. And John Morales watched all this play out with growing frustration and then alarm. And this is not a feeling his audience knows him for.

[G]

I've become a person that's not just a weatherman. I am almost a trusted family member that folks go to for advice when it comes to preparing for hurricanes.

[A]

He's earned their trust through a combination of accuracy and humanity, but most of all, his calm.

[G]

My audiences have always known me as the just the facts, non alarmist meteorologist.

[A]

But as Noah started to crumble, so did John's ability to give the people who trust him the forecast they can rely on. So just the facts. John decided to do the thing he almost never does. He sounded the alarm.

[G]

So June 2nd was a Monday. It's the first weekday of Hurricane season.

[A]

That night he began his forecast with a story from the past, an old broadcast from 2019, when it seemed that Hurricane Dorian was going to hit Florida. John predicted it would turn back out to sea.

[G]

As you've grown accustomed to my presentations over my 34 years in South Florida newscasts confidently I went on TV and I told you it's going to turn. You don't need to worry, it is going to turn.

[A]

And it did. It turned. Then he said something you never want to hear your weatherman say.

[G]

And I am here to tell you that I am not sure I can do that this year.

[F]

You.

[A]

He laid it all out. The budget cuts, the data gaps, the layoffs.

[G]

Did you know that Central and South Florida National Weather Service offices are currently basically 20 to 40% under understaffed from Tampa to Key west, including the Miami office, 20 to 40% understaffed.

[A]

And then looking straight into the camera.

[G]

This is a multi generational impact on science in this country. I just want you to know that what you need to do is call your representatives and make sure that these cuts are stopped.

[A]

The broadcast went viral around the country, not just in Florida. He'd struck a nerve because to turn on your local news or check a weather app and trust what you see to know whether you're safe or need to evacuate or even whether to take an umbrella or just take sunscreen. Everyone cares about that. In a lot of ways, weather is the most universal human experience. We've come a long way from sticking your finger in the air to gauge wind direction or feeling in your bones that a storm is on the way. That progress has saved building cities countless lives. But as John said recently in an Instagram post.

[G]

Tonight and just about every other night of the 2025 Atlantic hurricane season, we are going to be tracking hurricanes like it's 1999. We are on a time machine to the previous century. Except this isn't a party and people could die.

[A]

There are glimmers of hope. NOAA has announced plans to rehire hundreds of meteorologists and technicians. Public agencies and private companies are working to plug the gaps, and even citizen scientists are pitching in, collecting hyperlocal data in places NOAA can't always reach. But until the system is back at full strength, forecasting will remain at risk. Because weather isn't just numbers on a screen. It's a shared story of survival, of trust, of adaptation. And right now, that story is shifting not just in the clouds above us, but in the halls of power. Which means there's a choice to make. We can surrender to fragmentation, or we can build resilience with AI human expertise, volunteers, scientists, lawmakers and communities working together. The forecast doesn't have to be just about storms. It can be about hope.

[B]

That was host of the Click Here podcast Dena Temple Rastin. Coming up, a weather modeling expert explains what a future without NOAA data could look like. That's just ahead. This is one A.

[A]

Support for Click here comes from GiveWell. Let's say you're a detail oriented person. You don't just go to a movie. You read the reviews first. You recon the menu before going to a restaurant. I'm guilty of that. So how do you do your homework when giving to charity? That's where GiveWell comes in. GiveWell is an independent resource doing rigorous and transparent research into charities. They figure out which ones do the most good for every dollar donated, and they only recommend programs with the biggest impact on helping people and saving lives. That's why over 150,000 donors have already trusted them to direct over $2.5 billion to great causes around the world. So check out GiveWell next time you're giving to charity to make a tax deductible donation Today, go to givewell.org and pick podcast and enter Click here at checkout. Make sure they know you heard about GiveWell from click here again. That's givewell.org to donate or find out more. Do you ever feel like you blinked your eyes and then woke up in some kind of sci fi movie? Suddenly it seems like the very existence of AI is changing everything, including our relationships.

[B]

I would like to think that I could not fall in love with an AI companion, but I really think that anybody could.

[A]

I'm Os Velocin.

[E]

And I'm Cara Frank.

[A]

We break down the tech news you.

[B]

Really need to know.

[A]

Listen to tech stuff in the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts. I'm Dena Temple Rastin. This is Click Here. Here's more from Cyber Monday, our twice weekly series with WAMU's 1A news program. 1A featured click. Here's reporting on how AI is used in weather forecasting. It's something that's colliding with the Trump administration's budget cuts. Host Jen White continued that part of the conversation with Jeff Masters. Jeff is a former NOAA hurricane scientist. He's also a meteorologist at Yale Climate Connections and the co founder of Weather Underground, the first open access weather website. Take a listen.

[B]

So the National Weather Service is just one of the agencies under NOAA to be hit hard by widespread layoffs and budget cuts since President Trump took office. Again, how are these cuts different than other cuts that happened during your time at noaa?

