Summary of the Unusual 2025 Hurricane Season

The 2025 Atlantic Hurricane Season officially ends on November 30, but the outlook is very quiet out there, so it very likely ended on October 31 this year. Overall, it was a slightly above-average season with some strange characteristics.

This post marks the end of my 30th year writing these updates on tropical Atlantic activity.  During that time, I have written approximately 1470 posts spanning 488 tropical cyclones including 228 hurricanes, 107 major hurricanes, and 56 retired storm names. I know some of you reading this have been following along the entire time, but whether you've been reading these posts for 30 days or 30 years, I truly appreciate your continued interest!

I began writing these posts in the summer of 1996 (prior to starting my junior year of college), and the internet and data availability looked very different then. We were still at least a decade away from social media becoming widely used. I was honored to be asked to write blog posts for the New York Times from 2007-2010 and then the Washington Post's Capital Weather Gang from 2012-2019.

I plan to continue this blog into 2026, but that also seems like a good milestone to wrap it up. The 1996-2026 period spanned many historic storms and events, and 2026 will also be the 100th anniversary of the infamous 1926 Great Miami Hurricane.

Track map of tropical cyclones during the 2025 hurricane season. Storm tracks are color-coded by intensity, and each storm's peak intensity, minimum central pressure, and Accumulated Cyclone Energy (ACE) is listed on the right.

Due to near-record-warm ocean temperatures and a neutral ENSO state (neither El Niño nor La Niña), we went into this hurricane season expecting it to be slightly more active than average. It ended much closer to average than many thought it would, and much of that difference was likely caused by a total shutdown of activity during the climatological peak of the season when there can be long-track intense hurricanes.

There were a total of 13 named storms, 5 of which became hurricanes, and 4 of those became major hurricanes (Category 3+ on the Saffir-Simpson Scale).  The climatological averages of those numbers are 14, 7, and 3. That ratio of major hurricanes to hurricanes this season is interesting -- 4 out of 5 is remarkable. In fact, looking back at the past five decades, 2025 saw a record high percentage of hurricanes that became major hurricanes (80%, compared to the average of 40%). So even though there were not a lot of hurricanes overall, the ones that formed found favorable conditions for significant intensification.

Three storms reached Category 5 intensity (Erin, Humberto, and Melissa), which is extraordinary.  First, over the past century, only 3% of Atlantic tropical cyclones reach Category 5 intensity, so that by itself is a high bar. Second, there has been only one other year with three or more recorded Category 5 hurricanes: 2005.

Enhanced infrared satellite images of the three Category 5 hurricanes at/near their peak intensities this season.  All are shown on the same color and map scale. Left: Hurricane Erin (160 mph on August 16 ~1900 UTC), middle: Hurricane Humberto (160 mph on September 27 ~2100 UTC), and right: Hurricane Melissa (185 mph on October 28 ~1400 UTC).

For the first time since 2015, no hurricanes made landfall in the U.S. The only hurricanes to directly impact land were Imelda (Bermuda) and Melissa (Jamaica, Cuba, Bahamas). And although most of the season was relatively benign in terms of landfalls and impacts across the basin, the season ended with a storm for the record books: Hurricane Melissa. Much more on that later.

Shifting away from storm counts, we can look at the Accumulated Cyclone Energy, or ACE.  This is a commonly-used metric that accounts for the overall intensity and duration of whatever storms there were.  One very intense long-track hurricane can rack up more ACE than several other storms.  For example, Erin's total ACE was roughly the same as Andrea, Barry, Chantal, Dexter, Fernand, Gabrielle, and Imelda combined!  Much like last year, it was a very strange season with little to no activity in the usual peak part of the season. Not every year follows climatology!

The chart above shows the daily ACE accrued during 2025 (yellow bars) and during the 1991-2020 climatological period (blue curve).  We saw a slow start to the season until Category 5 Hurricane Erin came around in mid-August. Activity wound down and went completely dormant from August 29 through September 16. The same data are shown below, but as a seasonal accumulation.  The season ended at 108% of average, and the last couple of years that had less ACE were 2022 and 2015.

Now let's take a little closer look at a few individual storms...

