Support this site:

Add our blog feed to your favorite reader.


Posts by Date

March 2013

2013 Atlantic Hurricane Season Likely to be More Active than Normal

   Posted by Levi at 8:15pm on March 17, 2013

It’s that time of year again to start talking about the upcoming Atlantic hurricane season. 2013 continues the streak of years since 2011 that have exhibited somewhat mixed pre-season signals, with lesser degrees of confidence in the potency of the hurricane season. However, I believe the evolution of some early warning signs during the last several weeks have pushed 2013 farther into the realm of certainty, at least with regards to its overall level of activity. Specifically, 2013 is likely to be an above-average hurricane season.

As usual, we will begin with the current sea surface temperature (SST) anomaly map of the globe, which is always the best place to start, as great deal of information can be extracted from it.

Figure 1. SST anomalies for March 14th, 2013.

In the Pacific, little difference seems to exist from last year at this time. The El Nino Southern Oscillation (ENSO) is in its neutral phase, neither El Nino nor La Nina, evident by near-normal SSTs across the equatorial Pacific. Cooler than normal water extends from Hawaii up the western coast of North America, indicating a negative Pacific Decadal Oscillation (PDO). The negative PDO tends to feature high pressure over most of the north Pacific, which during the summer helps focus low-level convergence (piling) up of air in the tropical Atlantic, which helps the air to rise and form thunderstorms. The neutral ENSO is not usually detrimental to Atlantic tropical activity, but the biggest seasons usually feature a La Nina event. The tendency or “bias” of the ENSO state this summer towards warm or cool will likely determine its effect on the Atlantic.

While the state of the Pacific is similar to last year, the Atlantic is not. Last year, mostly colder than normal water existed in the deep tropical Atlantic, with warmer than normal water in the mid-latitudes. This year, the opposite is true, with the main development region (MDR) currently running about 0.5° above normal, and cooler water appearing in the mid-latitude belt, with warm water appearing again in the arctic waters. This pattern is known as a positive Atlantic tripole, and was less defined before February. However, since February, powerfully negative phase of the North Atlantic Oscillation (NAO) has developed, which features blocking high pressure over Greenland and low pressure over the mid-latitude Atlantic, which reduces the strength of the trade winds over the tropics, reducing evaporative cooling and allowing SSTs to warm. This pattern has coincided with the turn to cold weather across the eastern United States, which had previously been warmer. Such a turn in the winter pattern over North America is very relevant to the upcoming hurricane season, because if it persists for several weeks at the end of the winter and through March (which it is forecasted to do), the alterations it makes to the Atlantic SST profile can last into the summer and enhance tropical activity. Warm water in the tropics supports convection, and cold water to the north of the tropics promotes sinking air there, which upon reaching the ocean surface spreads out back to the south and increases convergence in the tropics, enhancing convection. If this SST profile persists through the spring, coupled with a negative PDO, then this alone is almost enough to guarantee an above-average hurricane season.

Figure 2. MSLP anomalies from February 1 through March 14, 2013.

Analog Package

The set of analog years I picked for 2013 were based on a simple algorithm. I looked for neutral ENSO hurricane seasons since 1950 which featured a late winter and spring Atlantic SST profile similar to this year. These years were 1952, 1966, 1996, and 2005. Figure 3 illustrates the similarity between the SST profiles of this year and these analog years.

Figure 3. Atlantic SST anomalies during March-April for the analog package (left) and SST anomalies for March 1-14, 2013 (right).

Interestingly enough, it turns out that the evolution of the winter pattern over North America during the analog years is similar to this year (see figure 4). Notice how the winters started off warm over most of the U.S. as this year did, but turned quite cold in the late winter and early spring as we are seeing right now this year. This illustrates the tendency of the pattern, moving towards a new state that is shaping the upcoming summer.

This analog set may seem strange in that it contains both 1952, a year with officially only 6 storms, and 2005, a year with 28 storms. Firstly, 1952 is curiously blank in the eastern Atlantic, and likely had some more storms that went undetected back then. Secondly, all of 1952’s storms were hurricanes except one, and 3 of them were majors. Strong hurricanes permeate all of the analog years. The most striking aspect of the analog package is the concentration of high ACE storms in the SW Atlantic, which we will talk about more later. My general feeling is that 2013 will be bigger than 1952 and 1966, probably a bit bigger than 1996, but of course not close to the level of 2005.

Figure 4. Surface temperature anomalies for analogs Dec-Jan (top left), analogs Feb-Mar (top right), 2013 Dec-Jan (bottom left), and 2013 Feb-Mar 14 (bottom right).

Model Forecasts

Having the ENSO in a neutral state during the spring is rather awkward, as this time of year is a transition period in the ENSO cycle during which the models perform very poorly in predicting its evolution during the coming summer. Perhaps the best model we have is the ECMWF, which forecasts continuing neutral conditions if we take the ensemble mean, but with a slight warming tendency throughout the next several months. ENSO may or may not be a wildcard, as its behavior is too unpredictable in a state like this. My current feeling is that we will see a near-neutral summer, and regardless of whether it is warm or cool biased, the atmosphere has been behaving more like La Nina than El Nino, and thus it will not significantly impact the Atlantic. The important thing is that heat is focused in the tropical Atlantic, such that it is a more favorable place for upward motion than the tropical Pacific. As long as the tropical Atlantic stays warmer, it should win that battle.

Figure 5. ECMWF forecasted Nino 3.4 SST Anomalies (initialized in March 2013).

The ECMWF is also usually fairly skilled at depicting the MSLP anomaly pattern in the tropics months in advance of the summer. Figure 6 shows last month’s forecast, which called for higher than normal pressures across the tropical Atlantic. However, this month’s forecast has changed significantly, and now calls for near-normal pressures. It is likely that the actual pattern will not be as “normal” as the ECMWF suggests, but I believe the recent trend in its forecast is partly a result of the currently changing northern hemisphere pattern, something that the model wasn’t catching before.

Figure 6. ECMWF February (top) and March (bottom) percentile forecasts for June-July-August average MSLP.

The National Multi-model Ensemble (NMME), an average of North American seasonal model forecasts, indicates a wetter than normal intertropical convergence zone (ITCZ) in the eastern Atlantic during the first half of the season, a sign of an active pattern and healthy tropical waves from Africa.

Figure 7. NMME precipitation anomaly forecast for June-July-August.

My Forecast for Overall Activity

My general philosophy is that we will see a season more active than last year in terms of ACE per storm. While last year technically had 19 named storms, many of them were short-lived and weak, forming north of the tropics, and it was not terribly above average in terms of ACE. Hurricane Sandy overshadows the memory of 2012, but it was not a huge hurricane season compared with history. In that light, I expect the deep tropics to be more active this year due to the SST pattern, and with model support for at least marginally favorable atmospheric conditions, I expect an above-normal season with 14-16 named storms. However, you never know what the final count will be these days. More and more “wimpy” storms have been acquiring names in recent years. Of these storms, I expect 8-10 to become hurricanes, with 3-5 major hurricanes (category 3 or higher). The median of these numbers is 15-9-4.

Where will these storms go?

Forecasting overall season activity is all well and good, but what really matters is where these storms will track and which land areas they will threaten. This is a much more difficult job to do with significant accuracy. I show in the video at the beginning of this post that it is hard to find a forecasting method that works 6 months in advance of the peak of the hurricane season, which is where we are now. However, if we have a negative PDO like we do this year, an aspect of the pattern in the Pacific may be a useful tool for determining whether the pattern will tend to steer storms away from the eastern seaboard or closer to the U.S. and Caribbean. This is shown in the video.

Another potentially viable method is examining our analog package. Here are the track maps:

Figure 8.Analog year hurricane track maps. Courtesy of Wunderground.

Again, 1952 was not a particularly explosive year in terms of numbers, but what is striking is the similarity between 1952, 1966, and 1996, with several strong hurricanes passing through the SW Atlantic and/or northern Caribbean. The track clusters are very similar, and all focused to the western side of the Atlantic. 2005 obviously blows up and has storms everywhere you look, but the largest concentration of power (strong hurricane track segments) in that year is also to the west, in the Caribbean and Gulf of Mexico. This tendency in the analog years indicates the effects of a negative PDO pattern (or trending negative) coupled with a positive SST tripole pattern in the Atlantic, which allows storms to form in the deep tropics and track farther westward before recurving.

