Stand Up a National Climate Service

That we need to rebuild national capacity in the natural sciences is obvious to anyone who has been paying attention to the cascade of alarming headlines over the past year and a half. In that short time Donald Trump has done a tremendous amount of damage: shutting down National Forest Service offices, cutting off funding for climate-related research in both academia and the government, and attempting to sell off assets from the National Center for Atmospheric Research to the highest bidder.¹

But even before the destruction began, our approach to climate science needed a strategic rehaul. While American scientists were among the best resourced and regarded in the field of atmospheric modeling—a field that explores how our Earth will change over time as we continue to pump out greenhouse gases—we were arguably understaffed and uncoordinated in helping people figure out what to do about these changes on the ground.² On questions of climate mitigation, we know as much now as we’ve ever needed to know: We should stop burning fossil fuels, as fast as possible. We don’t need more models to tell us that.

But when it comes to climate adaptation, there is a lot of work to be done. The United States remains one of the very few countries in the world without a national adaptation plan.³ The majority of states similarly lack plans, which would at the very least tell local governments what to expect in the decades ahead.⁴ As climate scientist Adam Sobel has memorably argued, we need more usable climate science that can help particular people in particular places make decisions in the present.⁵

We need a National Climate Service that can coordinate modeled climate data from science agencies like the National Oceanic and Atmospheric Administration and NASA with ground-level engineering and exposure information. Such a service will serve many purposes, but I highlight two main goals here: The first is to provide free public-facing information about climate risk to end users like states, municipalities, and homeowners. The second is to bolster the federal government’s own planning capacity, both to make decisions about physical adaptation and also to aid in crucial oversight of our macroeconomy and financial sector.

Localized Information Is Power

Insurers already see this connection between climate science and our ability to anticipate and respond to local economic impacts. The American Academy of Actuaries—the professional association of insurance actuaries in the United States—submitted a comment letter protesting the Trump administration’s plan to dismantle the National Center for Atmospheric Research (NCAR). “NCAR’s continuity is not only a scientific necessity,” the actuaries wrote, “but an economic one.”⁶ The actuaries explained that NCAR’s supercomputer-generated data was essential for their evolving understanding of climate impacts like hurricanes, wildfires, and changing precipitation patterns. Without NCAR’s publicly available research, they argued, the insurance industry would be left in the dark. This heightened uncertainty could lead them to raise premiums. Or the lack of data could lead them to underestimate total risk, which could result in any number of effects, including inadequate capital reserves, insurer insolvency, or higher volatility in pricing.⁷

So far, the states have taken varying approaches to modeling asset-level climate risks. In 2025, California passed legislation funding the construction of the nation’s first public wildfire catastrophe model, which aims to be an “open-source, community data-driven modeling framework” built with expertise from universities in the state.⁸ While the impetus for the model was the desire to oversee premium pricing from private insurers, the legislation acknowledges that it can be used to inform a range of decisions, including emergency planning, wildfire risk mitigation, and broad oversight of the insurance sector, including solvency.⁹ While California’s innovation in this area should be lauded, not every state can mimic this approach. Catastrophe models require years of work from an interdisciplinary team of PhD-level experts, with costs reaching into the millions of dollars. California has set aside an initial $8.5 million over the next three years.¹⁰ This is for modeling just one hazard: fire. Especially for states with smaller budgets and fewer research universities, federal help is needed to model the many increasing hazards in the years ahead, from water scarcity to flooding to hail.

Connecticut has taken a different approach, simply licensing climate-risk information from a third-party private data provider and giving residents access to climate risk scores.¹¹ As I’ve written elsewhere, and others have echoed, solely relying on for-profit providers has serious drawbacks.¹² The government is endorsing information created by black-box models in which it has no control over data inputs or methodology. Developing these tools in the public sector would not only ensure transparency but would also help train future generations of climate scientists to improve upon and maintain the models. While the private sector should be encouraged to continue to develop and use its own analytics, government-maintained models serve as an important public option, providing a free source of information that can be used to double-check private-sector claims.¹³

Climate Services for Macroeconomics

Updates in granular physical risk modeling are needed not just for local planning but for economic forecasting as well. Last February, Federal Reserve Chair Jerome Powell admitted to the Senate Banking Committee that areas of the United States may become uninsurable: “[In] 10 or 15 years there are going to be regions of the country where you can’t get a mortgage.”¹⁴ But while Powell conceded that this “will have significant economic consequences,” he wasn’t ready to concede that this is “a financial stability issue.” I’m not so sure about that. It certainly would be nice to know how large and where we expect these areas to be. Yet Federal Reserve researchers themselves wrote in 2024 that “surprisingly little is known about U.S. climate-related financial stability risks,” as they were only just beginning to “integrat[e] climate-related risks . . . into modeling frameworks.”¹⁵

Concerningly, mainstream economics mostly works with climate-model outputs in the form of regional averages, which means that their models ignore local extremes.¹⁶ But these local extremes are likely the very things we care most about for certain economic impacts. Our built infrastructure was designed to withstand certain thresholds, including flood height, temperature level, and rainfall intensity. When those thresholds are surpassed—even just once—infrastructure can fail (for example, a dam can break), which can lead to cascading catastrophic effects.

It’s not just the housing market that exposes us to potential macroeconomic repercussions. Climate impacts can disrupt supply chains with systemic effects. For example, right after Hurricane Helene, the world learned that a majority of the electronic industry’s high-quality quartz comes from just two mines outside of Asheville, North Carolina.¹⁷ While the mines were able to recover quickly from flooding, it was a wake-up call for the industry. Whether it is low water levels in the Panama Canal slowing down international shipping or food price inflation from too-hot European summers, we’ve already had enough lessons to know that an extreme climate event in one place can have reverberating economic impacts in many places.¹⁸

The President’s Council of Advisors on Science and Technology (PCAST) under the Biden administration recognized this need for granular information on climate impacts. The council called for increased support for high-resolution modeling capabilities at government agencies and the designation of a lead agency to host an extreme weather data portal.¹⁹ This should certainly be done. But better information on the likelihood of extreme weather events is just the initial step in giving people decision-useful information. To understand a home’s flood risk, you need to know not only how intense a rainfall event could get, but also whether the surrounding drainage infrastructure is up to the challenge. In other words, risk depends on how vulnerable or resilient a given entity is in the face of the hazard.

Which means that a tremendous amount of nonweather data is also needed: drainage system design specifications, building material, topography, and more. Of course, building a model this granular for the entire United States would be a massive undertaking, but government agencies could start by prioritizing risk modeling for the areas and infrastructure of likely systemic significance. One example might be dams: There are more than 90,000 dams in the US, many of which are aging and most of which were designed many decades ago for a climate we no longer have.²⁰ A dam collapse from an extreme weather event will inevitably have downstream impacts and could destroy roads or even entire towns in its flood path.

President Biden’s Council of Economic Advisors (CEA) endorsed the PCAST recommendations and, in a notable first, called for coordination between the economic and environmental science agencies. The CEA called for the creation of a “National Catastrophic Modeling platform able to quantify extreme weather event risk . . . at a high spatial resolution across the country.”²¹ This model could then be used to inform macroeconomic modeling and adaptation investment planning.²² Catastrophe models have long been used by the insurance industry for modeling asset-level tail risks. While updating them for climate relevancy will require domain expertise from climate scientists, they are an important tool for modeling extreme events in particular places.

While the PCAST and CEA reports were exciting acknowledgments of the weakness of current US government economic modeling approaches, it’s not clear that much progress has been made in implementing their recommendations. The Treasury Department separately established a Climate Data and Analytics Hub, but, contrary to the recommendations of this essay, it was not developed in collaboration with the science agencies. Of course, over Trump’s time in office, he has implemented an extreme reversal on our capacity to understand our changing climate at the local level.

Conclusion

A new administration must bridge the methodological gap between the economic and science agencies. It’s time to make the National Climate Service (NCS) a reality. To effectively meet the needs outlined in this essay, an NCS should be designed around a hub-and-spoke approach. A centralized convening institution should focus on high-computational challenges and data coordination across science agencies, while a network of regional centers should work with local stakeholders on consultation, research, and communication. As advocated by climate scientist Bob Kopp, these regional centers could build upon our existing network of land-grant universities that already work to meet the needs of farmers and local governments in their areas.²³

To hit the ground running and show value early on, a National Climate Service might facilitate conversations around the future of insurance. Insurance companies are reluctant to share their modeled risk data, so a public data option could help local officials come to the negotiating table or design legislative reforms that can endure. It’s going to be a rocky few decades ahead of us: Let’s put our world-class science to use in helping us navigate them.