Andrea was the first storm, and it formed on June 24 in the north central Atlantic.  While that may sound typical, it was actually the latest first named storm since 2014 (that was Arthur on July 1)!  Over the past fifty years, the median date is June 15. 

Barry, Chantal, and Dexter were fairly forgettable, and then came Erin which formed on August 11 near Cabo Verde. After tracking west for several days, Erin intensified VERY rapidly, going from a tropical storm to a Category 5 hurricane in just over a day. It was also the earliest Category 5 hurricane outside of the Caribbean Sea and Gulf of Mexico (map). It reached a peak intensity of 160 mph and 915 mb northeast of Puerto Rico on August 16.

Gabrielle was the season's second major hurricane... it formed in the central tropical Atlantic, then turned north and northeast where it reached a peak intensity of 140 mph and 948 mb (Category 4) on September 23 east of Bermuda.

https://bmcnoldy.earth.miami.edu/tropics/radar/

Next in line were Humberto and Imelda. Humberto rapidly intensified from a tropical storm to the season's second Category 5 hurricane in two days. About 1000 miles to its west was Tropical Depression Nine, which would eventually become Category 2 Hurricane Imelda. Although Humberto's track forecast was fairly confident, Imelda's definitely was not (see blog post from September 27 for a refresher). A landfall with significant flooding in the Carolinas was a very real possibility, as was getting tugged out to sea by Humberto before getting close to the southeast U.S. coast.

In one of the more high-profile and high-stakes Fujiwhara interactions in memory, Humberto and Imelda did approach each other and began to rotate around a common center, their centers coming as close as 465 miles apart. For context, we starting looking for potential Fujiwhara interactions when the centers are 800-900 miles apart. Imelda's deflected track kept it far from the U.S. but it did pass directly over Bermuda as a Category 2 hurricane on October 2.

I created the map below to show the interaction between the two over the course of four days (originally posted on October 1). The inset plot shows their tracks in a geometric-centroid-relative framework.

Jerry, Karen, and Lorenzo were also fairly forgettable tropical storms in the middle of the Atlantic.

Then came Melissa.

Melissa began as a tropical wave that left the African coast on October 13. It traveled across the deep tropics and entered the Caribbean Sea on October 19 then finally became a tropical storm on October 21 around 70°W between Venezuela and Dominican Republic. It maintained tropical storm intensity for the next four days, battling vertical wind shear but fueled by extremely warm water.

As soon as the vertical wind shear relaxed on the 25th, explosive intensification commenced. It went from a tropical storm to a Category 4 hurricane in the subsequent 24 hours and kept on going. Even at Category 4 intensity (peak sustained winds of 160 mph), it underwent another period of rapid intensification to become one of the strongest hurricanes on record in the Atlantic: 185 mph peak sustained winds with a central pressure of 892 mb. This made it the third Category 5 hurricane of the season. 

Visible satellite image of Melissa at peak intensity and at landfall on October 28.

And all of that happened while the storm was nearly stationary south of Jamaica. From the evening of the 24th through the evening of the 27th, the storm drifted and meandered around, traveling just 270 miles -- that's an average speed of 3.8 mph... a typical walking speed. Not only was the sea surface temperature anomalously warm there, the ocean heat content was anomalously high (almost record-high when averaged over the entire Caribbean). Had this stall happened almost anywhere else, the storm would have upwelled cooler water from below and weakened, but it was parked over very warm and very deep water for a virtually endless fuel source.

The peak intensity of 185 mph and 892 mb is truly astounding. In my blog post on October 30, I shared these tables of Atlantic records. This first table is for lifetime peak intensity as measured by pressure (left) and wind speed (right). Preliminarily, Melissa is tied for 3rd on the pressure list and tied for 2nd on the wind list.

But when looking at a similar set of tables for landfall intensities (below), Melissa is tied for first on the pressure list and tied for first on the wind list.

