Hurricanes: Trend or Oscillation?

Now that the Atlantic hurricane season is officially over (though in fact hurricane Epsilon is still churning away and who knows if we won't sneak in another) a bunch of stories are appearing with retrospectives on the season. I particularly liked this graphic from the LA Times:

Would you say that shows hurricanes in an oscillation with no trend, or hurricanes rising on a trend?

The answer looks pretty clear to me: the best fit would be an oscillation riding on top of a rising trend. Note the way the last dip of the oscillation barely falls below the last peak, and the current peak is way above the prior peaks. Looks like it's going to be rather challenging developing oil and gas in the GoM for the next decade till we figure out how to do it with regular Cat 4 and 5 storms. Big version of the picture below the fold for those with wide screens.

Indeed, this looks like a nonstationary time series. You could actually filter the data with an ARIMA technique and forecast one period ahead.
Are you sure it's not just better detection?  In 1900, we'd never know about Epsilon unless a ship sailed through it and lived to tell the tale.  With no satellites, we had no way of knowing about hurricanes that didn't hit us.
There were more ships at sea in 1900.  Of course tonnage is vastly greater today, with some supermax ships up to 300,000 tons, as opposed to the 3000 ton freighter of the past.  But the seas have been well covered for a long time.  The trend is real, not just sampling error.
I believe it's a trend.

Note that one of the effects of greenhouse warming is this: as the north polar ice cap melts, it stops one of the key elements of thermohaline circulation which has historically carried vast amounts of ocean heat from the tropics to the arctic.

The large reservior of warm water which now builds up in the Caribbean and Gulf of Mexico leads to larger, more frequent, more powerful storms.

Wow, Stuart, great graphic.  Unfortunately the right-hand side is cut off on my screen (I'm set to 1024 x 768 pixels and I can't seem to go any higher.) What's the original source for that?
It is a great graphic.

You can right-click the image and "save picture as...", then view with another viewer.  The entire image comes across.

Here's the source of the image in a PDF format (there's more to it that what Stuart grabbed for TOD):

Here's the source of the actual LA Times article from which the image originated (NOTE: LA Times requires that you use their free registration to view their online articles):,1,2486171.story

I believe the "oscillation" is over a much longer time period. Like sixty or seventy years. Which is, for all practical purposes, equivalent to a trend if we're trying to deal with the next twenty years.

Here's one article.
Exactly. What I see in the picture is an oscillation with maxima around 1880, 1950, and 2020(?), and minima in 1915, 1975, or thereabouts. That's roughly a seventy year period. However, each peak looks higher than the last, and the second minima is much higher than the first.
Now pullup the tracks of the 2005 storms. At least 9 of them simply petered out over the Atlantic, and usually not in major shipping lanes.In 1933 there may have been as many.To suggest that there is no sampling error is silly. A significant part of the trend is better detection. We should compare with 1933 or 1887 for how many made landfall. Murray
I think you considerably underestimate Victorian efficiency and the extent of commerce. The Beaufort wind scale was developed in 1805. The portable aneroid barometer was invented in 1843. Systematic weather forecasting began in the 1860s (pioneered in the US). The very first hurricane warning in 1873 was of a storm that never made land.

The waters between the US and Europe have been covered with shipping for centuries. I am from the Liverpool area, and my mother, grandfather, and great-grandfather all worked in the shipping industry. Record keeping was generally excellent by the last decades of the 19th century - passenger lists, crew lists, bills of lading, all significant events during the voyage were logged. Trade between European ports (especially Liverpool) and the US South and Carribean (hurricane country) was extremely extensive because those trade links became established during the slave trade, which was extensively facilitated by Liverpool ships. Sugar, tobacco, cotton both were major imports to Europe from the South and Carribean, and manufactured goods were extensively exported (Liverpool was known as the Cast Iron Shore at one period).

I would be surprised if many major Atlantic storms were not noted and recorded by Victorian society (let alone twentieth century society). And while I don't agree with NOAA's current official position on the cause of hurricane upsurge (which I assume has to pass muster with their political bosses), I do think they've made very extensive and thorough efforts to find and document past hurricanes. I think the low years are low because the basin was quiet, not because of missing data. Here's an example of the kind of data that they has been able to reconstruct for the period of interest (tracks for the first year of the LA Times map). You can see there's coverage of a number of storms that did not make land - even tropical depressions.

Here's 1884 - a quiet year in which nonetheless, two of the known tracks don't make landfall.

And here's 1887, the busiest year in the 19th century.

Sorry for the long comment, but I think it's a very important point.

