It's only February, and it's still the middle of winter, but already we can see signs within the developing weather pattern that can give us clues about what may be in store for us during this year's Atlantic Hurricane Season. So far these signs have been rather disturbing, pointing towards a likely active, above-average season, and possibly a year of many landfalls for countries bordering the western Atlantic. Let's take a look at some of the things that have been going on this winter.
The brutally cold pattern this winter over the south and eastern U.S. has cooled waters in the Gulf of Mexico and off the SE U.S. coast down to 2C below normal. This cold water stretches out across the mid-latitudes of the north Atlantic, due to the negative NAO which has dominated this winter, causing lots of blocking over the North Pole and intrusion of arctic air into the mid-latitudes of the Atlantic. The negative NAO has also resulted in a weaker Azores High, and consequently weaker trade winds over the eastern Atlantic, which has allowed SSTs to warm over a large area of the tropical Atlantic from the African coast to the central Caribbean.
If this pattern continues into the hurricane season, it will be very good for tropical waves coming off of Africa, not just because of the warm ocean water caused by the weak trades, but because of the trade winds themselves. Strong trade winds greatly limit surface convergence (piling up of air) and don't allow tropical waves to amplify and strengthen. This is easier to understand if you think about air molecules as being tiny cars. Cars traveling at high speeds on a highway can't turn very fast, and have to make large, gradual, sweeping curves. The same principle applies to air molecules rounding the axis of a tropical wave. When air is moving fast with the trade winds, it is forced to take a gradual curve around the wave axis, which causes air molecules (traffic) to be less conjested, and the tropical wave can't amplify. Weaker trade winds give air molecules the freedom to take a slower and sharper turn around the wave axis, causing more conjestion (convergence) of the air at the beginning of the curve, which results in more upward motion, and in turn, lower surface air pressure, which allows the tropical wave to amplify and strengthen.
Figure 2. Depiction of a typical tropical wave, showing the area of surface convergence that occurs east of the axis, and the surface divergence that occurs to the west of the axis.
The current SST profile of warm anomalies in the tropical Atlantic and Caribbean with cool to the north is forecasted by the ECMWF to continue into the beginning of the hurricane season, with the GOM and SE U.S. coastal waters forecast to catch up from their current below-normal temperatures and become warmer than normal by the start of the season. This pattern is very favorable for an active season with the heat concentrated in the tropical breeding grounds of the Atlantic, which is the complete opposite of last year when we had cold water in the tropics and warm water over the top. This is one of the most lucrative patterns you can get for a high number of named storms. This can be puzzling, since one would think that having the entire ocean warmer than normal would yield the best chance for more storms. The reason this is not true is because having all the heat spread out over the entire Atlantic is not necessarily the best way to go. We know that warmer-than-normal water promotes net upward motion, and colder-than-normal water promotes net subsidence (sinking). When you get a belt of cold water across the mid-Atlantic, it forces air south and concentrates heat over the deep tropics, promoting convergence and net upward motion. With a dying El Nino, pressures will be on the rise over the eastern Pacific, which also drives air across Central America to converge over the Caribbean and the tropical Atlantic. Precipitation will obviously increase over this area as well, as more convergence = more convection. Unfortunately this also sets up a corridor of low pressure directing storms toward the United States. The image below shows the ECMWF forecasted SST, surface pressure, and precipitation anomalies for the period of June-July-August, which reflect this scenario.
Figure 3. European model forecasted SST, MSLP, and precipitation anomalies for the period of June-July-August, showing the concentrated area of convergence and upward motion that sets up over the tropical Atlantic during a dying El Nino and a positive Atlantic Dipole SST pattern. Model background image courtesy of ECMWF
As we all know, models screw up all the time, and going on just a computer forecast is a bad way to predict the weather. That's why we look at history to try to find where similar conditions have happened before, and use the past to help predict the future. This kind of forecasting usually takes the form of what we call analog years, years where conditions were similar to what they are now. This is one of the most useful tools for long-range and climate forecasting.
This year I have come up with a set of 9 analog years for the upcoming hurricane season, mostly based on water temperature profiles and the ENSO, since some of the other climate factors don't come into play much until May, when most of the forecasting for the hurricane season is done. The set is comprised of the years 1958, 1964, 1966, 1970, 1978, 1995, 1998, 2005, and 2007. The table below summarizes the ENSO indices for each year as well as the previous year in order to cover the entire winter and get a feel for the general trend.
The common demoninator of these years is an El Nino peaking around December, and then declining through the latter half of the winter, eventually becoming neutral or a weak La Nina during the hurricane season. Global satellite SST anomaly maps only go back to 1996, but looking at what was going on at this time of year in 1998, 2005, and 2007, you can see the same kind of thing that's happening this year: a dying El Nino in the Pacific, warm water developing over the tropical Atlantic, and cold water to the north over the mid-Atlantic, Gulf of Mexico, and off the SE U.S. coast, with warm water in the far north Atlantic. This same Atlantic Tripole SST pattern is evident in the NCEP/NCAR SST reanalysis data from July through October of the analog year set, suggesting that the setup in February does indeed last through the Summer:
Figure 5. Composite reanalysis sea-surface temperature departures from normal for the months July-October in 1958, 1964, 1966, 1970, 1978, 1995, 1998, 2005, and 2007. Image courtesy of NOAA ESRL
February sea-level pressures from these years also show strong similarities to this year with negative anomalies over the mid to north Atlantic, meaning weaker trade winds, which supports the SST idea above:
Figure 6. Composite sea-level air pressure departures from normal for the month of February in 1958, 1964, 1966, 1970, 1978, 1995, 1998, 2005, and 2007. Image courtesy of NOAA ESRL
I guess a look at the actual hurricane seasons of these years would help a little too ;)
Figure 7. Composite of all tropical storm and hurricane tracks for 1958, 1964, 1966, 1970, 1978, 1995, 1998, 2005, and 2007. Areas of greatest track concentration are circled in blue. Image courtesy of the NOAA Coastal Services Center
Storm tracks from these years show a nice, healthy concentration of tracks in the MDR (Main Development Region) over the tropical Atlantic, which one would expect based on the SST patterns and concentration of heat over the tropics that these years had. The other most concentrated track sprays occur in the western Caribbean up through the Gulf of Mexico, and just off the SE U.S. coast. This says a lot about what the NAO and position of the Azores/Bermuda High was doing during most of these years, directing many storms westward towards the United States and Caribbean countries. 6 out of the 9 analog years had a predominantly negative NAO during the preceding winter, and 5 out of those 6 also had negative to neutral NAOs during the peak months of the hurricane season, August and September. Here are some interesting statistics from the entire set of 9 analog years. Out of a total of 131 named storms, 85 either formed or tracked west of 70W. 35 of these made landfall in the U.S., but nearly all 85 either made landfall or had a strong impact on at least one country. The average number of named storms per year was 15 (13 if you discount 2005). The average number of named storm U.S. landfalls per year was 4.
Do preceding winters have an effect on the hurricane season?
The answer is yes and no. Sometimes preceding winters correlate nicely, but they don't always cooperate. In fact, the correlation of the hurricane season's effect on the following winter is much stronger than the preceding winter's effect on the following hurricane season. It is my theory that when looking at analogs, the correlation of a certain winter's effect on the following hurricane season is much higher if you take years in the same phases of certain global climate cycles, such as the Pacific Decadal Oscillation (PDO). There is actually some really interesting stuff regarding the winters preceding our analog years for this hurricane season. Some of the years in the set were already analogs for the 2009/2010 winter, including some really big hitters like the winters of 1957-58, 1963-64, and 1977-78, all historically cold and snowy for the southeast United States. Sound familiar? In fact, when you take the first half of the analog set, 1958, 1964, 1966, 1970, and 1978, and look at the composite temperature anomaly of December to February compared to this winter so far, this is what you get:
Figure 8. Composite surface air temperature departure from normal (left) for the months December-February of the winters of 1957-58, 1963-64, 1965-66, 1969-70, and 1977-78, compared to the 2009-2010 winter so far (December 1st through February 21st, right-hand side). Images courtesy of NOAA ESRL
How close is that? If you could see further to the north on the left-hand image you'd even see the massive warmth over Canada just like this winter. It's amazing how similar these winters are, and they are all hurricane analogs for this year as well. So what did these 5 years have in common? They were all within the cold cycle of the PDO which began in the late 1940s and lasted through the late 1970s. Keep in mind that all of our hurricane analogs had El Ninos the previous winter. Now watch what happens when we take the winters of the 4 years that we left out, 1995, 1998, 2005, and 2007, and look at the composite temperature anomaly:
Figure 9. Composite surface air temperature departure from normal for the months December-February for the winters of 1994-95, 1997-98, 2004-05, and 2006-07. Image courtesy of NOAA ESRL
And wallah, look at how warm the United States is. Why is this latter half of the analog set so different from the 1st half in terms of the preceding winters? Even though all 9 winters were in an El Nino? Because these 4 years were within the WARM cycle of the PDO which began after 1978 and lasted through the late 2000s. We are now on the way back down into the next cold cycle of the PDO, and lo and behold, as we saw above, this winter so far has been extraordinarily similar to this year's hurricane analogs from the last cold PDO! Amazing isn't it? It sure would seem that there is at least some sort of connection between hurricane seasons and their preceding winters, as long as you are looking at years that are within the same cycle of the PDO. It is certainly worth further investigation. This could also open up a whole other discussion on what a different kind of beast El Nino is when it is within a cold PDO. Remember all 9 analog years were El Nino winters, and yet half of them were exceedingly cold for the U.S., and the other half were exceedingly warm. But that's another blog for another time ;)
So....what can we determine from all this about the upcoming hurricane season?
Well I think I've showed pretty well how SST patterns in the Atlantic this year will focus heat in the deep tropical breeding grounds, favoring an above-average year with a strong Cape Verde season and several classic long-track storms. With the strong negative NAO this winter, the steering pattern this summer is likely to favor a more south and west concentration of tracks, aimed closer to the U.S. and Caribbean, but we will need to watch what surface pressures in the Atlantic do in May to see if this changes. I think the entire U.S. coastline is at high risk for a major hurricane landfall this year, and there could very well be multiple majors threatening close to home. The Caribbean is also at higher-than-normal risk of being affected by a major hurricane. This should be a total opposite of last year's hurricane season, when the Caribbean and deep tropics were a burial ground for any storms, and heat focused to the north resulted in a below-average year. This year will definitely be far more dangerous for everybody.
Overall....with the world coming down off of one of the strongest El Ninos observed, the tropics are going to be left with a surplus of heat and moisture, making them primed for an active year world-wide, possibly with the exception of the eastern Pacific. The Atlantic hurricane season will likely be well above-average, and possibly a bad landfall year for the U.S. and the Caribbean. My preliminary forecast is for 14 named storms, 8 of those hurricanes, and 4 of those major hurricanes. I do think at least one major hurricane will make landfall on the U.S. coastline. Of course it's still quite early, but things are already starting to line up in the atmosphere. I hope all of you will take the necessary precautions and prepare well for this hurricane season, as you should every single year. Be safe, and be smart.