Using central pressure as a tie-breaker to bump Dorian down, Melissa is tied with the 1935 Labor Day Hurricane as the most intense landfalling hurricane in the Atlantic. Of course, 1935 was well before we had aircraft, radar, or satellite data, so this was a historic case that was well-documented by aircraft, radar, and satellite.

https://bmcnoldy.earth.miami.edu/tropics/radar/

Melissa made landfall in southwest Jamaica at 1700 UTC on October 28, and both the radar and satellite presentation made it seem like the island was not even there (satellite loop). There was no disruption to the storm from the island until after landfall, and curiously, it also looked like Melissa never completed any eyewall replacement cycles, which would be expected of an intense hurricane over the course of a few days. Those restructure the storm and cause a momentary period of weakening.

Below is an estimate of Melissa's surface wind swath, for the portion of its life that included its landfalls. Although, the landfall in Jamaica was by far the strongest ever on that island and the focus of most of the attention, the Category 3 landfall in eastern Cuba shouldn't be overlooked either: it was the strongest landfall on record in the central Sierra Maestra region of eastern Cuba.

Melissa's approximate surface wind swath produced by my parametric model. I use the observed track, intensity, radius of maximum wind, and tropical storm wind radii, and apply a uniform 30% reduction of wind speed over land (the reduction factor should be more over rough terrain).

Finally, I calculated preliminary track and intensity verification statistics of the forecasts made by the National Hurricane Center and a handful of models throughout the season.  I also plotted the NHC's average errors over the past five seasons for reference (black points with dotted line).

A brief explanation of the cryptic four-letter acronyms is in order, since the vast majority of people don't know what they mean -- and shouldn't need to. (By the way, needing to know and understand what these acronyms are and the models they represent is why I'm not a fan of "spaghetti maps" being shown and shared widely on TV and online, but that's another topic).

• OFCL is the official NHC forecast in bright red.
• Then I'm showing five hurricane-specific models (run only on active tropical cyclones): HWRF, HMON, HAFS-A, HAFS-B, and COAMPS-TC.
• For track, the next two are the GFS model and the GFS' ensemble mean, followed by a consensus of skillful track models (TVCN). 
• For intensity, the next two are the Decay-SHIPS and LGEM statistical-dynamical models, followed by a consensus of skillful intensity models (IVCN).
• Then HCCA is a skillful consensus that is historically heavily relied upon by NHC.
• Last but certainly not least is this year's new Google DeepMind AI model (GDMI).

For track, most models' errors were above the 5-year average, with the GFS global model (AVNI) the worst of the lot. The only three below the average error were the HCCA consensus, the NHC (OFCL), and the leader of the pack: the Google DeepMind ensemble mean (GDMI). This is absolutely remarkable and has definitely caught the attention of many people in the field. This AI model's output was only made publicly available this past June, and it out-performed all conventional models and even beat or tied the NHC at all lead times.

The story is a bit worse for intensity forecast errors. It was a challenging season packed with explosive intensification events, and all models and the NHC had higher errors than the 5-year average. But, we once again see the Google DeepMind model leading the other models, and very competitive with the NHC forecast.

AI-based models are relatively new, but are taking the weather forecasting industry by storm (pun intended). This DeepMind model is a perfect example of that and we will see more and more of them come out. There are some concerns and unknowns with them, but so far the pros outweigh the cons. For a new model to be leading the pack for both track and intensity is unprecedented.

Hurricane Season 2026 begins on June 1, and the first few names are Arthur, Bertha, and Cristobal.  There is one new name on the list, Leah, which replaces Laura after its retirement in 2020.  As you may know, we use a set of six rotating lists for storm names, and the modern naming convention began in 1979.  The 2026 list was therefore first used in 1984, then 1990, 1996, etc.  For a little storm name trivia, this list has had 11 names retired from it over those seven years: Diana (1990), Klaus (1990), Cesar (1996), Fran (1996), Hortense (1996), Isidore (2002), Lili (2002), Gustav (2008), Ike (2008), Paloma (2008), and Laura (2020).

List 6 of the six rotating lists used for naming Atlantic tropical cyclones. The extent of each season's activity is highlighted in cyan. If a name was retired, it's in red bold font and the subsequent replacement name is in green italics font. If a name was retired upon its first time on the list, it's in purple bold italics font (Ike happens to be the only one on this list).

• Visit the Tropical Atlantic Headquarters.
• Subscribe to get these updates emailed to you.
• Follow me on Bluesky