My forecast for the 2018 Atlantic hurricane season will be presented here, including detailed analyses as to why I am predicting a specific level of activity or landfall risk. It should be emphasized that these predictions are not exact, and in some years they will inevitably fail.
a. Forecast methodology
1. Sea surface temperature
The first and most important parameter for hurricane season forecasting is the sea surface temperature anomaly (SSTA). This parameter is important because, although tropical cyclones can and do form over water temperatures that are cooler-than-average, the general tendency will be for them to develop over areas where SSTA values are positive (i.e. over 0.0C). Also, positive SSTAs indicate an unstable atmosphere, and hence corresponds also to negative Sea Level Pressure (SLP) anomalies, which indicates upward motion, the initiation of deep convection, and subsequent potential for tropical cyclone development. Conversely, negative SSTAs suggest above-normal SLPAs, and thus suppressed convection and truncated tropical cyclone potential.
Figure 1. Global SSTAs as of May 10, 2018. Image credit: NOAA NESDIS SST analyses.
As can be seen in Figure 1, negative SSTAs have developed in the Main Development Region of the Atlantic, with only a small area of near-normal SSTAs. According to Phil Klotzbach and the now deceased William Gray, there is a strong correlation between Springtime SSTAs persisting into the heart of the Atlantic hurricane season, the months of August, September, and October. Therefore, this negative pool isn’t something that can be so easily ignored. That being said, climatology is not an irrelevant tool, and it strongly suggests that the this area will warm by the time of arrival of the peak season. But even if it does, considering the correlation above and the fact that SSTs have cooled considerably during the last month suggests that it will be hard to completely reverse this situation. It is also possible that persistent continental baroclinicity during the month of April played a role in cooling this area by increasing the surface high pressure belt over the central Atlantic, i.e. the Bermuda High. It will be interesting to see if warmer than normal SSTs return in abundance at the end of this month, since the persistent cold air continental trough appears to have dissipated now. If it does so, actual activity may be higher than I am currently predicting.
2. El Nino Southern Oscillation (ENSO)
Another important parameter for tropical cyclone forecasting is the background state of the ENSO. This is composed of the SSTA values surrounding the equatorial Pacific Ocean. When these values are positive for long periods of time, El Nino events can develop, whereas La Nina events can develop when these values are persistently negative. These events are significant because they can enhance or suppress hurricane activity in the Atlantic. La Nina creates cold water and sinking air over the Pacific, which produces a compensating area of upward motion over the tropical Atlantic. This pattern leads to enhanced convection and weaker vertical wind shear, which generally favors active hurricane seasons; El Nino is the opposite, and greatly suppresses Atlantic hurricane activity. As can be seen in Figure 1, the La Nina event of 2017-18 has completely faded, due to a persistent subsurface warm pool that has been steadily advancing eastward from the International Dateline (Figure 2).
Figure 2. Progression of the subsurface equatorial warm pool from February to present. Notice that there are no longer any significant cold anomalies beneath the surface, a telltale sign that La Nina has died. Image credit: NOAA’s Climate Prediction Center (CPC).
With this in mind, La Nina is unlikely to return this Summer, so the question now remains whether or not the current warming portends an El Nino event. Although positive (westerly) trade wind anomalies exist east of the Dateline (Figure 3), these winds are only weakly positive, and have so far been unable to trigger the formation of a full-fledged El Nino event.
Figure 3. Equatorial Pacific trade wind anomalies, five-day mean ending on May 6, 2018. Image credit: Australian Bureau of Meteorology (BoM).
Given the above and climatology favoring somewhat less than half of all El Nino’s since 1950 to not develop during the Summer, I am not currently expecting an El Nino event to develop and suppress hurricane activity in the Atlantic during the Summer. That being said, I do anticipate that the ENSO background state will remain on the warm side of neutral, and the development of an actual El Nino event during the late Fall remains plausible. As of mid-April, there is still some disagreement among the ENSO models as to whether we get El Nino this Summer, and if we do, whether it’ll persist into the Fall; also, some of them even suggest that it will fail to develop at all, which is also plausible. My own prediction is roughly between the dynamical average and CPC consol (shaded red and blue lines in the image, respectively). It should be noted that these predictions are typically inaccurate during the Spring months in what’s known as the Spring predictability barrier. Also, while I currently consider it an outlier, it should be noted that the ECMWF has shown some statistical skill in seeing past the barrier, and that model is showing us climbing into El Nino rather quickly.
Figure 4. ENSO model prediction plume as of mid-April 2018. The red line represents the dynamical average, while the green line represents the statistical average. The former are run on actual simulations, while the statistical models merely extrapolate history, hence their name.
3. Analog years and predicted activity for 2018
Another important aspect of hurricane forecasting is the use of analog years. While the atmosphere will never follow these years exactly, they can be a useful guide for a generalized awareness of what could transpire during a given hurricane season. For ease of convenience and also so I don’t skew the average using too many years, I have selected 5 different analogs that I feel best describe the level of hurricane activity we can expect in 2018. In order from greatest to least (defined by the current and anticipated ENSO background state, Atlantic SSTAs, SLPAs, and vertical wind shear anomalies), they are:
1. 1979 (9/5/2)
2. 1990 (14/8/1)
3. 2004 (15/9/6)
4. 2014 (8/6/2)
5. 2012 (19/10/2)
My forecast: 14/7/3, ACE index of 125
The mean activity of these years was 13 named storms, 7.5 hurricanes, and 2.6 intense hurricanes, respectively. My own predicted activity is slightly above this mean, as seen above. Also, as you can see above, I am predicting a slightly-above average ACE index of 125. For the laymen, ACE is short for Accumulated Cyclone Energy, and is sometimes a better indicator of actual activity than numbers alone. The stronger a hurricane and the longer it exists for, the higher the ACE it generates.
Common themes of these years was for a storm cluster near and east of the Lesser Antilles, with a secondary maximum along a subtropical belt from the Gulf of Mexico to the Bahamas. In other words, this would represent a season where hurricane formation in the deep tropics is generally suppressed, where higher values of vertical wind shear are expected to be prevalent. The western Caribbean also saw the existence of a persistent upper-air trough which produced upper-level westerly winds in these years and inhibited the development of deep convection. Although 1979 and 2004 featured significant hurricanes in that area (Frederic, Charley, and Ivan), they were already strong enough to modify their environment enough to avoid dying to the shear; they didn’t actually form there.
4. Landfall hotspots
Based on the above, I think the primary landfall hotspots this year will be the northern Gulf Coast, the Lesser Antilles, including the northeastern islands and Puerto Rico, along with the Bahamas and sections of Florida. Central America and Mexico appear to be at a low risk, with Bermuda at a somewhat higher risk. These predictions are highly uncertain, and residents should prepare equally for a hurricane every year, no matter where they are.
b. Final thoughts
I predict that the 2018 Atlantic hurricane season will be above the long-term average, but near the 1995-2017 average. It is possible that this prediction could be too low if the Atlantic warms more than predicted.
Kori Winstead's Weather Forecasts 