A Costly Climate Characteristic
For the Caribbean and areas of the USA and Mexico that are close to or bordering the Gulf of Mexico or North Atlantic, the hurricane season regularly inflicts a heavy cost, both socially and economically. The past ten years, for example, have averaged 542 fatalities and $47.405 billion in damages (see table below).
You can also see that even ‘quiet’ seasons may bring more than a few fatalities and hundreds of $million in damages.
The per-season statistics are sourced from wikipedia.org
When it comes to prediction, it’s the most active and therefore most costly seasons that draw focus. In the table, bold text highlights the four most active seasons of the past ten. These include the costliest (2017) and 3rd costliest (2012) seasons on record.
Three of those four had 10 or more hurricanes. That’s statistic has the strongest correlation with the degree of impact, when considering both fatalities and economic cost.
Anticipating Busy Seasons Using Analogue Years
MetSwift’s research has identified 5 climate drivers (teleconnections) to have a significant effect on the amount of activity in a hurricane season (out of 15 investigated).
By comparing how these are expected to behave during an upcoming hurricane season with how they behaved during historical years, a selection of most similar years can be identified. These are the analogue years. The average activity of these years can then be used as a prediction for the season.
The December-issue MetSwift predictions have found to offer an appreciable (12%) improvement on using the climatological average, when comparing predictions for 2010-2019 with the average of 1950-2019.
Predicted (by average of MetSwift analogues, rounded to nearest whole) versus observed number of hurricanes each season. The 1950-2019 average is 6.24 hurricanes, which is 0.76 further from the observed mean of 2010-2019 than the MetSwift analogues average prediction for those seasons.
MetSwift’s December prediction outperforms the forecasts issued by 4 leading organisations four or five months later (see below), several of which performed worse than climatology during 2010-19.
For the seasons 2010-2019. A positive value is an improvement on climatology. NOAA are the leading competitor with their outlooks issued in May. It’s a strange but true discovery that TSR’s December-issued outlooks have performed much better than their April-issued ones! Please note that NCSU and NOAA forecasts consist of ranges instead of single numbers – averages have been taken for this analysis.
Having seen this impressive track record, you may be interested in what the MetSwift analogues predict for the upcoming hurricane season…
The MetSwift Prediction for the 2020 Atlantic Hurricane Season
Broadly speaking, the Atlantic Ocean is expected to see patterns of warmth conducive to more hurricanes than usual – but not as much so as in 2017 or 2012.
Meanwhile, ocean heat content and surface temperatures in the eastern half of the tropical Pacific are expected to become increasingly below normal, as a phenomenon known as a La Niña event develops.
Overall, this is a recipe for an active but not exceptional hurricane season.
As you can see in the table below, the analogues were already picking up on this in December, with a slightly above-average season being predicted. Since then, it’s become clearer that La Nina event is likely to establish by the autumn. In response to this, the latest update predicts a higher number of hurricanes and major hurricanes.
A Modest Outlook Compared to Most Others
Interestingly, the MetSwift analogues prediction is among the lowest to have been issued for hurricane season 2020. As you can see here, most organisations are going for a highly active season. The one exception is UKMO, which have just released a prediction almost identical to MetSwift’s!
This is likely down to uncertainties regarding just how fast the expected La Niña event develops and how warm the Atlantic tropical waters (TNA) are during the peak activity months of Aug-Oct.
The TNA has shown a cooling trend between March and May. Perhaps the other organisations issuing outlooks in April did not foresee this?
Regardless, the 2020 hurricane season is likely to be at least a little more active than usual. If you’re in a risk area, good luck and stay safe!
Head Meteorologist at MetSwift
* For relevant locations worldwide, MetSwift’s NatCat Risk model offers postcode-specific hurricane (or other tropical cyclone types) risk assessment based on official historical records. Regional overviews are also available.
Cover Photo by Unknown Author is licensed under CC BY-SA