Thank you.  Another point is that since the age of steam, ships have gone pretty much everywhere.  There is no longer a need to follow the trade winds.  Aircraft and satellites are nice, but by no means necessary to document storms.
Add 3 for 2005, and the decade is on track to equal 1931-1940. The eye of Ophelia never quite made land. It probably doesn't mean anything but taking easy 50 year intervals we find 1851-1900 97, 1901-1950 95 and 1951-2000 72.
For the past to be fully represented, there had to be a ship present at the time and in the location when a storm hit max intensity, every time. Not bloody likely. Surely there were hurricanes that only got reported as tropical storms, and tropical storms that got reported as depressions. Also some storms are very short lived. Gaston, last year was only about 3 days, and in times past such a brief hurricane could well have been entirely missed if it didn't come ashore, or came ashore in a sparsely populated area.
To compare with global warming one has to take a world wide view. I have read, but don't have the reference now, that globally there has been no increase in frequency, and that 2005 was not remarkable. Also I read that the Bay of Bengal is always warmer than the GOM this year, but does not produce more intense storms.
I don't think we know enough to draw any conclusions.Murray
To get some idea for the possibility of underreporting consider to lowest years before regular aircraft patrols:
1914, 1925, 1929/30, 1939, 1946. Except 1925 each of them can be at least partly explained by a probable drop in shipping.  Murray
Well we can nip this one in the bud by just comparing hurricanes that made landfall over the period.

Anyone care to gather the data? did that, but they don't have any recent data included.  They think this "hurricanes are increasing" thing is bunk.

Junkscience? Their calculation is dated Sept 4. Also, whether they are "fair and balanced" on the science seems suspect because they have so many apparent ties to Fox News and such a one-sided perspective against climate change theory and evidence. There have been several exchanges among serious academics and so forth in the past few years, with what seems strong evidence of correlation between global warming and increasing hurricane strength. Correlation is not, of course, causation. But it seems Junkscience indeed not to hypothesize a link and investigate it.

Yes, I know that site has an agenda.  And I assure you, I believe in global warming myself.  But I don't think the evidence that it's causing more hurricanes is there yet.  The scientific consensus is that it may be increasing the intensity of hurricanes, but the number is just part of the natural oscillation.

As Dot said...someone should run the numbers for hurricanes that have made landfall in the U.S.  That should settle it.

On the other hand, including just landfall counts in the US might skew the results unless there's a consistent fraction of all hurricanes hitting the US.

Also, I wonder why we never hear of possible correlations (or at least statistical checking) of global warming and typhoon activity. Is it because the Pacific is much bigger with less intensive trade activity (and thus ship crossings) over the past century plus? If not, there should be fairly good data for comparative checking. In any event, surely there's good data for at least the last century. If hurricane hits on the US alone are a valid check, then typhoon hits on Japan, Korea, Taiwan and other North East Asian states would seem a useful comparative stat. Or is there something that makes the Pacific an invalid check?

If I were to place a bet today it would be for more/stronger hurricanes, just because that would seem to follow higher ocean temperatures.

Unfortunately, for things like global warming and peak oil, nothing is 'proven' until we see it in the good old 'rearview mirror.'

Anybody read Utterback or Christensen on innovation?  We are the early adopters.  We might test these ideas, but we don't move the market.

My cynical position is that no amount of advocacy will create mass action on GW or PO.  It will take an obvious and observable environmental change in each case.

Until then early adopters do a service, no doubt, by trying out possible solutions.

Yeah, but looking at just the last ten years the trends are obvious. even if climatologists can't go on record yet.  As an engineer I have no problem saying that the most likely possibility for next year is ~20 tropical storms, and a return to the average is a minor probability.  If next year is like this year, over half of those tropical cyclones will get loose in the GOM, and those will almost certainly intensify into at least Cat threes.  So you could say it's likely that existing GOM oil and gas production is going to go bye-bye to some degree from now on.
As an engineer, I would have a problem saying that.

It might be that what we think of as "average" is in fact way below average.  As Jared Diamond points out, a cycle that is decades or centuries long ends up being invisible to humans.  We assume the way it's been the past few decades is the way it's always going to be.  And then are surprised when it changes.  We move into the coastal areas during times of low storm activity, then get socked when activity returns to normal.  We settled the west during an unusually wet period, then are left scrambling when the normal dry pattern resumes.  

However, we may get the straight dope on hurricanes yet.  Scientists have found that you can count hurricanes in tree rings:

It could let us see the hurricane pattern for the past 500 years.

That would be nice but it's irrelevant for now.  Modelling a time-dependent complex system of PDFs where a lot of the constants are empirical has one big rule:  the most recent data gets the most weight.  For the mundane purpose of addressing what happens next year the historical "average" really doesn't matter.
As far as predicting what happens next year goes, it doesn't matter if it's global warming or natural variation.  Clearly, we're in an active phase now, for whatever reason.  

Why do we care whether it's a natural variation or global warming?  Because that tells us what, if anything, we can do about it.  If it's global warming, well, maybe we can reverse the trend.  If it's not, all we can do is get out of the way.  

Do something about it..... I certanly believe Global Warming is happening but a little working around the edges is about all we can do.  Kyoto, it's barely cutting into the growth rate assuming countries meet their obligations.
So I took a quick look at what they did. The LA Time graph appears to include all tropical cyclones, which gives about 91 storms a decade. The graph looks at just US landfalling hurricanes (which certainly is the thing of greatest practical interest), and that gives an average of 17.7 events per decade (over the period 1851-2004. Ie only about 1 in 5 tropical storms becomes a US landfalling hurricane. So then the problem is lack of statistical power. If we assume that it's iid random whether a given storm becomes a US landfalling hurricane, then the distribution would be binomial. The standard deviation for the relevant binomial distribution (picking 17.7 things from 91 things) is (sqrt(np(1-p)) = 3.8. What that means is that essentially the folks could as well be just plotting noise. In particular, no decade has a statistically significant departure from the long term average. The peak in the 1940s is 1.65 sigmas above the average, and the lull in the 1970s is 1.5 sigmas below normal. So you'd have to do a more careful analysis to see either oscillation or trend through the noise.

So the LA times may actually have made an interesting discovery by plotting it this way. It seems worthy of further investigation (and NOAA helpfully make the entire track database available online, so we can do so). In particular, I can't help thinking that the prior probability of the Ivan-Katrina-Rita sequence must have been truly miniscule.

Here's another take on the hurricane trends.  This MIT professor looked at the actual energy dissipated during each storm and season (instead of simply looking at the frequency of storms).  Measuring from the 1930s through the 2004 hurricane season, the researcher found that the hurricane seasons began showing a marked increase in energy dissipation since the mid 1970s.

What is interesting is that part of this upswing took place during a so-called period of "below average" hurricane activity.  It is definitely an interesting thesis none-the-less and I believe it to be a more scientifically rigorous way of measuring hurricane trends.

Another interesting aspect to the study, were the indications that increasing sea surface temperatures (SST) only accounted for part of the increase.  The researcher believes the increases could also be due to decreasing vertical wind shear and a potential decrease in the negative feedback cycle of deepwater upwelling after a storm passes.  In other words, storms bring cooler, deeper waters up to the surface and decrease the potential for storms in that same path for a brief period of time afterward which accounts for the negative feedback.  But since it appears sub-surface water temperatures are in a warming trend as well as the SST, there may be a weakening of this negative feedback.

Check out the short 3-page article for yourselves:

Here's the abstract:

"Theory and modelling predict that hurricane intensity should increase with increasing global mean temperatures, but work on the detection of trends in hurricane activity has focused mostly on their frequency3,4 and shows no trend. Here I define an index of the potential destructiveness of hurricanes based on the total dissipation of power, integrated over the lifetime of the cyclone, and show that this index has increased markedly since the mid-1970s. This
trend is due to both longer storm lifetimes and greater storm intensities. I find that the record of net hurricane power dissipation is highly correlated with tropical sea surface temperature, reflecting well-documented climate signals, including multidecadal oscillations in the North Atlantic and North Pacific, and global warming. My results suggest that future warming may lead to an upward trend in tropical cyclone destructive potential, and--taking into account an increasing coastal population--a substantial increase in hurricane-related losses in the twentyfirst

Thanks for the abstract.  Sounds interesting.  (The article is for paid subscribers only, BTW.)
It's a pity they didn't choose their decades from 1895 to 2005. Any one to take a whack at it? I would guess we are back to peak activity.  Murray
CDMA cell phone technology is called wide-band. My layman's understanding is that the signal/data is weakly transmitted over a wide number of frequencies. The thought is that at any one frequency, there could be spontainous noise. Trying to send a powerful signal over one frequency is the brute force method. But a weak signal could be clearly detected by simultaneously scanning many frequencies if the noise at different frequencies is uncorrelated.

What's my point? Trying to look at one noisy climactic phenomenon - hurricanes in GOM - and deduce a signal/trend is unproductive. Rather, a signal of global warming should be crystal clear when many phenomena are scanned and the uncorrelated noise of each is negated. Thus slight ocean current changes + melting of arctic ice + GOM hurricanes + precipitation changes + average temperatures + local changes + ... There must be hundered of measurable and noisy climactic phenomena which taken together ought to be sending a clear - noise free - signal that climate change is happening.

I understand the first part having had it explained to me by my brother who like me is a techno-geek, but he reads a whole lot more papers on these breaking science projects than I do, for his job.

 I don't have the skills to corrolate the Data of the Second half of the post.

