Fay continues the record hurricane-free Atlantic hurricane season

On Saturday afternoon, a small low pressure system in the north central Atlantic (tagged as Invest 99L) was upgraded to Tropical Depression Six, and then upgraded again on Sunday morning to Tropical Storm Fay... the sixth named storm of the season.

As of Monday morning, Fay is located about 400 miles southwest of the Azores and not moving much. It's expected to start moving toward the southwest and dissipate in the next couple of days.

Fay has very likely peaked in intensity, just shy of hurricane status. So, there still has not been a hurricane this season. The previous record-latest first hurricane formation was September 11 (during the more reliable satellite era), so that record continues to be broken every day.  Prior to the satellite era, some storms were likely missed, but 1907 and 1914 have zero hurricanes in the database. Will 2026 join that small list?

The next decent shot at activity appears to be a wave coming off of Africa at the end of the week. I can trace it back to the Ethiopian Highlands on September 15-16. Most models develop and intensify it fairly quickly once it leaves the coast, so we'll see if that ends up becoming the first hurricane??  The next name on the list is Gonzalo.

The season statistics through today are 6 named storms, 0 hurricanes, and 0 major hurricanes. The average by now is 9, 4, and 2. In terms of Accumulated Cyclone Energy (ACE), that is at 7% of average for the date. Looking at the ACE timeline, the average season is about 2/3 over by today. If this current season is also 2/3 over, it will end up around 7% of average by the end of the official season on November 30.

The last time a season has less ACE that this was well before the satellite era: 1929, 1925, 1914, 1907, and 1905.

As I've described in previous posts, the reason for this anemic season is almost entirely the building strong El Niño. It's a well-established teleconnection that strong El Niño act to suppress Atlantic hurricane activity, and La Niñas act to enhance it. Well, there has never been an El Niño as strong as what this one is shaping up to be, so it seems reasonable that we're also seeing a record-quiet hurricane season.

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

Centennial of the Great Miami Hurricane of 1926

[this post is a reproduction of the webpage at https://bmcnoldy.earth.miami.edu/tropics/miami26/, which I first shared with the community on August 18]

This was the seventh known tropical cyclone to form during the 1926 Atlantic hurricane season, and it was the second out of three major hurricanes to impact Florida that year. An easterly wave likely left the African coast around September 6 but was not officially tracked until September 11 when it was identified as a tropical storm about 1,000 miles east of the Lesser Antilles.

The Great Miami Hurricane's track and intensity (colored by Saffir-Simpson category, which did not exist at the time); dates are labeled along the track at 0000 UTC each day.

In 1926, routine weather satellite coverage was still five decades away, so observations came from "ships of opportunity" or nearby islands which could provide helpful observations of wind and pressure from which the storm center, pressure, and intensity could be estimated.

The storm first reached land on the afternoon of September 16 as it crossed the Turks and Caicos Islands as a Category 4 hurricane with 150 mph sustained winds... this was the peak intensity of the storm's lifetime. On September 17-18 it crossed the Bahamas, still at Category 4 intensity with 145 mph peak sustained winds as it tracked over Andros Island.

Surface wind swath over the lifetime of the storm produced using a parametric model that utilizes observed storm parameters such as intensity, radius of maximum wind, radii of tropical storm force wind, and track. Any missing parameters were carefully estimated. The color families correspond to the Saffir-Simpson Scale: tropical storm in blues, Category 1 hurricane in greens, Category 2 in yellows, Category 3 in reds, Category 4 in purples, and Category 5 in grays. 

Surface wind swath zoomed in over Florida, produced using a parametric model that utilizes observed storm parameters such as intensity, radius of maximum wind, radii of tropical storm force wind, and track. Any missing parameters were carefully estimated. The color families correspond to the Saffir-Simpson Scale: tropical storm in blues, Category 1 hurricane in greens, Category 2 in yellows, Category 3 in reds, Category 4 in purples, and Category 5 in grays. 

