Assessing the annual possibility of a Katrina.

August 17, 2007

Inspired by Brendan Loy’s Was Katrina a 1-in-400 years storm for N.O.?

Why I’m examining this

The US Army Corps of Engineers asserts that Katrina was 1 in 400 year storm, (see Facts and Reliability Fact sheet, for example). Brendan wonders how they came up with that.

What the USCE means when it says “400 year storm”

The calculation to make classify an event as a 100 year, 200 year … event is straight-forward. It is simply 100 divided by the annual probability of the event occurring. Thus, the USCE is asserting that the annual probability of a Katrina repeat is 0.25%

I’ll look at this in two different ways. One using published numbers and another ‘home-brewed’.

Using published numbers

The published numbers come from the United States Hurricane Landfalling Hurricane Probability Project, which comes from Colorado State’s William Gray and Phillip Klotzbach (of seasonal forecasting fame). The project calculates the probabilities of tropical storm/hurricane/intense hurricane landfall for the east coast of the United States (methodology). It’s done by dividing the east coast into 11 regions (based on intense hurricane frequency) and calculating the probabilities for each region. Using that, calculations are made for subregions and counties on the assumption that storms have an equal chance of hitting any particular spot in that region. (For example let’s say that a region has a 10% chance of a hurricane in a given year. If a subregion takes up half of the region, then it’s probability of a hurricane would be 5%).

So, if we look at the map, we find that the Katrina made landfall in region 3 (which extends from mid-Louisiana to the western Florida panhandle), sub-region 3c (eastern-most Louisiana). According to the methodology the data used to calculate the probabilities for this region is from 1880 to 2004.

Looking in the spreadsheet, we see that the annual probability of an intense hurricane hitting Plaquemines County is 0.1%, which converts to a 1000 year event. However, the probability of the subarea as a whole is 0.3%, which would be a 333 year event.

Making our own numbers

Our tool here is the Historical Hurricane Tracks webpage.   We start by selecting New Orlean and then selecting category 3-5 hurricanes for ‘all years’.  For the radius, 25 nautical miles is probably the best selection; 50 nautical miles is a bit far but I’ll include that just for comparision’s sake.

The search for intense hurricanes within 25 nautical miles of New Orleans yields a very short list. Katrina, and a storm from 1860. The record goes back to 1851, so we’ll say so there were 2 storms in a 151 year period. That gives a recurrence interval 75.5, which yields a 1.3% annual probability.

Expanding the radius adds a few more storms to the list, three from pre-1900, as well as 1969’s Camille and 1985’s Elena.  That puts us at 7 storms, which yields a return period of 21.6 years, a 4.6% annual probability.


As we can see from the first calculation, one can take reputable numbers and get a probability similar to that claimed by the Corps of Engineers.  However, one can take a different approach, include a little bit more data, and find a significantly higher likelihood.

Our probability utilizing the 50 nm radius is almost certainly too high, while Camille and Elena both affected New Orleans, they were did not have the same impact as Katrina.

So, the ‘true’ probability probably lies somewhere between that suggested by the calculation using a 25 nm radius and that suggested by the Probability Project, so somewhere between .33 and 1.33%. The difference lies in methodology, with the 1.33% number being dependent on going a few years back in the record than the Probability Project did. (This isn’t the only example of reaching just further back changes the probabilities substantially.  Georgia was not impacted by an intense hurricane in the 20th century.  However, three such hurricanes struck between 1854 and 1898.)

One consideration that could to serve to pull the estimated probabilities closer to that of the Corps of Engineers is that in our calculations, we are treating all category 3-5 hurricanes as the equal of Katrina. That is an assumption that is somewhat shaky and open to argument.  Dennis of 2005, for example, made landfall as a category 3, yet it had nowhere near the large extent high winds that Katrina did. Because of that, a Dennis landfall in the same spot as Katrina, would not have had the same effects that Katrina’s landfall did.  Because of this, one could argue for the necessity to adjust our calculated numbers to filter out storms that were not truly Katrina’s equal.  In the case of the number derived from the Probability Project, such an adjustment would likely put the revised probability very near that claimed by the Corps of Engineers.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: