What you see is that in El Niño years, the numbers of storms, hurricanes and major landfalls go down pretty significantly. It's more than just reducing the basin-wide frequency; it also does tend to reduce the odds of storms making landfall. You still get landfalls, but they tend to be weaker — which causes less damage. In addition to reducing US landfalls, strong El Niño events also significantly reduce landfall activity across the Caribbean and Central America.
SB: Does the ENSO cycle also affect storms in the Pacific, with implications for insurers who might have exposures in Japan or other parts of East Asia?
PK: Yes, if you have an El Niño, it tends to decrease your vertical wind shear across a broad swath of the Pacific Ocean. So, in the eastern North Pacific (off the western coast of Mexico) you tend to get more storms tracking farther to the west, potentially increasing threats to places like Hawaii. But obviously, Hawaii consists of several small islands in a very big ocean, and consequently it's really hard to hit Hawaii. In the western North Pacific (off the coast of East Asia), typically in an El Niño, you get a really robust monsoon trough, which is a broad area of rotation that tends to spin up storms.
If you look at the number of storms in the western North Pacific, there's not a huge difference between El Niño and La Niña, but if you look at the numbers of typhoons and super typhoons (winds of at least 150 mph), it really tends to go up in El Niño years, because storms forming further south and east just have a longer residence time over those really hot waters. A lot about where storms make landfall is driven by where they form. So those storms forming farther south and east are more likely to track towards Japan and South Korea, with potentially fewer landfalls in Southeast Asia.
SB: Switching back to the US and the Caribbean, we have seen some significant oceanic heat build-up in the Gulf this year. What are the chances of this leading to major storms affecting the Gulf coasts?
PK: Most storms affecting the Gulf coasts come out of the Caribbean or further east in the Atlantic, and those tend to be fewer in El Niño. But you can get storms forming in the Gulf, often early in the season. Typically, storms forming in situ in the Gulf tend to have fairly low predictability, especially in the early season. Tropical Storm Arthur, which recently formed in the Gulf, is a good example of this. Usually, they don't tend to be super strong, and they tend to be more rain events and probably not catastrophic wind events. I'd say that storms forming in the Gulf is kind of always a wild card. In general, they tend to be weaker storms and not massive catastrophes for the US — but it's something we will certainly be paying attention to.
SB: You mentioned that 2025 was a year of fewer storms than predicted, but more 'heavy hitters'. How did your forecast perform last year?
PK: Everything in our forecast essentially comes from ACE — the accumulated cyclone energy across the basin. ACE is an integrated metric that accounts for storm frequency, intensity and duration. This metric is heavily skewed towards strong storms, so weak, short-lived storms generate very little ACE, while strong, long-lived storms generate large levels of ACE. We forecast ACE from our different models, and then we just regress all of the other parameters (e.g. numbers of hurricanes) from ACE. So last year, we got ACE almost dead on the number, but it was a weird year in how it distributed.
In April 2025 we forecast 17 named storms and nine hurricanes, but we got 13 storms and five hurricanes. So, the number wasn't a great forecast, but the storms we got were very strong, so we also under forecast the number of major hurricanes — and their length. It was a year of low frequency, but very high severity in terms of the intensity of the storms. So that's a challenge. It's hard to say exactly how that ACE is going to break down, with some years having more (but generally weaker) storms and other years having fewer (but generally stronger) storms.
SB: What insurers would most like is maximum certainty on how many damaging, landfalling storms they can expect. Is this always going to be a limitation of seasonal forecasting, or are you working on this?
PK: To be able to say 'we think this is going to be a year with more, weaker storms' as opposed to 'a year with fewer, stronger ones' — that's always going to be a challenge. But the larger the region that you're looking at, the more useful the seasonal forecast is going to be. If you want to know if a hurricane is going to hit an individual property in Houston, Texas, I don't know of a seasonal forecast that can do that. But if you're looking at, say, a portfolio for the entire US coast, or you have insured property in the Caribbean, then these models are going to be a lot more useful for you.