 But my off the cuff guess from all I have read is that We have been heading into a trend.  I read Climbing magazines They have pictures of Ice Climbing Routes for the last 80 years, They have Places where the routes were set in with steel anchors in the Ice, these routes and climbing aids are gone.  The ice that was there is not there, its edges have moved further up the mountain.

 Tree lines have risen,  Glaciers have melted or Moved in ways not predicted,  Spring time come earlier now in the north country.  Reefs dying off becasue the water temps have gone out of their best range. A lot of little things.

 Where is the signal in all the data noise?  I Don't have a "scientific answer",  But my gut feeling is that we are in for a change and it does not matter why it is happening in as much as it is happening.  IF man did it, we can't stop it now, its set to motion in a big way and even if we cut the power all off, it would still happen.

 14 Hurricanes this season, twice the average.

You would think that someone would be putting together a Plan B for the contingency of the GOM oil and gas production being knocked out on an annual basis from now until eternity.  The Atlantic water temps are still 2 degrees above average, and show no trend to going down anytime soon.
Bottom Line:

There is no Plan B.

NEW YORK - Ford Motor Co. is likely to close five plants that employ about 7,500 workers, or about 6 percent of the company's North American workforce, the Wall Street Journal reported on FRiday.

In November, the company announced plans to eliminate 4,000 salaried jobs, or 10 percent of its North American white-collar work force, as part of the larger restructuring plan.

Assume 5 immediate dependents and a worker's $ turns over 7 times in that community.

"* In the long term, the rate of petroleum reserve additions appears
to be falling. As a result, engineering considerations indicate that
world petroleum production will peak in the 1990-2010 interval at
80-105 million barrels per day, with ultimate resources estimated at
2,100 billion barrels."

THE GLOBAL 2000 REPORT TO THE PRESIDENT: Entering the Twenty-First

A report prepared by the Council on Environmental Quality and the
Department of State. Gerald O. Barney, Study Director.

Penguin, 1982 (?)

Courtesy -PeakOil: Life After The Oil Crash

While I would say that this looks like a trend, it might also be that there is one, or more, secondary oscillations whose periods are larger that one hundred years.
Whether it is a trend or a secondary oscillation, isn't it by now recognized that there are no short term mitigation strategies of green house gas emissions?  So, whatever the cause, won't we will still be dealing with more extreme hurricane seasons in the short term (say the next decade or so)?  If we were to insitute a crash program to reduce CO2 emissions tomorrow (which is decidedly NOT going to happen) wouldn't we still be dealing with the current green house gas levels and those that are still stored in the Oceans?  If the melting subarctic and arctic permafrost is releasing large amounts of sequestered methane, are we not now in a positive feedback loop that is well beyond any human intervention on a global scale?  If the resulting warming leads to warmer tropical Ocean temps, hurricanes will be stronger, and perhaps, more frequent regardless.  So, isn't the short term and even medium term consideration best expressed by our asking, "What are we going to do about it?"  How do we best prepare for what is sure to be the inevitable?  And, aren't the answers to these questions absolutely dependent upon your particular location socially and geographically?  Does it make sense to have a "national" or "global" plan B when there are so many variables associated with social and geographic locations?
I don't know whether there are long term peaks and valleys for hurricane info that we are unable to see right now, but I do know that we are in the midst of a short term upswing.  This will probably last for the next 10-20 years.  So reagardless of what is causing this peak, whether it is part of a larger trend of ever higher peaks and valleys, or just an isolated extreme, we will have to learn how to deal with either the loss of production in the GOM or strengthen our platforms and pipelines to sustain at least category 4 storms with minimal damage.
   The next seasons weather might give a clue....If we get 4to8 cat 5,[bad] storms, some serious rethinking of how the rebuilding of NOLA should proceed,as well as the rest of the GOM.Underground utilites,no-build areas,seawall protection of of plants....a whole bunch of new ideas will have to be thought out

   Hopefully the data will be convinceing enought to quiet the skeptics...mabe NOLA as the new Venice?

William Gray and Philip Klotzbach of Colorado State University are predicting 17 tropical storms, 9 hurricanes, of which 5 may reach Cat 3 or higher for 2006. Those are almost the same numbers they predicted for this year. They also expect fewer hurricanes to make US landfall.

Here's the story on those predictions. Personally, I don't think these people are looking at the strengthening of tropical Atlantic surface temperatures that Stuart discussed. Because of that, I am wondering if they will be off the mark. Living here in Houston we all constantly marveled at the astounding sea surface temperatures in GOM this year from August through October. Every time we saw a hurricane hit that warm water, my co-workers and I had to consider our shutdown procedures for our data center. And we're all talking about how many times we might have to repeat the evacuation of last year.