Have Atlantic Hurricane Seasons Gotten Longer?

Probably. Or, to be more exact, the seasons seem to end later these days than they did in those days, if "those days" is defined as the first decade or so during which reasonably good (though not perfect) data were collected compared to now. A more immediate question may be: When will this year's hurricane season end?

Officially, the Atlantic Hurricane Season runs from June 1st through November 30th.¹

I am not certain, but it is probably the case that those dates were picked some time ago in order to include all of the days during which we were likely to have Atlantic storm activity of interest to the Hurricane Prediction Center or other agencies of the past or present. In other words, the beginning and ending dates are meant to be inclusive and liberal, not averages or conservative estimates. As long as everyone who is supposed to deal with active major tropical storms in the region is ready to do so on June 1st, and plans to be available and not away at conferences or off fishing somewhere through November 30th, we're covered. Therefore, it is of interest to note that several times in the last dozen years or so the Atlantic Hurricane Season played out in overtime. The following data show the last full 12 years of hurricane activity, giving the year, number of storms that year, start and finish dates of the last storm, and the name of that storm.

Of the 12 seasons listed, almost half (5) extended past the season, with three of those years seeing a storm begin its activities after the season was over. We might ask if this is because there is a shift in the season's dates, a shift in the intensity of storm activity causing late season low pressure events to mature into cyclones more often, or some other meteorological factor that has caused the estimate of a good, guaranteed to be almost always inclusive season end date to become a bad estimate.

Indeed, it does seem that if one was using experience, an end date of November 30th would be reasonable. The following table gives the storm season stats for the first 13 years for which data of roughly comparable form as the data given above are available.

Here we see that there are no storms that extend past November 30th. Eight of these 13 years see the end of the season before November 1st, and only one year has the season extend past the middle of November.

This could be a function of quality of data. The first data set does include satellite observations, but nothing like we have now. I'm not sure if there are any buoys collecting data in the early data set, but there are observations coming from shipping. Ships with radios have been reporting storms since 1900 or so. Aircraft were being used to survey hurricanes for this entire time, though the most advanced methods for doing that came somewhat later. So, while the period of 1959 to 1970 may seem to be a sort of dark ages for storm identification and tracking, it is not as bad as one might think.

The following graph, prepared by Robert A. Rohde, shows the relationship between observations of Atlantic cyclonic storms and technologies used to identify and track them.

While adding technology certainly increased the level of observation over time, for the more recent period, introducing radar and satellites seems to have had little effect on the trends seen in this chart, and even aircraft observations start about ten years after an apparent increase in overall storm activity. Oh, and if you adjust this graph only a little, to look at the last few decades only, there is a bit of a hockey stick effect! Overall, this graph shows that there is a lot of variation in the level of storm activity, a rough correlation between overall activity and frequency of hurricanes as a subset of that activity (with "major hurricanes" following along in proportion, as one might expect). The graph also shows that there is more activity in the latter part of the 140 years than the earlier part, and that this difference is probably partly accounted for by the addition of new technologies.

One could argue with those assertions, but either way, the data I give above are entirely from within this latter part of the data set, so I think it is reasonable to suggest that the actual, meteorological end of the Atlantic Hurricane season was comfortably prior to the nominal end of November 30th during the 1960s, but is as of late, in the early part of the 21st century, not.

It is very hard to imagine that a Category One or Two hurricane would float around in the Atlantic for a week or so and not be noted by mariners. So, while the given number of storms and end dates of the seasons may be a little inaccurate for the early data set, it is unlikely that half, or even a quarter of these years actually saw significant cyclone activity after November 30th that simply did not get into the data set. Maybe one or two storms, but not five or six.

Of this year's storm, Tomas would have had a reasonably good chance of being missed when it formed, but not after it hit St. Lucia and later, Haiti. Julia may well have been missed because it was small and did not go near land and seems to have been away from major shipping channels. Similarly, Shary may have been missed. However, even these smaller storms tracked across thousands of miles of ocean and would have been picked up by ships. Small and remote storms like these may very easily be unrecorded for pre-radio days, but for periods with ships radios and aircraft, it is likely that they would be recorded.

This is what the 2010 season looks like so far in the same format as the previous tables.


2010 21 10/29 11/07 Tomas


With 21 storms, this year is the second most active year in well over a decade, and is about twice as active as the most active years of the earlier data set. And, there is another storm that has a reasonable chance of forming.

Showers and thunderstorms over the southern caribbean sea off the coast of Colombia and Panama are associated with a broad area of low pressure. Conditions appear favorable for some development of this system over the next couple of days as it drifts slowly westward or west-northwestward. There is a medium chance...30 percent...of this system becoming a tropical cyclone during the next 48 hours.

If this turns into a named storm, it will be called Virginie, which I imagine to be pronounced "Virgin nyeee" as one might say the name of Old Dominion with a Tennessee accent. And, if Virginie does form, it will place this year into the upper half of modern day (21st century) years in terms of lateness of season.

¹This fact and the data used to make the tables in this blog post are all from the National Hurricane Center web site.