Atlantic Hurricane Season Outlook for August 2017

ABOVE AVERAGE HURRICANE SEASON IN TERMS OF BOTH OVERALL ACTIVITY AND U.S. LANDFALL RISKS EXPECTED…

Outlook Overview

  • With the busiest part of hurricane season just ahead, WeatherTiger’s final seasonal prediction for the 2017 hurricane season continues to be for above average activity in the Atlantic Basin. Solidly neutral El Niño conditions, warmer than average waters in key regions of the Atlantic, and weaker than average trade winds in the Caribbean over the summer all confirm this above-average activity signal. 
  • U.S. landfall risks are also elevated relative to normal, with two or more hurricane landfalls even money. The expectation that the current west Atlantic ridging may persist into September means that steering patterns are likely to be at least intermittently favorable for U.S. threats over the heart of the season.

Previously on the 2017 Hurricane Season…     

The 2017 hurricane season is now just over a third over in calendar terms, though over 85% of overall tropical cyclone activity and over 90% of major hurricane landfalls have historically occurred after the first week of August. Examining at our proprietary dataset of net TC energy expended over the U.S. since 1900 (above), the baseline risk increases steadily through early August, jumping to peak risks in mid-September. So far, 2017 has served up a number tropical aperitifs, with five tropical storms and two U.S. tropical storm landfalls.

While Cindy and Emily were Gulf Coast rainmakers, and the other storms were short-lived (and low-energy) weirdos, a quick start in terms of U.S. landfalls has often preceded more impacts later in the season. Among the 26 years since 1900 in which two or more storms made U.S. landfall before 7/31, an average of an additional 3.0 tropical storm, 1.8 hurricane, and 0.65 major hurricane U.S. landfalls occurred after 8/1— over 150% of climatological risks. Is the recent past prologue for the rest of the year? 

Updated Seasonal Outlook for Tropical Cyclone Activity

Since December, our seasonal outlook for the 2017 has been consistently for above average activity, but this forecast was least certain in the spring as the threat of a summer/fall El Niño loomed. As we discussed in our last outlook, the moderate or strong +ENSO event showing up in many models would have greatly increased the chances of a below average season.

However, our scepticism towards El Niño ’17 was well-placed, as the sea surface temperature anomalies of the equatorial Pacific are neutral, and none of the members (right) of the recent ENSO model ensembles show anything other than neutral or neutral-negative conditions for the remainder of the year. There is no remaining threat of a negative impact on the season from an El Niño.

Given that, there are a number of other factors that influence overall activity. Two important considerations identified by Weather TigerTracks modelling system are global mean-relative sea surface temperatures in a horseshoe across the subtropical and tropical Atlantic, and the strength of low-level Caribbean trade winds in July.

The historical relationship between July rSSTs and TC activity in the Atlantic from August onward is shown at right, with cooler rSSTs in the eastern Pacific (La Niña) and warmer waters in the central and eastern Atlantic Main Development Region (MDR) having the strongest positive correlation with active hurricane seasons. As shown above, this Atlantic rSST tripole was generally warmer than average in July, and continues to be in a favourable configuration in early August.

Relatedly, years in which summer low-level winds across the Caribbean Sea and western MDR, which blow from east-to-west, are weaker than average tend to precede above average hurricane seasons. This is because stronger trades both make it difficult for tropical waves to organize at the surface, and because it is associated with enhanced vertical wind shear. In July of 2017, the tropical easterlies were weaker than normal in the eastern Caribbean and western MDR, and near normal in the western Caribbean. All-in-all, this augers a more active than normal season.

TigerTracks objectively synthesizes these inputs, yielding an optimized and probabilistic forecast for the activity ahead in the 2017 season. The plot above shows the results of a replicated real-time prediction of Accumulated Cyclone Energy (ACE) for the years 1983-2017, with the 50% (red) and 80% (grey) confidence ranges. Skill over climatology during that period in terms of root mean squared error reduction is 35%, without “cheating” by bringing in information from the future. For 2017 (red dot on the plot below), our model is predicting total seasonal ACE of around 140 units, with a 50% chance of between 105 and 180 units. For reference, the median season has ~100 units of ACE, so this is a significantly above normal forecast. In terms of counts [with 50% confidence intervals], look for 17 [15-21] tropical storms, 7 [5-9] hurricanes, and 4 [3-5] major hurricanes total this year.

Updated Seasonal Outlook for U.S. Landfall Risks

Of course, we don’t care about counts per se, we care about impacts. Two very similar years in terms of ACE, like 2004 and 2010, can have wildly divergent outcomes, with 2004 yielding six U.S. landfalls and 2010 none. This is particularly apparent over the last decade, in which the Atlantic has been more active than average but a record low percentage of that activity has occurred over the United States, as seen by the declining red line at left. To put that in perspective, since the last U.S. major hurricane landfall in October 2005, the Spider-man film series has rebooted over forty times (fact). Forecasting of seasonal landfall risks is a tricky proposition, because U.S. landfall activity has surprisingly little to do with overall TC activity—variability in ACE explains only about 30% of the observed variance in U.S. landfalls, with the rest depending on genesis location, intensification, and steering currents. For this reason, some of the factors like Caribbean trade wind strength that do a good job predicting overall season activity have a minimal relationship with U.S. landfall risks, while others like Atlantic rSST have a much weaker correlation with landfall risks than overall activity. However, there are certain summer leading indicators of whether steering patterns will broadly be more or less favourable than average in the core of hurricane season.

One such metric is July sea-level pressure (SLP) departures from normal, as shown at right. There are two relationships corresponding to more U.S. landfalls over 1900-2016 at play here: first, lower pressures in the Tropics, and the second, lower pressures over northeast Canada and the northern Atlantic. The high-latitude July SLP anomalies are particularly interesting, as they can indicate the tendency for the negative phase of the North Atlantic Oscillation to be more common in the fall. U.S. hurricane landfalls are more likely in -NAO regimes, as the western Atlantic (Bermuda) high tends to be stronger and further west than average. This makes it less conducive for westward-moving storms turn north while they are still east of the U.S. coast. This year, high-latitude North American SLP has been well below normal, and the NAO has followed suit by remaining mostly negative in August so far. Long-range modelling suggests that western Atlantic ridging will wax and wane into September, but overall the further west than normal position of the Bermuda high in early August is a cause for some concern.

Not surprisingly, TigerTracks objective modelling is projecting higher than average U.S. landfall probabilities, though not on par with some of its most aggressive historical forecasts (like 2005). Another 2-5 tropical storm landfalls are likely, with an 80% chance of at least one and around a 50% chance of at least two U.S. hurricane landfalls. There is also a 50% chance of at least one major hurricane landfall in the U.S. this year, compared to the baseline odds of 35%. Geographically, these risks are most elevated relative to normal along the Gulf Coast and in Florida, with closer to normal risks in the Southeast U.S. and below normal risk in the Northeast U.S. While model skill is limited for U.S. landfalls, particularly at the granular regional level, the TigerTracks U.S. risk projections do agree with our qualitative observations and impressions of the season thus far.

The Bottom Line for the 2017 Hurricane Season

The core two months of hurricane season (August 15th-October 15th) are very nearly upon us, and recent/ongoing activity in the Atlantic Basin should be a reminder to be prepared. From our perspective, leading indicators are in good agreement that an above average hurricane season is underway in the Atlantic, in which around 140% of the activity in a normal season is expected. In terms of landfall risks, confidence is (as ever) lower, but here too there are signs that the chance of U.S. landfalls is elevated relative to climatology, particularly along the Gulf Coast. When will the U.S. break the 12-year major hurricane landfall drought? Brace yourselves, as it could happen as soon as before the next Spiderman reboot. At WeatherTiger, your friendly neighborhood tropical experts will be watching the season ahead with great interest and some apprehension.

Forecaster: Dr. Ryan Truchelut (ryan@weathertiger.com)

For a complete weekly forecast of Atlantic tropical activity, our tropical update video is posted each Wednesday by 6pm from June through November. You can also follow us on Twitter and Facebook for the latest.