Hurricane Dean: Six of the Ten Most Intense Atlantic Hurricanes Occurred In the Past Decade*

You have to hope Dean drops his rain in the mountains before he reaches Mexico City (track I'm looking at).

Best,

D

Thanks for this, Chris.

I've got my own take, with pointers to and a brief quotation from your storm punditry, at my blog.

Click my name to read "Hurricane Dean's Global Warning (Note Spelling, n not m)."

Central pressure measurements in the past were only possible if and when the hurricane made landfall at a location where pressure measurements could be made. Subsequently, aircraft measurements were made at sea, but those measurements were necessarily sparse, and probably often missed the peak. Now they estimate central pressure from satellite pictures, which means it's a continuous estimate. So the statistic is hopelessly biased.

The measured maximum of any time-varying function will depend on how often you measure. Plot central pressure at landfall for hurricanes hitting the US, and you would get better statistics.

Following up my earlier comment: the appended link has estimated central pressure at landfall for hurricanes hitting the continental United States in the last 150 years or so.

http://www.nhc.noaa.gov/pastint.shtml

Including the 2005 hurricanes, only one hurricane in the last decade had a central pressure at landfall in the US in the top ten (Katrina, 920 mb, 2005). There were three in the top twenty (Charlie and Rita added) and four in the top thirty (Ivan added). That is a slight excess over the expected 150 year average of two in the top 30, but I doubt it's statistically significant.

This isn't a perfectly unbiased data set, but it's better than what you're using.

Gerald,
I'm not sure, do you think I'm unaware of this?

All hurricane data is shot through with these types of problems. We have to take the best data, and the best theory, and see where it is that together they point.

I contend that the reasonable position, when regarding both sets of findings, is to worry. Not to state anything definitively, but to be concerned.

Your title says: "Six of the Ten Most Intense Atlantic Hurricanes Occurred In the Past Decade". If you are aware that's almost certainly not the case, well, um, aren't there some ethical issues here?

I probably should have used the word "recorded" but if you read the actual post I'm linking to, you see I explain all the uncertainties.

Still, just to be safe, I added a clarification to this post. Obviously, I am talking about *recorded* hurricanes.

Now they estimate central pressure from satellite pictures, which means it's a continuous estimate.

The NHC only uses satellite estimates for minimum central pressures when aerial recon is not available. Most Atlantic tropical cyclones receive regular aerial recon visits as soon as they are west of about 50W (this usually occurs early in their life.) So most minimum central pressures estimates from the Atlantic, or from the Northeast Pacific are based on aerial recon data. (This is quite different from the NW Pacific, where aerial recon was discontinued in 1987, and in the rest of the world, where regular aerial recon has never been used, and, as you say, most minimum central pressures are based on satellite pictures).
Satellites do not take continuous photos; US GOES West and East take a picture every 30 minutes. (Some aerial recon data is continuous, depending on the instrument, but only for a few hours on each visit. Dropsondes, considered the most accurate way to measure minimum central pressures do not make continuous measurements.) More importantly, the Dvorak method requires observing a storm for at least 3 hours. Satellite based estimates cannot achieve more than 8 samples per day. For minimum central pressure, 8 Dvorak estimates a day is more than the 2-3 dropsonde measurements per flight, but it is not as much more as your claim of 'continuous' implies.
Finally - satellite methods require (among other things) measuring the eye temperature, and thus, accurately imagining the eye with IR photography. Because satellites have limited resolution, satellite methods often underestimate TCs with small eyes, and those undergoing rapid intensification. For example, the highest Dvorak satellite classification recorded for Wilma of 2005 was 6.0 , corresponding to 115 kt sustained surface winds (borderline category 4 strength) and 948 mb minimum central pressure. Yet aerial recon data led analysts to estimate Wilma's sustained surface winds at 160 kts, and minimum central pressure at 882 mb.

It is true that the numbers in the top 10 list differ significantly in their sources (thus it is a table of apples and oranges), and it is generally agreed that the older data is more prone to underestimate storms than overestimate storms (but for a counter example, see Landsea's paper on the effect of using mis-calibrated cup-anemometers to measure wind speeds. Many old cup-anemometers were found to overestimate wind speeds by up to 50% .), but as long as we are talking about minimum central pressures in the Atlantic, or the NE Pacific, it is better aerial recon that is to blame, not satellites.

In closing - I do not know of any hurricane scientist who uses minimum central pressures to look for long term trends. Wind speeds have their problems, but not nearly so many.

Gerard Harbison:

Following up my earlier comment: the appended link has estimated central pressure at landfall for hurricanes hitting the continental United States in the last 150 years or so.

US landfalling data is a tiny fraction of total Atlantic hurricane activity. (Which itself is only 11% of global activity.)

Thanks for the correction, Chris.