After the eye exited Andros Island, it was a short 7-hour, 140-mile trek across the Straits of Florida. The Category 4 hurricane made landfall 13.5 miles south of downtown Miami at 6:30 am on September 18. Peak sustained winds were 145 mph, and the radius of maximum wind in the eyewall was 23 miles at the time. That put Miami directly in the calm eye for approximately 35 minutes, and in the northern eyewall with the maximum onshore winds immediately before and after that.

Estimated surface wind field during a 12-hour period centered on the time of landfall in Miami-Dade County (called Dade County then). The decay of wind speed over land is modeled using approximate land use data from the time.

The first storm warning for the Miami area was issued just 18 hours prior to landfall, and a hurricane warning was issued only 7 hours prior to landfall. Several ingredients converged to make this an especially destructive and deadly storm: 1) the extreme intensity and storm surge, 2) the urban area it impacted, 3) the large number of people living there who had never experienced a hurricane, 4) the rush of people into the open during the calm eye, not knowing the worst would come abruptly just minutes later, and 5), the very brief warning previously mentioned.

There is nothing like the 1926 hurricane in the historical record for Miami – it was very intense, very large, and passed directly over downtown. It came at a time when forecasts were primitive, warnings were insufficient, and people were inexperienced. This hurricane struck around sunrise on September 18, 1926 and instantly halted the seemingly boundless land boom, shattered the image of a tropical paradise, and plunged the area into an economic depression three years before the Great Depression began for the rest of the country.

This storm also has special significance to the University of Miami. The university was established in 1925, and construction on the campus began shortly after. The new buildings suffered extensive damage, but amazingly, the university still opened for classes in October 1926. This is the origin of the "Miami Hurricanes".

I will not attempt to retell the impacts in words and photos, that has been done in many places, most succinctly by the NWS Miami Office. Experts have shown that if this exact hurricane were to occur in 2026, it would be the costliest hurricane in U.S. history.

The only Category 3+ hurricane since 1926 to pass directly over Miami was Hurricane King in 1950. It was slightly weaker than the 1926 hurricane at landfall (peak sustained winds of 130 mph compared to 145 mph), but still a Category 4 hurricane. Furthermore, it was significantly smaller (eyewall was about 12 miles in diameter compared to 46 miles) and the warning time leading up to it was twice as long (36 hours compared to 18 hours). Those factors all helped to make King much less deadly, destructive, and infamous than the 1926 hurricane.

Then in 1992, Category 5 Hurricane Andrew did not pass directly over Miami, but the eye made landfall about 20 miles south of downtown and was very impactful. Hurricanes can be small, large, and anything in between. The Great Miami Hurricane of 1926 was very large. King was extraordinarily compact and Andrew was very small. One measure of a hurricane’s destructive power is called the “integrated kinetic energy”. The 1926 hurricane had approximately 6 times more energy than King and 4 times more than Andrew at their landfalls in southeast Florida (170 TJ vs 30 TJ vs 43 TJ).

Estimated surface wind field at landfall for the Great Miami Hurricane of 1926 (left), Hurricane King in 1950 (center), and Hurricane Andrew in 1992 (right). Select storm statistics are listed in the upper right of each panel: intensity ("VMAX" in knots), radius of maximum wind ("RMW" in nautical miles), and integrated kinetic energy ("IKE" in Terajoules). The storm track is shown by the thin maroon line.

I want to thank John Knaff, Chris Landsea, and Dave Nolan for various contributions to bringing the wind field of this historic hurricane to life. I worked with John to iterate on several values of the radius of tropical storm winds ("R34"), Chris provided feedback and approval from a seasoned veteran of hurricane reanalysis, and Dave provided a WRF-derived "roughness length" dataset to properly handle the decay of wind speed over land. To adapt the modern high-resolution roughness length data to 1926, I reduced any urban areas to woodland to somewhat reproduce the smaller and fewer buildings of the time. I then applied a smoothing filter to remove the impression of spatial precision. Both the data and the model have assumptions, approximations, and estimates, so this is only intended to provide an educated rough guess of what this hurricane's wind field may have looked like.

