After an interestingly successful analog forecast for Hurricane Season 2014, I will post a similar forecast for Hurricane Season 2015 where I try to use the same methods. Last March, I forecast that “tropical cyclone activity should be near or below normal”. The two sets of analogs used, ‘warm winter’ and ‘wet winter’, and an average of the hurricane activity for those two sets of analogs were provided:
The analog years for that forecast included Hurricane Arlene (1963), Hurricane Faye (1975), and Hurricane Dean (1989) that had impacts on Bermuda. They accounted for 4/14 analog years as 1989 appeared in both the ‘warm’ set and the ‘wet’. While it is impossible to say whether this method of seasonal forecasting has any skill from just the one hurricane season’s forecast, I am compelled to try again this year to see if we get similar levels of accuracy.
The mean air temperature in Bermuda for January and February was 65.1°F (18.4°C) which is near normal as January was slightly warmer than normal and February slightly cooler. The total precipitation, however, was much greater than normal with 13.20″ accumulating for the two month, most of which fell in February which was the second wettest February on record at the Bermuda Weather Service, and one of three Februaries with more than 9″ of rain since 1949.
Continuing the method from last year, 7 years with the closest mean Jan-Feb temperature and 7 with the closest precipitation total will constitute the two analogs. Here they will be known as the “Temperature” analog and the “Precipitation” analog, rather than “warm” and “wet”.
Temperature Analog Set (Top): 1976, 1957, 1986, 1995, 2012, 1983, 2008.
Precipitation Analog Set (Bottom): 1972, 1977, 1969, 1993, 1998, 1973, 1963.
Interestingly, neither set of analogs has a similar 500mb height anomaly pattern for January and February to 2015. The temperature analog captures the idea of higher heights over the East-Central North Atlantic, and lower heights to the north of that, over Greenland. However, it does not indicate lower than normal heights over Eastern North America or even come close to the magnitude of the anomalies anywhere. Further, the 500mb height anomaly pattern in the precipitation analog set is almost opposite that for 2015 with lower than normal heights across much of the North Atlantic and higher than normal heights over Greenland. This shows that these analogs are not necessarily representative of the prevailing atmospheric set up across the North Atlantic, which is not all that unexpected as the analog years are chosen based on conditions at one point. I expect this forecast to demonstrate less skill than last year’s as a result.
Using the same reasoning from last year’s post, I will look at the hurricane season in two parts, the first half of the season June-July-August, and the second half August-September-October.
Looking at the first half of Hurricane season and mean sea-level pressure anomalies for the Temperature Analog (Top) and Precipitation Analog (Bottom), we see that common to both analogs are lower than normal pressures across the Caribbean, Southwest Atlantic, and West Africa. The big glaring difference, however, lies near the Azores where in the Temperature Analogs, there is much lower than normal pressure and in the precipitation analog there is much higher than normal pressure. This may be a North Atlantic Oscillation (NAO) signal; when the NAO index is negative the Bermuda-Azores high is weaker than normal, and the opposite is true generally speaking. The sign of the NAO mainly impacts the trade winds in the Central Atlantic (negative NAO: weaker trade winds, positive NAO: stronger trade winds). According to Climatology, most of the tropical cyclone genesis locations are in the Gulf of Mexico, Caribbean, and Southwest Atlantic for the first half of the season, not in the Central Atlantic where trade winds become a factor.
[What is the North Atlantic Oscillation?]
In Summary: the lower than normal pressures in the Southwest Atlantic (including the Gulf of Mexico and the Caribbean) in both analog sets suggest that there will be greater than normal chances for tropical cyclone activity in that region. Lower than normal pressures over West Africa suggest that an active monsoon pattern will set up over than region and allow for enhanced disturbance activity in the Central and Eastern Tropical Atlantic, particularly in the Precipitation Analogs. Should the NAO be in a persistent negative phase, enhanced activity could extend eastward into the Central Atlantic with weaker than normal trade winds allowing disturbances to organize. This could result in an early start to the peak part of Hurricane season. Conversely, should the NAO be persistently positive during this period, stronger than normal trade winds across the Central Atlantic could result in a later start to the peak part of the season.
