Subtropical storm Andrea kicked off the 2019 hurricane season on May 20, becoming the first named system of the year.

Hurricane season doesn’t officially start until June 1, but Andrea isn’t the first storm to form before the season begins, though it is not common.

I’ve included some maps, charts and data below to answer some of my former students’ questions about Andrea and an early start to a hurricane season.

**Important to note is that Colorado State’s April 2019 projection of hurricane activity shows an average (more-or-less) upcoming season. It’s also important to note that for the last five years, they’ve under-predicted hurricane activity for the upcoming season every single year – some years they predicted less than half the number of hurricanes. To their credit, this is a complicated process that includes a robust knowledge of atmospheric physics and processes, as well as the appropriate statistical tools to analyze those conditions. This year, they are projecting 13 named storms, five hurricanes, and two major hurricanes, with an ACE of about 80. The average from 1950-2018 was: 11.4 named storms, 6.4 hurricanes, and 2.8 major hurricanes. The average ACE was about 100. The average for warm ENSO periods only during that same period was 10 named storms, 5.6 hurricanes, 2.2 major hurricanes and and ACE of 87.4. Of ENSO years with pre-season hurricanes, the averages were 10.6, 6.1, and 2.6 for named storms, hurricanes and major hurricanes, respectively. The ACE for those years averaged 96. That’s a lot of numbers, so here’s a chart summarizing that information:


Some quick facts about the known* first storms of each season since 1851:

  • 22 systems that started the year’s hurricane season (first-storms) have formed before Andrea’s naming-date (May 20)
  •  A total of 43 first-storms formed before June 1, the official start of Hurricane Season today (that date range was chosen in 1965)
  • 31 of those were doing the month of May.
  • If you count all storms, including storms that were the second, third, etc., of the season, there have been 52 total cyclones that formed before the start of hurricane season.  For perspective, the historical record used for analysis includes 1,182 storms, so that’s about 4.4% of all known storms occurring from January through May.
  • Another 31 storms formed in December  (about 2.5% of the total number of recorded storms each)
  • That means 83 cyclones formed during the six-month “off-season,” from December-May, which is about 7% of all hurricanes in the historical record.


So what does an early first system mean for the season?

Some quick and dirty math in R shows some correlation between the number of off-season storms and the total number of named storms (0.4368) for that same season, but not nearly as much correlation as other variables. For example, the correlation between ACE with hurricanes (0.8045) and ACE with major hurricanes (0.80755) is much higher.

FYI:  A hurricane means maximum sustained winds reach 65 knots, or 74 miles per hour (a Category 1). A major hurricane is a Category 3-5 storms with max. wind speeds greater than 95 knots (110 miles per hour).

Still, early hurricanes with total named storms is a higher correlation than another big hurricane activity indicator: El Nino strength. For the 1951-2018 period (for which ENSO data is available), off-season storms correlated as closely with the number of named storms in a season ( 0.3340972) as mean El Nino with the number of hurricanes (-0.33760802),  and more closely than ENSO with the number of major hurricanes (-0.28571113).

Click here to see a PDF version of the this mapfirststorms


Still, the best indicator for hurricane activity is ACE, but it’s not predictive. And if early hurricane activity and ENSO are both about the same correlation with the total number of tropical cyclones in a season for the North Atlantic, than either is a fair bet at guessing how this season will turn out (if you’re not in a lab with a super computer, that is).

But right now, those two variables – ENSO and early cyclones – are pointing in opposite directions based on raw correlation data.

El Nino has been holding steady since about December of last year, and it’s currently in a warm period, which typically reduces hurricane activity in the Atlantic. Early hurricane activity would make you think more storms, though, so which is it?

Let’s look at a few years of data as an example.

Assuming El Nino continues to maintain course, or even become slightly stronger or slightly weaker, the closest years in the recent past with similar conditions to that of today would be: 1951 and 1953 (slightly stronger ENSO), and 1991 and 2009 (slightly weaker ENSO). There were a combined total of five pre-season storms and 11 major hurricanes for those years. Both of those figures are way above average compared to other ENSO and off-season storm years, and much higher than the average across all years.

Interestingly, if ENSO intensifies, and our season looks more like a 1951 or 1953 season, we could be worse off than if ENSO weakens. This of course defies traditional logic with hurricanes, but any hurricane climatologist (like myself) will tell you that “the big one” can happen any year, regardless of what ENSO is doing (See: Hurricane Betsy (1965) and Hurricane Andrew (1992)).

The 1951 season claimed more than 300 lives and cost more than $80 million at the time (in today’s dollars that’s close to $1 billion).

Both years had late May storms and an El Nino conditions similar to what we have today.

So where did those storms hit in those two years?

Well, the Gulf-side of Florida was hit once in 1951 and  FIVE TIMES in 1953. With a three-year winning streak of getting hurricanes back-to-back, maybe we’ll get a break. But I wouldn’t hold your breath.

Click here to see the full PDF version of this map



What is a subtropical storm? It’s a little more complicated than simply whether the storm formed in the tropics or subtropics. You can read more about the differences between the two (and extratropical storms) here. 

*While the Atlantic hurricane record extends as far back as 1851, it is unlikely that a system like Andrea would have been detected that far back in history, as many storms are likely not included in that record. My dissertation seeks to deepen and extend that record so we as climatologists can have a fuller picture of hurricane activity in the Atlantic Basin over time.