This year’s (2023) tropical cyclone season in the North Atlantic and Caribbean witnessed a relatively high number of named tropical cyclones: 20. In spite of the current El Niño, which tends to give lower numbers. But it appears to follow a historical trend for named tropical cyclones with an increasing number over time.
The curve presented above is an update of the analysis presented in 2020 and posted here on RealClimate.
The relative high number of tropical cyclones in 2023 seems to contrast the most recent assessment report from the intergovernmental panel on climate change (IPCC), which concludes that the number of tropical cyclones is likely to decrease or remain unchanged with global warming:
The proportion of tropical cyclones that are intense is expected to increase (high confidence), but the total global number of tropical cyclones is expected to decrease or remain unchanged (medium confidence).
(IPCC assessment report 6 – technical summary TS.2.3, p. 70)
Moreover, the recent 5th National (US) Climate Assessment, NCA5, concludes in a similar vein that
[HighResMIP with resolutions between 20 and 50 km from CMIP6] simulations support the conclusion of a global decrease in tropical cyclone frequency together with an increase in intensity with warming.
These conclusions are mainly based on global climate model simulations, but current state-of-the-art global climate models are not really designed to capture all processes playing a role in such storms (e.g. convective processes and cloud microphysics).
A relevant question, therefore, is whether the evaluation of simulated historical tropical cyclones records demonstrates sufficient skill.
The observed tropical cyclone record
On a global scale, the number of tropical cyclones appears to be fairly constant as shown in the figure below, a reproduction from Bulletin of the American Meteorological Society (BAMS) State of the Climate 2022:
In other words, the statistics presented in BAMS State of the Climate seems to be more in line with unchanging numbers rather than a decrease.
It’s an interesting situation if there are different trends for all tropical cyclones on the one hand and the most intense ones on the other, or if there are trends that point in different directions over different ocean basins.
Such differences need an explanation because they would be caused by different physical conditions, processes or mechanisms.
Intense versus total number of tropical cyclones.
All cyclones, including the most intense ones, start as weak disturbances which intensify over time. A decrease in the total number of tropical cyclones, but more of those with greatest intensity, may imply that the threshold for cyclogenesis taking place is raised, reducing “infant tropical cyclones”, with a simultaneous adding of “fuel” so that they grow faster.
For instance, less frequent easterly waves off the African continent needed for triggering, or increased wind shear tearing them apart, may keep the number of new cyclones developing low.
Warmer sea surface, on the other hand, may boost their growth once they have been formed.
But recent hurricane seasons indicate both more tropical cyclones as well as more intense ones, at least over the North Atlantic and Caribbean Sea.
We have discussed the statistics on tropical cyclones in previous posts here on RealClimate: ‘Climate Change and Tropical Cyclones (Yet Again)’ (2008) and ‘Does global warming make tropical cyclones stronger?’ (2018).
Support for the warm-area-connection?
The said assessment reports seem to have ignored empirical studies on tropical cyclones statistics, such as indications of more tropical cyclones with greater ocean surface area with sufficiently warm water to sustain tropical cyclones (Benestad, 2009).
It seems that area-based climate indicators have traditionally not been widely appreciated in the climate research community (link).
A recent paper may provide support to the notion of the surface area of warm ocean playing a role for the total number of tropical cyclones. Shan et al. (2023) observed a significant seasonal advance of intense tropical cyclones since the 1980s in most tropical oceans. They found rates of earlier occurrences 3.7 and 3.2 days per decade for the Northern and Southern Hemispheres, respectively, associated with earlier onset of favourable oceanic conditions.
The widening of the time window for the season is analogous to a spatial extension of warm sea surface area. The connection between the number and warm oceans can be found both in temporal and spatial dimension.
The quality of tropical cyclone observation
One question is whether the observed record gives a representative picture of the past and if there were many tropical cyclones in the past that were not caught by observers.
Tropical cyclones are not subtle phenomena. They create huge waves and swells. After all, surfers often enjoy their legacy.
An increase in the number of tropical cyclones in the North Atlantic and the Caribbean points to increased risks for Florida which is highly exposed to tropical cyclones and sea level rise.
References
- R.E. Benestad, "On tropical cyclone frequency and the warm pool area", Natural Hazards and Earth System Sciences, vol. 9, pp. 635-645, 2009. http://dx.doi.org/10.5194/nhess-9-635-2009
- J. Blunden, T. Boyer, and E. Bartow-Gillies, "State of the Climate in 2022", Bulletin of the American Meteorological Society, vol. 104, pp. S1-S516, 2023. http://dx.doi.org/10.1175/2023BAMSStateoftheClimate.1
- K. Shan, Y. Lin, P. Chu, X. Yu, and F. Song, "Seasonal advance of intense tropical cyclones in a warming climate", Nature, vol. 623, pp. 83-89, 2023. http://dx.doi.org/10.1038/s41586-023-06544-0
“One question is whether the observed record gives a representative picture of the past and if there were many tropical cyclones in the past that were not caught by observers.
Tropical cyclones are not subtle phenomena. They create huge waves and swells. After all, surfers often enjoy their legacy.”
I think you have glossed over this a bit too strongly. When it comes to tropical cyclone counts going back 100 years or more, there is the issue of lack of homogenity in observational techniques. Prior to the satellite era it is likely that at least some short lived weak tropical storms well out at sea would have been missed and these are more easily picked up in the present day. This could largely explain the upward trend in counts in the Atlantic basin.
https://www.gfdl.noaa.gov/historical-atlantic-hurricane-and-tropical-storm-records/
Figure 1 here suggests the upward trend is almost entirely due to these short lived tropical cyclones.
I’m not saying there isn’t a climate change fingerprint in the tropical cyclone records, but in order to detect it you really need to start by making sure your records are homogenous.
Rasmus,
I have brought this up a number of times here but no one seems to see the logical connection. You leave a hanging question about the quality of cyclone observations (and I assume you mean named storms rather than TC) I am not really sure what you are getting at in the last coupe of sentences, but it seems to me that you are implying the observational record is complete as there will be a coastal impact – is that correct? That seems a little like Mann’s ludicrous idea that pre-sattelitte records are adequate “because without aircraft and satellites to warn them off, ships often encountered storms at sea, at least peripherally” Eos, Vol. 87, No. 24
A way of quantifying the increase in detection ability is to look at the mean longitudinal location of storms. I can see no reason that NA storms would have a east-west shift (happy to be corrected) and therefore any observed shift is likely related to improved detection further from land observations. Here are two images generated, one is the annual mean longitude and latitude
https://photos.app.goo.gl/kEa9YJT9gkFbsecBA
Or the same data presented in map view as decadinal averages
https://www.google.com.au/maps/d/u/0/edit?mid=1Z5xcYcbsPIFxeq3HkWQmDvmokls2DYU&usp=sharing
What we can see is that the detection efficiency was fairly constant until about the 1940s/1950s when planes and then satellites where used to actively search. This improvement is continuing
That last sentence confused me. What improvements in detection happened in the last 20 years that could account for the recent increase in total number of North Atlantic/Caribbean named tropical cyclones?
The National Hurricane Center is utilizing newer techniques calibrated to surface winds, such as scatterometer estimates and higher resolution satellite pictures. When it’s a question of deciding between 30 and 35 kt sustained winds, the difference between a tropical depression and a named storm, they have more to work with than they used to. If you’re interested, follow their discussion page on any developing tropical cyclone next year.
Thanks and will look into the discussion page next year.
Yes.
“Measure all that can be measured and make measurable all that cannot be measured. ”
(Gallilei. )
Scatterometry is rather easy and compulsary radio and optoelectronicalo techniques for material studies.
I have once seen a hurricane at sea. Hardly waves to be seen on shallow ground, the sea was just a green foam. That must be rather easily scatteromeasured.
and when all small pools are blown empty and all water is flying through the air.
Todays rain- radars seem to catch that very well. The examining radar or light wavelengths to be scattered smust be in the magnitude of blown particle sizes.
Molecular electromagnetic scattering is a bit more advanced. There we must think in terms of a resonant radio receiver and transmitter where the dimensional size of the radio with antennas may be in the magnitude order of 1 /1000 to the received and transmitted wavelength.as with
yellow sodium plasma light and CO2 backradiation.
Hi John, this is interesting. Just to clarify, your response doesn’t explicitly make an assertion on the relevance of improved detection methods to storm count, per se, However, I don’t necessarily think that there is any such relevance. I would think that improved detection methods would primarily only be relevant to the study of the temporal evolution of a particular storm, that is, exactly when a storm qualifies as a named storm. Presumably, few storms would evolve exactly to this transition point and then subside. Even if that were true and there were many borderline storms. It would have to be shown that a systemic bias exists such that forecasters were more likely to identify a storm as a named storm.
Thanks, Jonathan.
No, I am not making any specific assertion about storm count., and your interpretation is close to what I intended. I have no doubt that detection and storm intensity evaluation increased greatly with the advent of airplane and then satellite observations. However, the ability to evaluate finer detail of tropical cyclones as they develop has continued to improve. This is supported by a program of continuous improvement in satellite instrumentation, and also the ability to interpret satellite observations by cross-checking with airplane observations, when available.
Also, bear in mind that I am getting most of my information from following National Hurricane Center discussions. Tom Knutson is a real expert, and has many peer-reviewed research papers to his credit. He listed a few of those.
I’m curious: Exactly how much “continuing” detection improvement do you really think there has been in the last half century?
Quickscat:
https://en.wikipedia.org/wiki/QuikSCAT
https://www.jpl.nasa.gov/news/after-two-long-careers-quikscat-rings-down-the-curtain
It’s replacement, Rapidscat:
https://www.wunderground.com/blog/JeffMasters/quikscats-replacement-the-rapidscat-ocean-wind-sensor-installed-on-.html
I have to ask again: How much actual hurricane DETECTION improvement do these satellites give?
True…modern satellites give better and better surface wind data (and much more) for more accurate track forecasting. But more accurate track forecasts and simple hurricane detections are 2 very different things. So your cites simply don’t apply to your argument.
Specifically, what actual percent change in hurricane detection rates in the North Atlantic–the actual subject at hand here–are you suggesting occurred over the past 50 years? My guess is the detection rate change there is zero or essentially zero, but I could be wrong.
That wasn’t the question you asked. The article was about named tropical cyclone counts which includes all tropical cyclones from tropical storm trength upward. You asked “Exactly how much “continuing” detection improvement do you really think there has been in the last half century?”. I took this as meaning tropical cyclone detection, not purely hurricane detection as the article is about the former. Claiming to ask **again** about hurricane detection is illogical as you weren’t asking about hurricane detection improvement the first time.
If we are to have a decent discussion on this or anything else, there have to be rules of logic in place so we can contribute constructively. This article is about tropical cyclone counts, my response to this article was the final bit about tropical cyclone counts that I felt was brushed aside as though it wasn’t important. Hurricane counts are a different thing so if you want to ask about evidence for improved detection for counting hurricanes, that is a separate question, not one that should be framed as though I have made an assertion about the reliability of hurricane counts which I didn’t. That is why I posted links to QuickScat and its replacement because those instruments are relevant for better detection of at least the weak tropical storms that may have been missed in prior years.
To answer your question anyway, I don’t have evidence for lack of homogenity of hurricane numbers for the Atlantic over the last 50 years and I would expect those to be better detected in the past than weak short lived tropical storms. Hurricanes are more visual and impactful and harder to miss especially given the last 50 years covers the satellite era. There might be cases where very short lived hurricanes were incorrectly analysed as strong tropical storms and there might be an upward trend in very short lived hurricanes but I have no idea as I haven’t looked into it nor have I seen a similar paper along these lines to what Tom Knutson has linked to in another post.
The “question I asked” was in response to the following assertion:
“What we can see is that the detection efficiency was fairly constant until about the 1940s/1950s when planes and then satellites where used to actively search. This improvement is continuing.”
So yes, I was addressing the point according to most commonly accepted rules. Further, your long screed here provides ZERO actual evidence for your assertion above.
I will add here that T. Knutson’s Nature citation below which actually studied the ratio of missing hurricanes to detected hurricanes in the North Atlantic shows the last 50 years as being fully sampled. See Figure 4.
jgnfld, I didn’t bother trying to address your question as John Pollack addressed effectively the same question from MEV. Did you not read that? Or understand?
Here is another indication, let’s look at the number of 2-day storms… In the 1980s there were less than one per year (mean=0.8). In the 2000s there were 2.3 and so far in the 2020s we are at 3.5. Surely this is an indication of better detection rather than in increase in short lived storms???
Are you under the impression that satellites were as good in the 1980s as they are now? Sort of makes you wonder why anyone would be spending money putting more and more of them up with more and more instruments and higher resolution..
The article, and all numbers and graphs have been for named storms. Unfortunately not everyone involved in the discussion has stuck to this definition. You are possibly right that all hurricanes this year may have been detected with early satellite coverage, We just cannot know.. Take for example Hurricane Don, this year. During it’s 11 day wander around the North Atlantic, it managed to reach Hurricane strength for “several hours” 700km off the coast of Newfoundland. Would the system have been detected in the 1980s – probably. In the 2000s – definitely. Would it have been classified as a hurricane in the 1980s – unlikely. In the 2000s – ????? you tell me.
But all indications show that we are detecting more storms than 20 and 40 years ago
KW: Are you under the impression that satellites were as good in the 1980s as they are now?
BPL: Red herring. They were good enough to provide reliable counts,
Keith Woolard, Are you aware that hurricanes are REALLY, REALLY Big?
Dude, a kid with a bad-ass Estes rocket with a camera could catch most of them.
All your screeds say nothing whatever about getting COUNTS correct. Read the Nature pub that TKnutson co-authored and posted below which provides quantitative evidence which directly contradicts your evidence-less assertion.
But you are free to post any actual evidence that hurricane counts have become significantly more accurate since the late 70s/early 80s. So far, you simply haven’t. Repetition of the same assertion is ‘splaining. It is NOT providing actual evidence.
Here are some references for those who are interested in the data quality issues for this problem — issues which can lead to spurious trends in basin-wide (as opposed to US landfalling) data. The studies range from a 2008 one for for tropical storms through a 2021 for major hurricanes. Another looks at short-lived storms.
https://journals.ametsoc.org/view/journals/clim/21/14/2008jcli2178.1.xml
http://journals.ametsoc.org/doi/full/10.1175/2009JCLI3034.1
https://journals.ametsoc.org/view/journals/clim/24/6/2010jcli3810.1.xml
https://www.nature.com/articles/s41467-021-24268-5
Rasmus:
I don’t know if the weather system that raked across Florida over this past 15-17 December weekend from out of the Gulf of Mexico, then proceeding with heavy rains and coastal flooding at least through eastern South Carolina, qualified as a “compound extreme event” or simply as an “extreme event”, but post-30 November it did not qualify as an Atlantic “tropical cyclone” (at least not by National Hurricane Center standards).
Rainfalls of 8″ and 9″ were recorded at least from near Charleston to north of Myrtle Beach, with gusts and/or sustained winds of 50 mph at times. Many localities saw 4′ and 5″ rainfalls in less than six hours, resulting in local flooding, power outages, tree falls, and property damage from wind and rain and coastal flooding (can’t speak here to the timing of local tides).
As I mentioned in comments to one of Stefan’s post some weeks ago, many of the Atlantic tropical cyclones in 2023 stopped well short of the US mainland and pinwheeled into the north Atlantic just east or west of Bermuda. As those paths might (begin to) qualify as “anomalous” in terms of cumulative historical paths, this past weekend’s event perhaps shows that extreme events can take the place of hurricane landfalls in the off-season and form from dynamics unlike classical Atlantic cyclone formation.