We use a combination of statistical and statistical-dynamical models, using both past data and current forecasts of sea-surface temperatures as the main inputs. We are beginning to experiment with AI models too, known as 'climate emulators'. So far, they don't show very different results to our existing models, but their main benefit is they run really fast. An AI model can effectively run a hundred simulations of every historical season on a laptop, instead of on a supercomputer. I think these climate emulators are going to be very useful in the future, especially as we go from versions one or two of these emulators to versions three, four or five.
SB: So, taking account of that uncertainty, what can you tell us about the chances of hurricane landfalls in 2026?
PK: While we don't do predictions like 'we can expect three hurricanes to hit Florida this season', we do look at probabilities of landfall in different regions. Since we are anticipating El Niño conditions and forecasting a below-normal season, the landfall probabilities are below normal in 2026.
Figure 3: Probabilities of Hurricane Landfall in Different Areas in 2026
Probabilities for at least one major (Category 3-5) hurricane landfall in each of the following areas
| Entire US coastline |
17% |
43% |
| US East Coast including Peninsula Florida |
8% |
21% |
| Gulf Coast from the Florida Panhandle westward to Brownsville |
10% |
27% |
| Caribbean (10-20°N, 88-60°W) |
19% |
47% |
Source: Colorado State University analysis
We currently estimate a 17% probability of at least one hurricane striking the US coast somewhere this season, which is well below the long-run average. For the Caribbean, the forecast probability is 19%, below the average of 47%.
Our forecasts are not static. We'll update it again on August 5. Bear in mind that at the August update, while we're over two months into the hurricane season, there's still about 95% of the major hurricane activity left to go. Our updated forecasts are more accurate, because we are that much closer to what we are trying to predict. Around late June and early July the tropical atmosphere tends to 'lock in' to whatever mode it is going to be for the rest of the season.
SB: Can you speak a little about your longer-term research agenda, too? What can you tell us about how hurricane trends are developing in a changing climate?
PK: I'd highlight two things. The first is the long-term increase in the numbers of late season (October-November) hurricanes in the Caribbean, which we studied in a paper published in February.3 We found that since 1979, these are happening more often, hitting land more often, and strengthening more quickly. Hurricane Melissa last year serves as a sobering reminder of how damaging these late season Caribbean hurricanes can be. These storms can turn northward and hit the continental US, like Hurricane Michael did in 2018, with significant implications for the level of insured loss.
Why is this happening? Starting with sea surface temperatures, the long-term trend in the Caribbean is definitely warming, so you're getting a more thermodynamically favorable environment, with more fuel for hurricane formation and intensification. If you look at Pacific Ocean temperatures, we are seeing some weak long-term cooling there. We talk about this as a La Niña-like trend. This year, it's all about El Niño, but since about 1980, we've been trending more and more toward La Niña conditions. This really bumps up your Atlantic storms at the expense of your Pacific storms, which are significantly down over the long term — by about 30% since 1990 — particularly for Western North Pacific typhoons. Conversely, lower vertical wind shear can prolong the Caribbean hurricane season. Melissa, the Cat-5 which struck Jamaica last November, was emblematic of this trend.
The other piece of work I would highlight is on rapid intensification (RI), when relatively mild storms turn into major hurricanes in a matter of just a few hours. Melissa was also an example of this. We have done extensive work into RI that forms part of the research we are doing as a member of Gallagher Re's Global Tropical Cyclone Research Consortium.
We first published a paper on global trends including RI a couple of years ago and have updated it every year since. What we've found is that if you take the traditional definition of RI — more than 35mph in a day — the trend since 1990 is basically a flat line, no increase. But if you look only at 'super' RI — more than 60mph in a day — we see a significant increasing trend. To get rapid intensification, conditions have to be just right — you need warm sea temperatures, but also low wind shear and sufficient moisture. That is what we expect with climate change in general; it loads the dice toward more extreme events.
SB: Thank you, Phil, for taking us through what we can expect from the 2026 hurricane season and beyond. Remind us again when the next forecast is published?
PK: We'll update our forecast again on August 5, and you can find that at our website.
