Wednesday, April 7, 2010

Intensity or Frequency?


I have previously argued that – in my estimation – there's a strong causal association between sea-surface temperatures and the number of named storms (or tropical cyclones) in the Atlantic Basin. Statistically, the association is quite significant, and graphically, it is evident once you apply very simple smoothing filters.

This is not the prevailing scientific view, which essentially says that the intensity of storms should increase with global warming, and the frequency of storms should actually decline. This prevailing view, largely based on computer modeling and not observations, is best summarized by the IPCC in 4AR WGI 10.3.6.3.

I was wondering if both could be true: global warming increases the frequency and intensity of tropical cyclones. How can we test this idea using available observations? You can't just look at, say, Accumulated Cyclone Energy (ACE.) If the frequency of storms increases, ACE should also increase, even if the average intensity of each storm doesn't change.

It occurred to me that a much better test would be to look at the ratio of hurricanes to all named storms, and the ratio of major hurricanes to storms. I've done this, but I'll leave it as an exercise for the reader. It's a very easy analysis. You can use the named storm count data from the Hurricane Research Division of NOAA. If you have concerns that tropical storms were under-counted in the past relative to hurricanes (a reasonable assumption), you can use data starting in 1944, which is when systematic aircraft recognizance started. But remember, causality matters more than the trend in this case.

To make a long story short, observations do not appear to support the view that global warming will cause storm intensity to increase. The historical data is telling me the opposite of what the IPCC claims. What, if anything, am I missing? Could it be that things will work differently in the future?

11 comments:

Anonymous said...

I think you might want to consider impulse. ie how quickly a storm forms and rises to a dangerous intensity.

This also highlights duration of storms at particular intensities.

steven said...

Looking at your orginal analysis I would take series issue with the application of a filter ( central moving average). First because it is an acasual filter and you are trying to establish a causal relationship and secondedly because anytime you smooth series like this you generate more confidence in the result than the data warranted.

http://wmbriggs.com/blog/?p=197

As Briggs would say "dont smooth your data hockey puck.

Joseph said...

@Anon: Is there data on that?

@Steven: The smoothing is not part of a statistical analysis. It's just used for visualization. What I actually use for statistical analysis is a linear regression of both series, after I've detrended them, with full noise. This way you take care of any systematic trend bias.

Now, while it's true that smoothing would tend to produce better apparent correlations, what you see in this case is subjectively just too obvious to be anything other than causality. Or are you aware of any other two series that have nothing to do with one another and look like this after you produce 15-year smoothings of each one? In fact, if I were to smooth the ACE series, for example, you no longer have as good a subjective correlation. If I were to smooth a global temperature series to replace the NH SST series, the correlation would not look as good either. So are we talking coincidence here, even though there's clearly plausibility? I think that's beyond unlikely.

steven said...

Ok, I always get suspicious of smoothed graphics.

The detrending is an interesting step from a phyiscal argument perspective. I'll have to look at your original stuff to see how you detrended it.

Other thing I would look at is the count data. lots of issues with that. I'm not so impressed with analysis when the underlying data has known issues. I've followed a bit of kenneth's work ( at CA) but not enough to render a judgment. On the surface, warmer SST = some kinda change in hurricanes. Since it has gotten warmer and since sea surface temp does play a role we expect to see some relationship. Did you look at global SST or SST just in the formation zones and impact Zones?
and data for the whole year or just in season?
Just curious not critical. Oh, I like the thing you did on the megacities.

Joseph said...

Other thing I would look at is the count data. lots of issues with that. I'm not so impressed with analysis when the underlying data has known issues.

I get that, but that's primarily a criticism you can apply to claimed trends. For example, if I said, "there's clearly a trend in the number of named storms," you could counter with "wait a minute - are you sure they counted all storms in the past?"

But I haven't made claims about the trend. I have made claims about causality. That's different.

Did you look at global SST or SST just in the formation zones and impact Zones?

I looked at Northern Hemisphere SSTs from HadSST2. I'm not sure if there's more granular quality data than that, spanning at least a century.

llewelly said...

On the one hand, your analysis only applies to the Atlantic, which represents only 11% of global tropical cyclone activity. On the other hand, the data available for the Atlantic is by far the most reliable ...*

Nonetheless - it would be interesting to see your analysis applied to data for the other areas of tropical cyclone activity, particularly the NW Pacific (by far the most active basin, accounting for 38% of global activity, and having the highest proportion of intense hurricanes), and southern hemisphere.



* Interestingly, only the Atlantic shows a significant upward trend in total number of tropical storms over the last 50 years. And the Atlantic also shows the largest, and least disputed upward trend in number of intense hurricanes.

Joseph said...

@llewelly: I haven't done that, but as you note, both the tropical storm data and SST data are probably much less reliable (with random noise) outside the Atlantic.

Steve Bloom said...

Joseph, Kerry Emanuel has just upended the TC science apple cart. See also Jim Elsner's related recent work (discussed by Jeff Masters here).

All the ducks seem to be in a row for this season.

Joseph said...

That's interesting, Steve.

I will note that "intensity" has a pretty clear 35-year cycle in the data, if you measure it as number of hurricanes per storm, or ACE per storm. It requires a bit of smoothing to see it.

Eric L said...

What I've heard is not that the average storm intensity would get worse, but that the worst storms would get worse. Keep in mind that the very definition of hurricane depends on its intensity -- must have sustained winds of 75 mph. When you look at hurricanes, you're looking at the upper tail of the distribution of all storms that form over the ocean, cut off at a fixed intensity that we don't adjust as the climate warms. If the Atlantic creates more storms, I would expect that to mean a lot more tropical storms and category 1 hurricanes, and a few more category 5, so it would be hard to see this effect by looking at average huricane intensity. Maybe looking at the top 3 for each year?

Anonymous said...

The key source for this is not the IPCC AR4, which in this particular fast-moving subfield of climate change science is already somewhat outdated, but rather the Nature Geoscience Review on "Tropical cyclones and climate change" by Emanuel, Knutson, Landsea, and several of the other big names in hurricane analysis. There is apparently still disagreement about the hurricane count even post 1944: the contention is that there has been a recent uptick of short (< 2 day storms) which can be attributed to satellite monitoring improvements.

Sadly, the paper itself is behind a paywall... perhaps it might be worth looking at the US GCRP report on tropical cyclones, which does show the SST/frequency relationship (http://www.climatescience.gov/Library/sap/sap3-3/final-report/sap3-3-final-Chapter2.pdf, pg. 58), but I think even this 2008 analysis may be obsolete in face of the Nature Geoscience paper.

A little more searching suggests that this paper may be one of the key recent analyses: http://journals.ametsoc.org/doi/pdf/10.1175/2009JCLI3034.1

-M