This past weekend’s storms seem not to have been as catastrophic as a full-fledged Category 5 hurricane making a direct hit at high tide, but where the flooding and wind damage have occurred, clean-ups are now underway at whatever costs of time and financial expense.
Along with keeping an account of “ordinary” historical trends, it is perhaps becoming worthwhile to keep track of changing and emerging patterns, as or if they attain sufficient regularity.
Sincerely,
Edward Burke
Edward,
The storm you mention was extratropical, with a long fetch starting in the Gulf of Mexico. The distinguishing feature of a tropical cyclone is a warm core to the storm aloft. This one had cold air in the circulation center aloft, even in its formation area.. It was a nor’easter, as they are called.
Other than the long fetch, the storm seemed to have more in common with an earlier-season nor’easter. The whole system was unusually warm for December, and resulted in no snow for New England. I don’t have statistics, but that is relatively uncommon by December. Of course, it also contained copious moisture, which was lifted over an embedded frontal structure to produce localized intense rain.
From my perspective, it is unhelpful to compare this to a major landfalling hurricane as far as impact. The winds weren’t even as strong as a category 1 hurricane, and the rainfall amounts weren’t any greater than you’d expect from a hurricane. I’ve seen plenty of nor’easters with lower central pressure, too. This one wasn’t very well organized, as it resulted from the merger of two systems. What I am interested in is whether this is a type of storm behavior that will appear more as cold air available to these systems decreases, but moisture increases and forms a larger proportion of storm energy.
John:
Thank you for the clarifying information.
The Wikipedia account of nor’easters goes on to tell us that last weekend’s storm qualified as a classic Miller Type A nor’easter. This pattern being so well attested, it should be comparatively easy to discern histories and frequencies of events, frequencies with respect to El Nino/La Nina patterns in play, and frequency and intensity compared to those of Atlantic tropical cyclone systems.
On the basis of measured and perceived changes in water distribution patterns (AMOC weakening, e. g.) and thermal distributions in ocean currents, are models of prediction forthcoming to account for changes (increases?) in nor’easter frequency or intensity? (As I’d mentioned in reply to an earlier post from Stefan, the US seems to be seeing an eastward shift of the “tornado alley” corridor, from the Central Plains to across and east of the Mississippi Valley. Central Tennessee had a tornado in the first half of December, somewhat atypical timing but perhaps such events are becoming steadily more common as weather patterns nudge and shift.)
Sincerely,
Edward Burke
“The said assessment reports seem to have ignored empirical studies on tropical cyclones statistics, such as indications of more tropical cyclones with greater ocean surface area with sufficiently warm water to sustain tropical cyclones (Benestad, 2009).
It seems that area-based climate indicators have traditionally not been widely appreciated in the climate research community (link).”
Indeed!!!! How does one construct/calibrate and validate a climate model? Empirical data
The framing of the question:
“A relevant question, therefore, is whether the evaluation of simulated historical tropical cyclones records demonstrates sufficient skill. ”
Detracts from the fact that the ‘”skill” test is to compare model to empirical data.
Here is a fine example (thiough not specific to hurricances) from history, of the climate research community aversion to empirical data: “As will be discussed below, the general methodology of utilising past climate analogues would come under significant pressure from the Western modelling fraternity during the IPCC process due to concerns over the robustness of resultant future climate predictions” (source: https://wires.onlinelibrary.wiley.com/doi/10.1002/wcc.829)
@ R. Benestad & al
Magnificant Polar stratospheric clouds over Oslo this evening. Called Perlemorskyer. I saw them today morning in the north allready but not coloured. That entails colour spectral diffraction rather than reflection.
The litterature says that the colours come better by polaroid filters. All that entails cristalline forms.
I call upon everyone to read about them now on Wikipedia.
They form at extreemly low temperatures in the stratosphere between 20 and 30 Km, and are shown to be formed by H2SO4 .HNO3 . H2O in stoechiometric, solid state cristalline composition. Strong nitration acid so to speak .But they can be chased by Mig 25 and similars and samples taken. Thus their chemical composition has been studied.