More time is still needed to observe the springtime pattern for signs of what the summer steering pattern over North America and the Atlantic may be like, but early signs indicate a healthy negative PDO pattern that is perhaps a little stronger than what we had in 2010-2012. Coupled with active deep tropics, this implies storms that originate in the MDR and recurve in the western Atlantic, threatening the eastern seaboard as they have in the last couple years, along with the Bahamas, Florida, and eastern Caribbean. There is a risk that the eastern U.S. will be threatened more often than in 2011 and 2012, which each had one east coast hurricane, as the pattern may favor an extra nudge to the west in the track spread.


Overall, I expect 2013 to be an above normal hurricane season, and specifically more active than last year in terms of ACE per storm. Fewer total storms may develop than last year, but more of them will be stronger. I expect 14-16 named storms, 8-10 hurricanes, and 3-5 major hurricanes. There may be 2-3 hurricane landfalls on U.S. soil, and one or two of these could be major. The 1950-2012 average for the United States is 1.5 hurricane landfalls and 0.6 major landfalls per season. It is worth noting that the U.S. is still in the midst of a drought in major hurricane landfalls that is now the longest in observed history. Hurricane Wilma in 2005 was the last major landfall. As always, be prepared, as the atmosphere is inherently a chaotic system that can acquire almost any configuration at any time, and could direct a storm at you during any week of the summer and fall, no matter what year it is or how active the season is as a whole. Be safe during 2013.

We shall see what happens!




  • StormW says:

    Very stellar analysis… great minds……

  • Papi says:

    Great analysis. Many thanks for sharing.

  • Eric Webb says:

    Thanks so much Levi, very good analysis and video. Now, with the years you’ve mentioned which dealt with the trough and ridging pattern over the northern Pacific and North America as a whole (which would tell me to pay attention to the PNA) in which their was an increase in the number of hurricane landfalls on the US, I have taken this a step further in that, I compared those years to the sunspot cycles. I did this because I have noticed a very interesting inverse correlation relationship to the PNA & the sunspot cycles, even over short scales of time, and have found the 30 day sunspot cycle to be very useful in predicting the overall state of the PNA. When I looked at the years mentioned like 1953, 1954, 1955, 1959, 1964, 1971, 1979, 1985, 1989, 2004, 2005, and 2008 I noticed something very interesting about all of those years, in exception to 1979. In that, those years occurred near the peaks and the minimums of the normal 11-year solar cycle (22 years accounting for magnetic field changes), which is extremely interesting and may in fact hold the key to determining this pattern over the north Pacific and North America as you were talking about, and considering that this year we are currently near the peak of the solar cycle suggests to me that this hurricane season may very well follow suit of those mentioned years given also the overall state of the PDO, AMO, and the fact that the US, especially the east coast is overdue for a major hurricane. I am actually somewhat surprised by the lack of activity, probably as you are in the aspect that this pattern is very similar to the 1950s, yet there have yet to be any major hurricanes on the US east coast, while in the 1950s, there 8 major hurricane on the east coast, with the other major storm being Hurricane Audrey of 1957 in the gulf.

  • Eric Webb says:

    This sunspot cycle theory also works in the 1930s, with the solar minimum between solar cycles 16 & 17 having 1933 in the middle, and of course in that season 3 major hurricane struck the US coast. 1934 & 1935 were also both years near that same minimum, and the barrage of hurricanes continued on the US coast including Labor Day in 1935. The solar peak of cycle 17 around 1938 featured the Long Island Express and Hurricane 3 hitting the US coast. It does not stop there, the next region of solar maximum and minimum was towards 1944, which had yet another barrage of hurricanes on the US coast, with 3 hurricanes hitting the US in that season, along with the Great Atlantic hurricane of 1944. It only continues from here with the next solar peak or minimum towards the period of 1947-1950 and during that period of the cooling (not yet cold) PDO and warm AMO, reminiscent of the hurricane landfall pattern of the 2000s, another batch of intense hurricanes struck the US coast, especially towards Florida and into the Gulf of Mexico, with 5 major hurricanes in just 3 years striking the US coast, including Hurricanes King and Easy in 1950. Then, by 1953-55 at the next solar peak and minimum, there was yet another period of extreme landfall activity on the US coast, with storms like Connie, Diane, Ione, Hazel, Carol, and Edna. Just extremely interesting to see this correlation seeming to show itself in just about every instance going back to the 1930s. If it wasn’t for you talking about the pattern over the north Pacific getting me into the mindset of the PNA, would have never thought of this. Extremely hard to ignore such a glaring trend in hurricane activity on the US coast in relation to the peaks and minimums of the solar cycle.

  • Levi says:

    Thanks all for stopping by :)

    Eric – correlations need a physical explanation, and there doesn’t seem to be a reason why both solar maximums and minimums would cause the same phenomenon. Furthermore, the temporal distance between local extrema is less than 6 years, which, if you allow a couple years of fudge factor around each min and max, takes into account almost ALL years anyway. At that point, almost any year you pick would qualify for your claimed relationship.

  • Eric Webb says:

    @ Levi Cowan
    Very true indeed as I did note there are certainly exceptions to this during the 1910s & 1920s, but objectively looking at those years post 1950, the vast majority of them at the peaks and minimums of the cycles were active (not say there weren’t some active years here or there in between). Yet, considering such relationships like this are primitive and we still are probably a good ways away from fully explaining such phenomena (among other solar relationships), it is interesting to observe, although you are right in that it does take into account a lot of years, but the peaks and min years specifically going back to about 1950 with solar cycle 18 triple peaked in 1947, 1948, and 1949, the next minimum in 1954, the next maximum in 1969, solar cycle 22’s double peak in 1989 & 1991, the following minimum in 1996 (one of the analog years mentioned) Next peak 2000 & 2002 (obviously not a whole lot happened there), and the next minimum towards 2008. Thus, there obviously is no direct correlation or perfect factor playing into hurricane seasons, as it is pointless in doing so as there are so many other variables that have to be taken into account. However, in science and in weather forecasting not every oscillation nor predictor used will be perfect, thus it is important to look at the overall trend in things. There certainly is something to be said about this, and knowing there is a relationship between the PNA and the 30 day sunspot cycle, this certainly has my attention to some degree and it will be interesting to see if this does come to pass in one of the next few years where a supposed “secondary solar cycle peak” according to NASA is going to occur, but only time will tell.

  • Oracle_De_Atlantis says:

    It makes my head hurt sometimes, but I appreciate your efforts.

  • Eric Webb says:

    I know you may not believe or completely understand what I was trying to get at with the sunspot cycle-hurricane landfall influence, but I have an answer for you as far as physical drivers go and their effects on the northern hemisphere pressure pattern (particularly over the north Pacific and northwestern North America over the source region of the PDO) with my thoughts on the upcoming hurricane season, I think you’ll really like it, just like I absolutely loved your analysis on thoughts above on the upcoming hurricane season.

  • […] And for a Hurricane season 2013 outlook by Levi go to… […]

  • Andre Brooks says:

    This is going to be an active year.And thank you Tropical Tidbits for the information.I’m expecting 16-20 storms,10-14 Hurricanes,and 6-10 major hurricanes.

  • BahaHurican says:

    Great job, as usual, Levi… Still trying to figure out how u r going to work 4 NHC and stay in AK… :o)

    Looking at your analog years, none were horrible for the Bahamas, which leaves room for hope. Still, I’m watching to see if the positive dipole pattern u mention persists.

  • Tucker says:

    Great job. I really enjoy reading your blog and watching your video’s you seem to really care about what you do. I have learned a great deal about storms through your post, do you have any idea when your next update will be. I just read what CSU forecast are seems your in agreement, they used 5 different years and you used 4 to make your forecast. I hope your doing well in school and once again great job on the forecasting.

  • Boyroy says:

    Excellent job Levi!! Thanks for all you do and keep up the good work. It’s great to know that I can come here to get your informative view on the tropics and the upcoming hurricane season.

    Thanks again,

  • mchammer says:

    BALLER analysis Levi. Keep on doing great work, I appreciate the time you put into your work and the product you put out. Excellent.

  • StormW says:

    Levi, outstanding! Can you tell me, where did you get the MJO maps starting at minute 2:00?

  • Leave a Reply to Eric Webb Cancel reply

    Your email address will not be published.

    Basic HTML is allowed.