Thursday, April 06, 2006

Blow the house down:

If you follow climate news, you will have heard some noise about increases in the number and intensity of hurricanes. If you follow these things closely, you will have heard that there is discussion of whether the increase in intensity is due to climate change. If you really pay attention to detail, you will have heard that there are some who believe that the increase in storm strength is a cyclical thing, and not due to climate change at all.

I went to a talk by Kerry Emanuel today. Kerry is a professor at MIT, where he does all sorts of number crunching on hurricane data, and high-falutin' theorizing on the genesis, growth, and decay of tropical cyclones. I realized as I walked into the room that in my seven years at MIT as a grad student and post-doc in the same building as Kerry, I had never seen him, even though I had several friends who were students of his.

Kerry is firmly in the camp that believes the increases in storm strength are due to climate change - specifically from the warming of the sea surface. But he's interested in more than simply that - he wants to know how the energy budget of the ocean-atmosphere system works today, and how it worked in the distant past.

Some points from the presentation Kerry Emanuel made here at NSF:

  • Early April is a global minimum for hurricanes
  • 10% of the global total occur in the North Atlantic, but they get 90% of the press coverage
  • Hurricanes do not form within a few degrees of the Equator because they need Coriolis forces - they also need warm sea water (Quizlet alert...)
  • There are generally no cyclones in the South Atlantic because the water is just slightly too cold except in a narrow strip off Brazil. The first recorded hurricane there, Catarina, only just classified as a hurricane before it went onshore
  • Hurricanes are basically Carnot heat engines - they are driven by the difference between sea surface temperature and the atmospheric temperature, and are therefore directly related to the greenhouse effect
  • Insurance losses vary roughly as the cube of maximum wind speed (that is why this number is predicted)
  • Over 50% of all damage in the US was caused by the 5 great storms prior to Katrina (I researched these numbers, they are not Kerry?s):
    • Andrew, 1992 ($20.3 billion, 2006 dollars: $28.9 billion*)
    • Charley, 2004 ($6.7 billion, 2006 dollars: $7.1 billion)
    • Hugo, 1989 ($6.2 billion, 2006 dollars: $6.6 billion)
    • Ivan, 2004 ($6 billion, 2006 dollars: $6.4 billion)
    • Frances, 2004 ($4.4 billion, 2006 dollars: $4.7 billion)

  • Katrina itself is equal to all those together, with insured losses estimated at $40-60 billion, and total losses at over $200 billion
  • The next five in the list are:
    • Georges, 1998 ($3.27 billion, 2006 dollars: $4.0 billion)
    • Jeanne, 2004 ($3.24 billion, 2006 dollars: $3.43 billion)
    • Opal, 1995 ($2.5 billion, 2006 dollars: $3.3 billion)
    • Iniki, 1992 ($2.09 billion, 2006 dollars: $2.98 billion)
    • Floyd, 1999 ($2.1 billion, 2006 dollars: $2.52 billion)

  • This list does not include damages outside the US, or from the other known Category 5 hurricanes:
    • Labor Day storm, 1935
    • un-named storm, 1947
    • Dog, 1950
    • Camille, 1969
    • David, 1979
    • Gilbert, 1988
    • Mitch, 1998
    • Isabel, 2003

  • [I note as an aside that insurance losses are not a good index for increases in storm strength, even after correction for inflation. They do not account for increases in population in vulnerable areas.]
  • Over 90% of the damage has been caused by category 3 storms or greater (and there have been only 30 of these since 1870, causing statistical problems for insurance)
  • We can get a good count for the North Atlantic back to the 1870s because the ship traffic was so dense
  • The annual global total of cyclones has been 90+/-10 since 1970 (when we first had continuous global coverage) -- for statistical purposes, this frequency is constant, and we have no evidence that the number of storms is increasing
  • The argument about a trend (and from there, about natural cycles in hurricanes) is based on an extremely limited dataset that is not statistically significant
  • The resulting pattern of storm numbers has no recognizable trend, but it is a very good match to sea surface temperature, which is also well known over that period
  • There will be no quiet decades (hurricane-wise) in our lifetimes (but there will be quiet years from ENSO or volcanic activity)
  • "Global warming" means different things over the oceans ? heating in the high and medium latitudes, but cooling in the tropics. If sea surface temperature increases, this makes for a stronger Carnot cycle.
  • The argument about solar variability is still constrained mostly by measurement problems
  • 50 million years ago, during the Eocene, the tropics were a few degrees warmer, and the higher latitudes were about 15C warmer. Climate models cannot replicate this temperature distribution without hurricanes to transport the required heat

The take-home is that while the frequency of storms has not increased noticeably, the number of strong storms has. This has great implications for coastal development and insurance.

*: Constant dollar values calculated with the Bureau of Labor Statistic?s handy Inflation Calculator. Base figures are from the Insurance Information Institute.

No comments: