(NASA)

Saturday is considered by some to be the “peak” of the Atlantic hurricane season. In an average year, half of our hurricane activity will have already occurred. This means that for coastal residents, the risk of landfall feels like it’s subsiding and the tropics are becoming much less interesting. But how we define “peak” is actually pretty complicated.

In an average season, there would be six named storms, two hurricanes, and one major hurricane — Category 3 or higher — by Sept. 9. This year, there have been eight named storms, three of which became hurricanes. Of the three hurricanes, Gaston was the only one to become Category 3 or higher on the Saffir-Simpson scale. So, by this definition, the Atlantic is already running ahead of average in 2016.

If we consider another way to measure a season, accumulated cyclone energy, we can paint a different picture. In terms of energy, this season is just 85 percent of average.

There is a lot of subtle meaning in the word average. Which period of record is being used, and which metric of activity is best? There are no right or wrong answers to these questions, just informed preferences.


Tracks of all Atlantic tropical cyclones during 2016. (B. McNoldy)

Period of record

In the Atlantic, tropical cyclone data extend back to 1851. The storm tracks and intensities have been carefully reviewed and reconstructed using all available data sources. There are admittedly gaps and missing storms in the first 115 years of the database (pre-satellite era), but the record is as complete as possible based on ship reports and land-based observations. So one period of record from which we can construct a climatology is 1851 to present.

In December 1966, the first geostationary weather satellite was launched into orbit and provided routine complete coverage of the Atlantic basin. This was a significant advancement for detecting and monitoring tropical cyclones. The record definitely becomes more reliable and complete from this point on, so another useful climatology to construct is 1967 to present.

There’s another way you could slice it — by NOAA’s defined climate period from 1981 to 2010, the “climate normal”. Every 10 years, NOAA officially changes its baseline climatology to reflect the past three decades of observations. In 2021, the climate normal will change to 1991 to 2020, but until then, we use 1981 to 2010.

The database I use here is called IBTrACS (International Best Track Archive for Climate Stewardship), which contains storm data for the entire world, but has a bit of a lag in uniformly admitting data from the most recent year. For the Atlantic, it is current through 2014 — but including 2015 would have a negligible impact on these statistics anyway.

Measure of activity

How active has the season been compared to normal? Turns out this also has no clear-cut answer.

We can count the average number of tropical cyclones of at least tropical storm intensity (39 mph+) that are active on any given day, and that gives a nice seasonal cycle that peaks in late August through mid-September.

Or, we could calculate the daily average ACE contributed to by all active storms over the period of record. The seasonal cycles from the two methods look generally similar, but have important differences too.


Seasonal cycle of Atlantic tropical cyclone activity, as measured by the average number of active storms on a given day.  Three different climatologies are calculated (thin lines), then a 31-day smoothed curve is overlaid (thick lines). (B. McNoldy)

Based on the named storm metric, four different dates emerge as peaks in the unsmoothed series (thin colored lines) created from the various periods: Aug. 31, Sept. 9, Sept. 10, and Sept. 14. The thin lines also have many irregular and nonintuitive bumps and gaps in the seasonal cycle. The fewer years that are used in making the climatology, the noisier it will be, so the longer record makes a smoother line. The thicker shadowed lines are 31-day averages of the thin lines — this removes much of the variability (wiggles) and highlights the essence of the underlying cycle.

The only drawback to using a long-term climatology (such as 1851 to 2014) is that it may not be representative of the modern climate. The 30-year “climate normal” is created specifically for that reason: to reflect the modern climate when referring to anomalies and averages.

Now let’s look at the accumulated cyclone energy. The maxima from the unsmoothed lines for the three different periods fall on Sept. 9, 14 and 15. The same irregular bumps and gaps show up as a result of relatively few years being included in the average. The same 31-day running mean is applied to these series to smooth out the wiggles.


(B. McNoldy)

Given all this, when is the peak of hurricane season?

Since nature doesn’t flip a magic switch on a particular date, it’s probably best to call the entire second week of September the peak of hurricane season.

All the peak dates from the various time series are listed in the table below, so as long as you include the appropriate details, you can claim any of those dates as the peak of the season. To quote Star Wars’ Obi-Wan Kenobi: “You’re going to find that many of the truths we cling to depend greatly on our own point of view.”


Perhaps in another 100 or 200 years, the daily data will fill out to produce a nice smooth seasonal cycle with an obvious peak, but until then, we just have to work with what we have, bumps and all!