Suppose that higher SST indeed causes a greater frequency of intense cyclones. How can that be statistically detected? My simple approach above is to check for linear correlation between SST and storm count. My assumption is that, if higher SST leads to higher intense storm count, then the population mean should shift and that shift should be evident in a linear correlation.

But, my sense is that in a Poisson situation I may need a large number of data points to truly explore for any SST/cyclone relationship.

My thinking is fuzzy and anything you or anyone else can say to de-fuzz me will be much appreciated.

Southern Indian Ocean:

Mean = 3.73 Counts per Year
Probability (p) from chi square test = 0.33

Near Australia:

Mean = 0.92 Counts per Year
Probability (p) from chi square test = 0.92

SW Pacific:

Mean = 0.78 Counts per Year
Probability (p) from chi square test = 0.34

Not much of a SST/cyclone relationship in those splits. If I combine the two then the r-squared rises to 0.27, but I need to learn a bit more about the region to see if it makes sense to do that combo.

This uses NCEP reanalysis SST, which I’ll check against other SST constructions, but I doubt that there’s much difference.

There doesn’t seem to be a relationship.

The contest will close at 12Z 6 June.

It’s now 12 Zulu time (which is the same as 12 GMT or 12 UTC). The contest is closed and the race for prognostication immortality is on. Staffan, I’ve noted your ACE entry as 133.

Much Below Average

Sam Urbinto

Below Average

matt vooro (ACE of 75)
Ryan Maue (72)
bender (78)
paminator (79)
Bob Koss (73)
Jonathan Schafer (72)

Average

Dan Hughes
1950-2007 climatology
John A
Paul Linsay
WSR
John Norris (90)

Above Average

(terry)
Steven Mosher
TSR
ma in va
1995-2007 average
Gray/Klotzbach
raleighwx
David Smith
Kenneth Fritsch (89-201)
US NOAA
Accuweather
Weather456
Bill F (112)
Staffan/Swedish Hurricane Centre (133)

Much Above Average

Mike B
Reference (202)

Since you are a climatologist with the Australian Bureau of Meteorology, might you please be so kind as to give the official definition adopted for “surface sea temperature” for the the purposes of comparison with storm intensity? Thanks Geoff.

One important point: the set which is currently (6 June) on the web at ftp://ftp.bom.gov.au/anon2/home/ncc/cyclone/cyclones_newformat.zip is NOT the ‘new’ set. The ‘new’ set was only formally adopted as the official Australian best-track set a couple of weeks ago and we haven’t yet uploaded it to that part of the web, but expect to do so within the next week in place of what is there now (when it does go up it will be a zipped Excel file rather than a zipped text file, so you’ll be able to tell when the change has happened).

Grant Anthony (#259) is broadly correct as to the nature of the new data sets (although there have also been some changes to tracks). A reanalysis of intensities wasn’t, in general, part of the scope of this part of the project – that will come later. However, we have done a fair bit of analysis about what issues exist with intensities in general, and have reached the conclusion that there are no significant systematic biases in intensity after 1980 in the Australian region (I can’t speak for the broader SH), although the error bars on the intensity of individual cyclones are still substantial. The major intensity biases are pre-1970 but the analysis under discussion doesn’t extend to that period.

The bulk of the amendments to the data set that is on the web now will be in the pre-1970 period, especially between 1955 and 1970.

The database work hasn’t got as far as a peer-reviewed publication yet but hopefully that isn’t too far away. In the meantime the best available documentation is an extended abstract I presented to this year’s AMS Annual Meeting (available via the AMS website).