It tells a lot about the tropopause and what can possibly get through there and what may happen to that further up. H2O is probably coming up there as CH4 and maybe NH3 as oxidized nitrogen seems also needed.
Day over tomorrow is winter solstice in the north.
The heavens on that day are mostly quite remarkable.
Benestad
Today it is peculiar news intheclimate from Norway again.
Yesterday it was very marvellous polar stratospheric clouds over Oslo now near winter solstice.
But today from Sola airport something called “Static electricity” that did stop the helicopters at Sola. And that it is rather normal over the north sea in winter they say..
I never heard of it so maybe I know that peculiar weather by other terms.
Maybe Rasmus Benestad being professional on NEPHELAI in the climate knows what it is?
But I try and phantacize in terms of quite steady weather with electrically polarized icy- microcristals that may line up macroscopic in steady weather and give near megavolts per meter vertically in the air.
As Water, a condensing climate gas, is a rather exotic material.
.Only vapour is free molecular H2O. following Daltons law. But liquid water is rather conjugated H12O6
molecules by Hydrogen bonds- flickering in hexagonal rings by brownian moovements. , That explains the quite abnormeously high frreezing and boiling points of water. Nobel price by Linus Pauling on the Hydrogen bond.
H2O also being an assymetric polar molecule explains the polar, di- electric properties. of common water.
Then solid state is rather a polymer on macrolevel. The snowflake is an ice-molecule in definite form singular.
I repeat….!
Molecules being alol invisible,that is today old supersticion ruled out by science. Rather look better.
But on nephelai and scirrus and halo- level the ice cristal particles are much smaller, they are microcristalline. But theese properties and electric moleculat properties of nebulous icy frost ought then to be rather quite important in the real climate.
Perhaps also explain better the pearl- mother clouds where microcristalline ice may rather line up macroscopically by electrostatics in space giving even polarized scattered colours. .
I happened to see the very high stratospheric cloud from Eyafjallajøkulol coming in from the west shortly after sunset. And could clearly see the difference to verry high icy scirrus in the sunset.. The very special Eyafjallajøkull cloud showed no perearlmother effects at all, only light uniform yellow ocker. Thus rather plausible very fine and chaotic solid state stone- dust different from ice- dust. And next day there was some light yellow ocker fine dust on the cars.
That stone- dust melts to glass in the jetliners and clogs the high yellow hot driving turbine blades.
In that way, Island stopped the very EU Jet traffic for some days. It was their way of participating in the EU, they could confess. .
But then we have 4 fameous greeks in the real climate allready.
1 Aristoteles who wrote that “What goes up must come down!”
2 Aristophanes who wrote about NEPHELAI
3 Zevs with up to 1 Megavolt/ Meter at disposal in the lower troposphere, and 6 giga- volt at hand. when it matters.
4 Electra… who shows up at Sola in winters. and downs the helicopters.
The global economic consequences of the year’s hurricanes and tropical cyclones may be eclipsed by the Panama drought. Water levels in the Gatun lake that feeds the locks and forms the main shipping channel of the Panama Canal have fallen to leves thought impossible in the days of its construction.
Persistant tropical rains were the bane of those digging it, , but besides the driest stretch of years in a century, Panama’s burgeoning population is extracting water from the lakes watershed at an unprecedented rate.
With the Southern approaches to the Suez canal made problematic by Yemeni attacks in reaction to the war in Gaza, Panama has become a very highly leveraged factor in the circulation of global trade—
Global economics may be more susceptible and sensitive to “tipping points” than the global circulation.
I understand why empirical models and physical models could differ due to the physical model’s failure to fully capture the nuances of cloud microphysics or storm convection. Likewise I could believe the empirical data are incomplete. But why would the physical models fail to depict changes in sea surface temperature (size of potential tropical cyclone development area) or increases in the length of the tropical cyclone season (again based on sea surface temperature)?
Re: Andrew Jan. 5
To which “failure to depict changes in sea surface temperature” are you referring to?
And still the Deniers deny.
Like Keith Woollard.
They will deny to the very end. Could they be any more irrelevant?