The weather we stopped trying to modify

The U.S. has spent decades getting better at weathering hurricanes. Building codes have been strengthened, evacuation systems refined, forecasting windows extended and made more precise. After Hurricane Andrew devastated South Florida in 1992, Miami-Dade County adopted some of the strongest building codes in the country. Yet, twenty-five years later, Hurricane Irma still caused nearly twice as much economic damage to Florida. When Hurricane Michael tore through Florida's Tyndall Air Force Base in 2018, meanwhile, it destroyed hundreds of buildings and caused nearly $5 billion in damage to a single military installation. The base is now being rebuilt to withstand future Category 5 storms. It’s a worthwhile response. And an emblematic one, insofar as it’s solely focused on adaptation and resilience. 

The pattern is consistent across the Atlantic basin. Despite sustained investment in adaptation and preparedness, hurricane-related damages keep climbing as economic development in vulnerable coastal areas grows, concentrating more and more at-risk infrastructure and people in the path of a perennial series of storms. Almost all investment goes toward downstream preparedness, i.e., efforts to prepare for, absorb, and recover from the damage hurricanes cause. There's another approach, however, one that asks a fundamentally different question: what if we could weaken hurricanes before they make landfall? 

It's not a new idea. For decades in the mid-twentieth century, the U.S. government, as well as the Japanese government, funded ambitious, multi-agency programs to investigate scientific questions surrounding hurricane formation, trajectories, and the feasibility of upstream interventions to weaken or redirect them. Then, for several reasons, all reasonably founded, the vast majority of that work ground to a halt. The science was inconclusive, the costs were escalating out of hand, and social and geopolitical questions were and remain valid, complicating factors. For half a century, this field has lain fallow.

This is not to say there isn’t ongoing, credible, rigorous research into intervention in hurricanes; there is significant, ongoing research at many institutions worldwide on several different potential pathways to viable intervention. But the fundamental conditions that historically stifled interest in storm modification have changed considerably, making it worth revitalizing, accelerating, and creating coordinating infrastructure to advance intervention efforts.

The history of hurricane modification

The first federally funded program to investigate the viability of hurricane modification was Project Cirrus, a collaboration between the U.S. Army Signal Corps, the Office of Naval Research, and General Electric to cover "research study of cloud particles and cloud modifications." In 1947, researchers with the project dropped roughly 86 kilograms of dry ice near the center of Hurricane King. The storm appeared to weaken briefly, though it then intensified the following day before making landfall in Georgia and South Carolina. It’s unlikely, and was impossible to conclude at the time, that Project Cirrus’ intervention altered the trajectory or power of the storm. Still, the general public blamed the seeding efforts and a lawsuit followed, establishing a pattern in which social license and geopolitical questions, as well as the need for more sophisticated methodological and meteorological capacity, came to the fore after any period during which intervention research gained a modicum of traction.

While Project Cirrus was discontinued in 1952, it sparked interest and appetite for federal storm modification research efforts. Between 1954 and 1955, six hurricanes (namely Carol, Edna, Hazel, Connie, Diane, and Ione) caused extensive damage across the eastern seaboard of the U.S., reinvigorating political and scientific appetite to, well, “do something.” In 1955, Congress authorized the U.S. Weather Bureau to create the National Hurricane Research Project. Within a few years, it had its own aircraft and was studying hurricanes for exploitable weaknesses.

Perhaps the most famous of past research efforts into hurricane modification was Project STORMFURY, an experiment conducted by the National Oceanic and Atmospheric Administration and the U.S. Navy from 1962 to 1983. The project aimed to weaken hurricanes by seeding their clouds with silver iodide, a substance known to induce cloud formation by causing supercooled water droplets to freeze. The same technique is still used today in cloud seeding operations that attempt to induce rain from supercooled clouds worldwide.

Researchers discovered large amounts of supercooled liquid water inside hurricanes, which led to a hypothesis that seeding the eyewall of a storm with silver iodide and freezing the supercooled water in it might force the eyewall to reform at a larger radius. That, by virtue of conservation of angular momentum, might help reduce peak wind speeds. In 1961, the National Hurricane Research Project tested this hypothesis on Hurricane Esther. Silver iodide seeding in the wall cloud appeared to reduce maximum wind speeds by about 10% over two hours. The result was promising enough to formalize the effort: In 1962, the U.S. Navy and the Department of Commerce launched Project STORMFURY.

STORMFURY’s ambitions were high, but its ultimate success was limited. Given stringent (understandable) safety conditions governing how far from land storms had to be to qualify for potential intervention, Project STORMFURY only attempted to intervene in three storms: Beulah in 1963, Debbie in 1969, and Ginger in 1971. Debbie was the high-water mark. Over two days of seeding, the storm’s wind speeds dropped measurably from 57 m/s to 43 m/s. Researchers were cautiously optimistic. But Ginger, a weak and disorganized storm, showed no objective response to seeding. 

As storm science matured, STORMFURY’s hypothesis also began to disintegrate. In the 1980s, improved observations revealed that most hurricanes simply don't contain enough supercooled water for silver iodide seeding to work as theorized. Secondly, and perhaps more consequentially, researchers observed that hurricanes naturally undergo eyewall replacement cycles, forming new eyewalls at larger radii absent any human intervention. The Japanese government had also been conducting typhoon and hurricane modification research, but its efforts petered out toward the end of the 1970s, roughly in parallel to those in the U.S.

STORMFURY’s failure to produce conclusive results, coupled with the scale of the testing that would have been required to make additional progress (hundreds of controlled hurricane experiments), weakened appetite for additional resource allocation. Moreover, geopolitical considerations cropped up, as countries downwind of proposed experiments (mainly China and Japan) objected to being in the way of “modified” storms. By the early 1980s, U.S. agencies redirected funding for weather‑modification programs toward forecasting and atmospheric science, where the returns were clearer, would come faster and with higher likelihoods of success, and weren’t politically fraught. After STORMFURY’s cancellation in 1983, no federal hurricane‑modification program with comparable funding or size replaced it.

The stigma that stuck

STORMFURY's cancellation didn't mark the end of all federal interest in storm modification, but what followed illustrates why the field continues to struggle to regain its footing. The back-to-back hurricane seasons of 2004 and 2005, which together caused hundreds of billions of dollars in damage, prompted the Department of Homeland Security to investigate whether the government could, once again, resume efforts to reduce storm intensity directly, in addition to investing in damage mitigation. In 2008, DHS's Science and Technology Directorate partnered with NOAA's Earth System Research Laboratory to scope viable paths forward. The resulting plan was modest and methodical; the coalition would develop models focused on the most promising potential intervention pathways, iteratively narrow the list for further testing, and eventually, ideally, identify one worth advancing to small-scale outdoor experiments.

DHS launched Project HURRMIT to model hurricane mitigation ideas, such as cloud seeding and wave-driven upwelling pumps, and later established a Hurricane Aerosol Microphysics Program (HAMP) that enlisted scientists including Joseph Golden, a retired NOAA researcher who worked on STORMFURY decades earlier. But the initiative ran into a wall that had less to do with science than with institutional culture. As William Laska, a program manager at DHS's Science and Technology Directorate, described it: 

The terms "geoengineering" and "weather modification" were still heavily charged in scientific and government circles. There was, as Laska also noted, "still a sour taste from STORMFURY." DHS ultimately retreated to aerosol and cloud microphysics research. That work is essential, too, but DHS explicitly reframed it as hurricane prediction support, rather than intervention. 

The lessons from the early aughts mirror those from the 20th century: Even well-resourced federal agencies, motivated by devastating hurricane seasons and armed with cautious, stage-gated plans, couldn't overcome lingering stigma and institutional risk aversion that persisted in the field over decades and still does to this day. 

What makes this decade different? 

Investment in tropical storm modification has stalled in the U.S., and abroad, multiple times for three main reasons:

1. The science: Experiments failed to conclusively showcase the capacity to modify storms or to bear out initial hypotheses, in part due to technological constraints, some of which have since been solved or are now solvable. 

2. The capital: The scale of capital required to iterate on these experiments again, as well as to expand them to prove intervention capacity, was monumental. 

3. The stigma: Stigma in academic and scientific circles, as well as social license and geopolitical concerns, coupled with the other factors, undercut appetite to keep investing in research, programs, and supporting infrastructure. 

So why is now, more than fifty years after the largest ever experiments, the right time to resuscitate storm modification and intervention efforts? For one, much of the core science, weather forecasting, and modeling capacity needed for research has advanced considerably. Computational models have grown in power by many orders of magnitude and continue to scale annually, if not monthly. Satellite and airborne observation systems can reach resolutions that Stormfury-era researchers would have salivated over. But as the frequency and severity of tropical storms accelerate as a result of cascading climate impacts, scientists are revisiting the idea of intervening in them, this time with the tools of modern science.

Three broad categories of intervention are attracting the most scientific attention. The first focuses on cooling sea surface temperatures along likely hurricane tracks before storms arrive, using methods including engineered upwelling or deep-water mixing to deprive storms of the warm water that fuels them. The relationship between sea surface temperature and hurricane intensity is one of the most robust in tropical meteorology, and autonomous ocean sensing now makes it possible to map subsurface thermal structure in real time. The second category investigates suppressing evaporation at the ocean surface using thin, temporary films that cut off the heat transfer powering intensification, an approach validated in lab settings decades ago but abandoned because the films couldn't survive realistic ocean conditions. That’s a materials science challenge that modern capabilities are much better equipped to solve, though concerns regarding potential environmental contamination persist. The third category updates the STORMFURY playbook: modify hurricane inner-core dynamics via cloud seeding, but with better observation, targeting, and attribution capabilities, more rugged deployment architectures, and potentially more effective and environmentally safer seeding agents.

Still, the landscape of science, research, and development remains a far cry from what it was in the 1960s and 1970s, and many bottlenecks persist that render step-change-type progress in the field unlikely near-term. For any modification and intervention efforts, regardless of the intervention pathway in question, to achieve greater viability, researchers will need more sophisticated sensor and data technology, which in turn will support more sophisticated modeling and attribution. Current lab work consists of large indoor tanks, static water tests, film layers, all of which, even together, can only loosely represent real hurricane conditions. Progressively larger and higher-fidelity simulations and testing infrastructure, ones that better represent ‘real’ conditions, alongside better parameterization in models, are essential before outdoor experiments can even be considered and anyone can credibly claim meaningful hurricane-weakening effects from any given intervention pathway.

The highest-leverage areas for this intervention research span several interconnected fronts. On the computational side, the field needs higher-resolution, AI-assisted hurricane models capable of simulating candidate interventions at operationally relevant speeds, as well as better integration of aerosol-cloud microphysics into numerical weather prediction. On the observational side, researchers need rugged, low-cost atmospheric sensing equipment that can be deployed in and around hurricanes, alongside real-time subsurface ocean thermal profiling across hurricane basins. Better materials are also essential, particularly turbulence-resilient, environmentally benign surface films and cost-effective ocean cooling delivery systems that can function at meaningful scales.

Perhaps most critically, the field needs experimental and statistical frameworks capable of detecting and attributing the effects of future interventions against natural variability — the problem that ultimately undermined STORMFURY and other past projects' credibility. Lower-stakes proxy testbeds, such as marine stratocumulus or non-hurricane tropical convection, offer a path toward validating intervention physics before scaling toward tropical storm environments. And all of this research will need to proceed alongside co-developed governance and policy frameworks involving local, state, federal, and international stakeholders, infrastructure that could take as long to formalize as the science itself.

Moreover, recent years have seen renewed interest and investment flow towards weather modification efforts more broadly, particularly cloud seeding, a sector in which private-sector companies are raising venture capital and inking deals with customers ranging from farmers to ski resorts to augment rain and snowfall. The World Meteorological Organization reports that cloud seeding operations have taken place in more than 50 countries. China alone has purportedly conducted ~27,000 documented operations since 2014, a figure that is likely significantly higher, given it is based on a limited data set. While markedly different in many ways from hurricane intervention efforts, advances in cloud seeding could be useful to several potential hurricane intervention pathways, and the broader signal that attitudes toward weather modification are shifting is a helpful tailwind. 

The optimal path forward to revitalize the field

Private companies, academic labs, and government agencies are all doing valuable work on hurricane intervention. Venture-backed startups are exploring ocean cooling technologies. Academic researchers are refining models and testing hypotheses across intervention pathways. Government agencies continue to fund foundational atmospheric science. But each operates within constraints that limit how far any one actor alone can advance the field. Venture capital drives companies to commercialize on a fund-lifecycle timeline, which can pressure companies to pursue commercialization before the science is ready, creating tension and inhibiting collaboration with other players in the field. Hurricane intervention efforts, especially at the scale of storms, are highly unlikely on medium-term timelines, regardless of how well near-term revitalization efforts proceed. What a revenue model for future interventions looks like is also opaque; even if meaningful intervention is achieved someday, precise attribution of what potential damages it averted and to what extent, may remain more challenging than meaningful intervention efforts themselves. 

Government agencies, as the DHS’s experience illustrates, face institutional stigma and are subject to political cycles that make sustained commitments, let alone scaling additional capital allocation, difficult. Academic labs produce essential science but often lack mandates to coordinate across disciplines or a clear, viable path to progress toward applied, field-ready research programs. Alongside scientific, R&D, and governance components, what’s missing from this field at present is a coordinating layer: an entity to convene otherwise disparate actors around a shared research agenda, to fund the foundational work that doesn't fit VC’s timeline or any single lab's mandate, and to build key coordination infrastructure for the field.

The economic stakes reinforce the case. Financial returns, including those potentially realized by averting economic losses driven by hurricanes, could be significant. Hurricanes scale approximately with the cube of their landfall wind speed, meaning even 10–15% reductions in maximum sustained winds at landfall can reduce total damages by ~50%. Who pays for interventions, however, remains an open question. That’s not to say that the benefits of intervention couldn’t be structured into financial instruments in creative ways; in the market for financial derivatives, which is massive, there is basically no limit to creativity. But the scientific and R&D advances most needed for the field today and in the medium-term require a “first mover” funder that isn’t motivated primarily by medium- or even long-term financial gain, given hurricane intervention is likely to produce more public than private goods. Even if real-world hurricane intervention never reaches operational scale, advancing the underlying R&D — such as weather and atmospheric data collection, modeling capacity, materials science — could find application across other fields, offering valuable co-benefits regardless.

Philanthropy is best positioned to fill this void. Philanthropic capital combines the catalytic characteristics of venture capital (relatively small investments with high potential upside) with the patience and public-interest orientation of non-private sector funding. A philanthropically-funded coordinating entity could de-risk private sector investment by advancing foundational science to the point where commercial applications become more viable. It could also bridge academic silos by creating shared benchmarks and sequencing research so that early results inform later decisions, and kickstart more of the essential work of building governance frameworks and stakeholder coalitions, including with governments, multilateral institutions, and affected communities, in parallel with the science. 

A government-funded entity could do more of this work, too. But it seems unlikely governments will invest more at this point in time, given how many cycles of prioritization and deprioritization they’ve already gone through with respect to hurricane intervention. Philanthropy, specifically, the creation of a new, philanthropically-funded entity, is the best option to create vital coordination infrastructure for the hurricane intervention field at this time.

The eyewall in the room

Now, it’s time to return to the eyewall in the room. The operational realization of hurricane intervention efforts and strategies will always have to navigate social and geopolitical controversy. They always have and always will. The planetary atmosphere is a global commons. Modifying a hurricane to spare one specific region could inadvertently redirect the storm to another, potentially yielding geopolitical blowback and exacerbating already scientifically and financially challenging questions around attribution and liability.

While that scale of intervention is a long way off, it’s another reason that an uncoordinated field of actors isn’t the ideal state of play to advance this discipline. No single actor is in the position to take on the level of risk larger-scale interventions will entail, nor to build and sustain global coordination, whether on research or governance efforts. Coordination on those fronts will need to proceed in parallel with accelerated R&D; as they could take as long, if not longer, to formalize than the science, technology, and attribution systems. The urgency and opportunity grow every year as more people, more infrastructure, and more overall economic value sit in the path of intensifying storms. As such, it's essential that all workstreams accelerate today, with coordination anchored by a new, philanthropically-funded entity.

Other opportunities

The Capital Valleys Forum is coming up on May 8, as part of the inaugural Sacramento Climate Week. It features a growing, curated selection of speakers ranging from Mayor McCarty (of Sacramento), Utilities, Cisco, CalSTRS, CARB, SMUD, CalEPIC, Infinium, California Forward, HSBC and more TBA; covering water, energy, jobs, agriculture and AI, bridging discussions of tech & policy for the prosperity of California. → RSVP here.

Catch you back here on Monday

— Nick