Additional information and resources:

    - NWS Miami Memorial Webpage
    - Monthy Weather Review article (C. L. Mitchell, 1926)
    - Florida Memory photo archive
    - Journal of Climate reanalysis article (C. W. Landsea et al., 2012)
    - U.S. Weather Bureau storm surge report, pages 19-26 (D. L. Harris, 1963)
    - Earth and Space Science damage estimates article (K. R. Mooney et al., 2026)
    - Wikipedia page

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

Hints of a first hurricane on the horizon?

As I mentioned in my post on September 11, this hurricane season is already one for the record books, but for a change, it is record quiet! There still has not been a hurricane in the Atlantic... the median date of the first one is August 6, and the record latest first one was September 11.

Well, it's September 16 and there is not even a named storm out there. However, Invest 98L has been lurking east of Bermuda for a few days and at least has the potential to become the next named storm, Fay, but maybe even the first hurricane of the season.

(the name Fay has been in circulation since 2002... it replaced Fran after the 1996 season)

Tidbit: The last time we were using this list of names was 2020 and there had been 20 named storms by now, including 8 hurricanes. In 2020, Fay formed on July 9! What a difference a strong El Niño makes!

Satellite animation of the northwest Atlantic... Invest 98L is the swirl about 500 miles east of Bermuda (the tiny red speck).

The National Hurricane Center is giving it a 30% probability of formation within a week, but that's a safe blend of diverging model guidance.

Invest 98L is currently centered at the yellow X, and the potential area of formation over the next seven days is the yellow "blob". Image from https://www.polarwx.com/tropical/

Some models' ensembles are not too excited about the prospects for this low pressure system, but one that stands out is the Google DeepMind ensemble, which demonstrated exceptional skill in its debut last season and is leading the pack for intensity and is among the leaders for track forecasts in this very sparse season. 

Intensity (left) and track (right) verification for the five short-lived storms so far this season. A variety of deterministic, ensemble, and consensus aides are shown, as well as the National Hurricane Center's forecasts ("OFCL" in red). The Google DeepMind ensemble mean is "GDMI" in brown.

Considering its demonstrated skill, even anecdotally for genesis forecasts, it's worth highlighting here. 

In the morning run, it showed a 46% chance that Invest 98L becomes a tropical storm and the ensemble mean peak intensity was a Category 3 hurricane at 107 kt. That was not the maximum, that was the mean of the 1000 members.

Now, the afternoon run was less bullish with a 19% probability of development, and the average lifetime peak intensity is at 109 kt. From this we can surmise two things: 1) there are competing factors in the atmosphere that the disturbance will be very sensitive to and 2) if it manages to form, the conditions would support impressive intensification.

The forecast track is a bit odd: drifting south for the next 4 or so days, then making an abrupt turn to the north then northeast beyond that (reminiscent of Joaquin in 2015, just displaced northeast). According this current forecast, it would reach its peak intensity near Bermuda in about one week... IF it forms.

Member tracks and ensemble mean from the 1000-member Google DeepMind ensemble. Dots are color-coded by intensity. Image from https://tcatlas.org/realtime_ir.html

The Accumulated Cyclone Energy is down to 6% for the date (4.4/71.9), and losing ground on climatology every day.  No other hurricane season in the satellite era (past 50ish years) has been so inactive... not even close. Even prior to the satellite era when storms were more likely to have been missed, only four years were less active by now: 1941, 1929, 1914, and 1907. It's likely that they were somewhat more active than we know.

ACE through September 16 from 1851-2026. The climatological average by today is shown by the dashed cyan line. The bar for 2026 is shown in yellow, which you can barely see.

Finally, a reminder about the centennial of the Great Miami Hurricane of 1926 coming up on Friday the 18th... I posted about that back on August 18 and there's a link in the post to a webpage with lots of historical tidbits and new surface wind maps created by yours truly.

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