25 October 2010

2010 Hurricane Factoids

Adam Lea, of University College London, shares these interesting hurricane factoids related the the remarkable dearth of US hurricane landfalls in recent years.  His comments are reproduced here with his permission:
As the 2010 hurricane season (with 10 hurricanes) starts to wind down I thought I would share a few statistics on how unusual this season has been historically for its lack of US hurricane landfalls:

1. Since 1900 there is no precedent of an Atlantic hurricane season with 10 or more hurricanes where none has struck the US as a hurricane. The five previous seasons with 10 or more hurricanes each had at least two hurricane strikes on the US.

2. The last precedent for a La Nina year of the magnitude of 2010 which had no US-landfalling hurricane is 1973.

3. Since hurricane Ike (2008) there have been 16 consecutive non US-landfalling hurricanes. Such a  sequence last happened between Irene (1999) and Lili (2002) with 22 consecutive non US-landfalling hurricanes, and between Allen (1980) and Alicia (1983) with 17 consecutive non US-landfalling hurricanes.

4. The period 2006-2010 is one of only three 5-year consecutive periods without a US major hurricane landfall (the other two such periods were 1901-1905 and 1936-1940). There has never been a six year period without a US major hurricane landfall.

5. Historically one in four Atlantic hurricanes strike the US as a hurricane. Thus the recent dearth in strikes should be 'corrected' in the next few years.

16 comments:

  1. What about the great Long Island hurricane of 1938? If that isn't a figment of my imagination, then 1936-1940 wasn't a 5 year period with no major hurricane landfall in the US.

    ReplyDelete
  2. Point 5 is not true at all. There is no reason to believe that the next few years will see enhanced landfalls, either at normal levels, or at levels required to "correct" the perceived problem by Adam Lea.

    This type of hyperbole is the same crapola furthered by the sages after the 2005 season suggesting huge increases in landfalls -- and the requirement that insurance rates are jacked up.

    ReplyDelete
  3. 4. looks wrong - New England Hurricane of 1938

    ReplyDelete
  4. So, if the chance of any Atlantic hurricane striking the US is 25%, the chance of no US strikes in a consecutive series of 16 hurricanes is (1-0.25)^16, or about 1%.

    Not particularly unusual. Sorry.

    Knowing the average number of hurricanes per year, I suppose one could determine the number of times this might happen in a 100 year period. It's a bit more work, though.

    ReplyDelete
  5. -2-Ryan

    I read Adam as simply saying that things should regress to the mean, hence the scare quotes around "corrected."

    ReplyDelete
  6. Hurricanes are like snow in Washington DC. This past winter we got more than 80 inches of snow. The two years before we got less than 5 inches. Counting hurricanes to the US and snow totals in the nation's capital is a very poor way to look for a climate signal. The range of what NOT unusual is just too great. (For what its worth, 80" of snow is really really unusual but it was a fairly normal winter temperature wise.)

    ReplyDelete
  7. How can there be Hurricane Ike in 2008, but a five year period from 2006 to 2010 with no US major landfall? Was Ike not major? Tell that to people who lost a roof.

    ReplyDelete
  8. Roger - if your reading is correct, the statement is poorly written - and this by a scientist who presumably has an understanding of probability.

    Personally, I don't read 'corrected' as regressing to the mean. A 'return' and a 'correction' are not synonymous in my book. If he was simply saying that there will be hurricane landfalls in the future, the statement would be devoid of real content - of course, there will be future landfalls. A 'correction' implies making up for lost events. If I shortchange you in the till, a 'correction' means putting in the money I took out, not just putting in money from future transactions.

    ReplyDelete
  9. Are the above hurricane comparisons (1900 to present) technology and monitoring adjusted?

    ReplyDelete
  10. -9-Pat Moffitt

    Since these deal with landfalls, only #1 would be influenced by monitoring issues, and that is certainly plausible.

    ReplyDelete
  11. Dr. Pielke-
    #5 seems to apply as well- unless there is some qualification I missed that "1 in 4" relates only to land-fall hurricanes.

    ReplyDelete
  12. -11-Pat Moffitt

    Have a look at Figure 3 here:

    http://sciencepolicy.colorado.edu/admin/publication_files/resource-2786-2009.47.pdf

    ReplyDelete
  13. Dr Pielke,
    Thank you for the paper-and Amen to:
    "Regardless, it seems straightforward that the more difficult it is to identify a signal in messy data the less practically useful is that knowledge."

    I noted your vonNeuman reference-- his proof that two or more variables cannot be simultaneously maximized is unfortunately not applied in environmental management. Managers are often given the impossible task of maximizing ten things absent a ranking. The fisheries resource consequences resulting from no ranking handled well by http://www.napawash.org/publications-reports/improving-fisheries-management-actions-taken-in-response-to-the-academy%25E2%2580%2599s-2002-report/

    ReplyDelete
  14. Robert #7

    Saffir Simpson Scale categories 3, 4, 5 are considered major.
    Ike and also Gustav in 2008 reached category four while at sea. They were both only category two at landfall. The factoids above only consider wind speed at landfall, not while at sea.

    The scale has been modified recently. They used to make subjective assessments of damage and adjust the category appropriately. Now they only use wind speed to establish the category.

    Here is a description of scale usage.
    http://www.nhc.noaa.gov/sshws.shtml

    ReplyDelete
  15. OK, like most scientists who look at a probability problem and decide it's too much work, I ran a Monte Carlo simulation. I took a random string of 1000 Atlantic Basin hurricanes, which would correspond to a period of 160 years at a mean frequency of 6.2 hurricanes per year, set the possibility of a US strike at 25%, with no correlation between strikes, and looked for the longest runs of hurricanes not striking the US. I got one run of 19, one run of 16, one run of 14, one run of 12, three runs of 11...below 11, it really takes off.

    So a string of 16 hurricanes not striking the US would seem to be the kind of thing that happens around once a century, or perhaps a little more frequently. Correlation, of course, would considerably increase the frequency. For example I expect that a weather pattern that diverts hurricanes might often set in for a period of weeks and divert several at a time, which would tend to increase the number of strike-free runs.

    ReplyDelete
  16. Going a bit beyond the statistics, the probability of land falling hurricanes in the U.S from the Atlantic basin is really dependent on the shape, intensity, and persistence of the Bermuda high. I have seen no science that relates directly to that factor. Has anybody else?

    ReplyDelete