Again, Temperature Analogs (Top), Precipitation Analogs (Bottom). Looking ahead to the second half of the Atlantic Hurricane Season, including the peak in activity (climatologically around September 10th). Much of the pressure anomaly in the Western Atlantic vanishes. Near normal pressures across the Western Atlantic, Caribbean and Gulf of Mexico in both sets of Analogs. In the Precipitation Analog, we see lower than normal pressures over Central America indicating a more active Central American Monsoon pattern. Monsoon circulations from Central America are known to spin up into tropical cyclones over the Western Caribbean and Gulf of Mexico if given enough time over water. However, monsoon circulations could limit total numbers of named storms by taking so long to develop and absorbing approaching disturbances before they develop. Further, we see this persistent signal in pressures near the Azores. The Temperature analogs continue to indicate lower than normal pressures here while the Precipitation analogs suggest higher than normal pressures.
In Summary: Near normal chances for tropical cyclone activity seems likely across the Southwest Atlantic, and while they may be enhanced by an active monsoon pattern (as indicated in the Precipitation Analogs), generally, near normal chances for tropical cyclone activity extends into the Gulf of Mexico and Caribbean. The West African Monsoon continues more active than normal as indicated by lower than normal pressures there in both analogs. However, the same question of the Bermuda-Azores high and NAO phase arises as in the first part of the season.
Overall: Expect a Near ‘Normal’ Season.
Slightly above normal activity seems to be the theme in the Southwest Atlantic for the early part of the season, followed by near normal activity there during the second part (main phase) of the season with eyes on what the Central American Monsoon does. Between the two analogs, one pushing for stronger than normal trade winds, and the other for weaker than normal trade winds, it seems fair to expect near normal trade winds. That combined with a more active than normal West African monsoon suggests a slightly more active than normal Central Atlantic for the entire season. The big questions are: How strong will the Bermuda-Azores high be? and How strong will the trade winds be as a result?. With an apparently more active than normal West African Monsoon, there should be plenty of disturbances moving from east to west into the main development region of the Central Tropical Atlantic and the frequency with which they develop could be limited or enhanced depending on the strength of the trade winds. A weaker than normal Bermuda-Azores high might also allow cyclones that form in the Central Atlantic to recurve further out to sea rather than tracking into the Caribbean or West Atlantic where they could impact land.
According to 1981-2010 climatology, an average season has 12 names storms, 6 hurricanes, and 3 major hurricanes, with a season total ACE ear 93.5.
(1976, 1957, 1986, 1995, 2012, 1983, 2008)
(1972, 1977, 1969, 1993, 1998, 1973, 1963)
These analog forecasts therefore show a near normal Hurricane Season for 2015. It is important to note that little can be said about the exact track of individual storms this far in advance, and it is important to remember that this forecast is based off of analogs from one spot (Bermuda) and therefore might not represent conditions across the Atlantic as a whole. This forecast also does not account for some important physical conditions that drive/impact tropical cyclone activity such as regional Atlantic sea surface temperature or ocean heat content and the impacts from whatever phase of the El Nino Southern Oscillation we are in during the height of the season.
Again, as mentioned at the beginning, these analog years were chosen because similar temperature and precipitation statistics were observed in Bermuda in Jan-Feb. However, neither set of analogs produced similar patterns of 500mb height anomalies for Jan-Feb. So this analog forecast is likely low confidence at best. Further, there were huge differences in activity in the years within both analogs, including some of the top ten most active seasons and the least active seasons. The commonality was they both featured seasons with hurricane impacts on Bermuda.
Residents in coastal areas should take the time before the season to prepare as regardless of the number of tropical cyclones that form. It only takes one cyclone to make for a bad season. It is therefore important to follow products on individual storms from the National Hurricane Center and from a local meteorological service such as the Bermuda Weather Service.