[H]

These are not simple cuts where if you get a new amount of money the next year, you can, you know, go back to where you were. We're actually assaulting the basic infrastructure to do science, to do research, to make weather forecasts. We're losing a lot of the most experienced people. The top end, most experienced people were encouraged to retire early with the Dodge buyouts this year, and a lot of people took those. And we're also losing the early scientists. The probationary employees with just one or two years experience were all let go. That's your seed corn for the future of the science. So you're losing people at the top end and at the beginning end, and that's gonna really majorly impact your progress going forward. You're attacking the basic infrastructure to do science, something that's not easily recovered from. And we're gonna see, you know, in many years to come the impacts of these cuts, even if we're fully funded next year.

[B]

We got this email from one of you. Cuts to weather service office in Alaska caused many small Inuit villages to be flooded because not enough weather balloons could be set up for a more accurate reading. You know, we've been the federal disinvestment in NOAA and this critical moment for us in terms of climate change. But Jeff, how are real communities around the country seeing the effects of delayed severe weather warnings or inaccurate forecasts because of funding cuts or staffing shortages?

[H]

Yeah, because of the cuts to the National Weather Service, they're not able to launch as many weather balloons. And depending on the model that we're using to make the forecast, those are the most important inputs that we have to make a good forecast. So we've seen about a 17% loss in our upper air balloon data in the past eight or so months. And that will degrade your forecast. You roll the dice and you're going to load them in favor of worse forecasts, worse outcomes. And we saw a case last week in Alaska where a former typhoon hit the coast bringing a six and a half foot storm surge, hurricane force winds. And that storm was not properly forecast. I mean, it was a major forecast bust. The position of the storm shifted suddenly by 300 miles just a day and a half before it was supposed to hit, resulting in a big surprise for a lot of those communities there on the coast. And that's a very unusual sort of situation. And it certainly didn't help that 5 of our 13 upper air balloon stations in Alaska were not properly taking measurements. So we can't say for sure that was the cause of that bad forecast, because forecast busts are complicated. But you load the dice in favor of worse forecasts and in a critical situation like that, you have to make sure you get things right.

[B]

Not everyone may know about the connection between NOAA and the weather app they use on their phone. So walk us through how the system work, how works, how does the weather data get turned into forecast? We can see on our phones or hear on the radio, or pull up on our computer.

[H]

Every day you take weather observations, like 10 billion different weather observations globally. You've got airport data, you've got radar, you've got buoys over the ocean, you got satellite data, you got data coming in from balloons, from commercial aircraft that send back data. All of this data gets put together by the National Weather Service in the US and by international weather services all over the country to formulate an initial picture of what the atmosphere is doing right now. And in order to make a good forecast, you have to have an accurate picture of what the weather is right now. Once you have that picture, then you can start to generate a forecast. You traditionally have used what we call physics based models, which subdivide the atmosphere into a grid where you've got a data at each point, that grid saying what the temperature, moisture, winds and so on are. And then you can solve the basic mathematical equations of how the atmosphere works and generate a forecast. Or in the new system, we can actually use AI to do that work as well. So the forecast that you get on your App can be either from the National Weather Service or from a commercial entity that runs their own model. And the two have their various strengths. And you pick the one which works better. Some work better in some situations than others, so. But both of these forecasts all start from the same point. You have to have an accurate picture of what the current weather is doing. And if you're making cuts to the observation system like we're doing now, you're going to degrade all forecasts, not just from noaa.

[B]

Well, you mentioned AI, Jeff. How are meteorologists using AI to make weather forecasts more accurate?

[H]

They're kind of neat because you don't have to have a supercomputer. Once you train the model in order to make a forecast, you can run that on a laptop. So a lot of companies can customize that sort of capability to make forecasts that are hyper local or focused on a particular sort of circumstance. And that's the big advantage that they have over the traditional models, that you don't need a supercomputer to run them.

[B]

We got this email from Donna who says, I live in the White Mountains of New Hampshire. There are a number of weather forecast resources, including the Mount Washington Observatory. There is one source that is all algorithms, no media, meteorologists involved. It is notorious for getting wind speed wrong, especially at higher elevations. The reason? The algorithm that power the models on which it was built discount extreme wind values. So it doesn't believe that high elevation wind speeds could possibly be that high. It puts outdoor enthusiasts, hikers and skiers in danger. Jeff, what are some of the limitations of using AI to predict the weather?

[H]

Yeah, your AI model's only as good as the data you train it on. So if you're not training it on, you know, particular observations, for instance, this high wind observation that Mount Washington had, or if you're seeing because of climate change completely unseen before sorts of weather situations, then your data is not going to be very good for your AI trained model. So that's a big concern with climate change. You're going to make weather more difficult to forecast, both because it's not been experienced in the past, and AI models don't do well with that. And also traditional models are tuned to some degree based on what the past weather is. So all models will have trouble with a changing climate.

[B]

Well, this makes me wonder about the human factor in weather forecasting. So you may use AI, you have to do these mathematical equations to figure certain things out. But why is it important to have a human being who's also a part.

[H]

Of the work that's happening, Humans can understand what the limitations of models are. Certain models have a particular sort of flaw where you see it over and over. And the human can say, well, you know, there's these two scenarios and you know, I don't believe this model because in the past we've seen it, has trouble in this sort of situation. We have that going on right now, for instance, in the Caribbean Sea where we've got a tropical disturbance that is expected to take one of two paths. Either it's going to power northward and hit the Dominican Republic, or it's going to continue into the center of the ocean where maybe it will threaten Jamaica or Cuba. And the two models that we have that give those two scenarios, we're tending to weight one over the other. Based on what the past has shown. That particular model has a weakness or a strength in.

[B]

We got this email from Clifford who says, I'm a former employee of Environmental Canada, which includes the Canadian Weather Service. National Weather Services have an international obligation as members of the World Meteorological Organization of the United nations to provide regulated precision, 12 hour upper air and hourly surface aviation observations. The aviation industry is particularly dependent on weather observations and specialized aviation forecasts. And Kevin emails. I'm a general aviation pilot and recently flew into near severe turbulence due to strong wind shear at high altitude. This is an area that must be researched as it seems that higher temperatures and dew points are causing more severe turbulence events to occur affecting all sectors of air travel. Jeff, we talked about weather as it relates to how we're living on the ground, but millions of people across the US travel by plane. How, how is the way climate change is sort of reshaping weather patterns affecting air travel?

[H]

We're experiencing more turbulence because of climate change, and that's a big concern. And also climate change is making extreme events more common and more intense, something our models don't necessarily do very well. So that is another concern. I'm also majorly concerned with how the cuts to the National Weather Service are affecting aviation meteorology. We've got 21 centers that are run by the FAA that have one to four meteorologists on staff. Right now, they're all supposed to have four meteorologists on staff. So we only have 69 of the 90 meteorologists that we're supposed to have at these FAA aviation weather centers, which means we're not getting as good a weather forecast for the aviation industry as we should be. And that will cause delays and it does threaten flight safety. I mean, those meteorologists were put on that job because of a major crash that happened back 49 years ago when a airplane flew through a severe thunderstorm on the way to Atlanta that was not properly forecast because the air traffic controllers couldn't get the weather information to the pilots. So we need good aviation forecasts and we are critically low. I mean only 69 out of 90 meteorologists on staff at these weather observation or these weather forecast centers for aviation.

[B]

Jeff, in a best case scenario, what kind of funding and staffing is needed to keep the data collection and system maintenance for weather forecasting running like it should?

[H]

Going into this year, the National Weather Service was down about 5% in staffing. We need to go back to that level. We need to reverse this 17% cut that they've seen in positions this year. We need to fully fund NOAA and the National Weather Service. We need to fully fund their research efforts as well, because that's what we're experiencing huge savings in. For instance, improvements in hurricane forecasts over the last 20 years have been responsible for a $5 billion per major hurricane savings in the last few years. And the entire NOAA budget is only 6.8 billion a year. So consider the 10 major hurricanes we've had hit the US over the last eight years. You know, you've got tens of billions of dollars in savings from these investments that we've made in weather forecast capability. So we have to realize that the cost to benefit ratio of money we put into NOAA is money well spent. And I've seen numbers as high as, you know, it's a 73 to 1 return on your money spent. So we need to fully fund also the National Science foundation, which is the funder for a lot of the research that happens at the university level. A lot of the training of the grad students which end up going into the National Weather Service or into the weather research organizations like noaa. That comes from our higher education system, which is under assault also from the lack of foreign students we're allowing in. I mean, about 30% of all grad students in my field are foreign students. We need to emphasize that that funding should continue. So a lot of things need to happen to maintain this world class system of weather and research that we developed over the last few decades, which have been the envy of the world.

[A]

That's Jeff Masters. He's a meteorologist at Yale Climate Connections and the co founder of Weather Underground. He spoke with Jen White, the host of WAMU's 1A. And you've been listening to a special edition of Click Here, an edited show we did with WAMU's 1A news magazine. You can hear the full segment with more listener responses and questions [e-mail] click here is a production of recorded future news and pra. I'm Dena Temple Raston and our producers are Megan Dietrich, Sean Powers, Erica Gaeda and Zach Hirsch. Special thanks today to one A producer, Lauren Hamilton. Click Here is edited by Karen Duffin, Fact Checked by Darren Ancrum, and contains original music by Ben Livingston with some other music from Blue Dot Sessions. Our staff writer is Lucas Riley and our illustrator is Megan Gough. Jesse knightswonger and Jake Cook are our sound designers and engineers. I'm Dena Temple Raston and Click Here. We'll be back on Friday with a mic drop. We'll see you then.

[F]

Looking for more of the cybersecurity and intelligence coverage you get on Click Here. Then check out our sister publication the Record from Recorded Future News. You'll get breaking cyber news from reporters in New York, Washington, London and Kyiv, among others. And you'll see for yourself why it attracts hundreds of thousands of page views every month. Just go to the Record Media.