Yes, the title is a bit excessive, I admit, but we had to put our feet in the dish!
This is not to discuss the quality and relevance of this kind of forecast. Overall, these forecasts are more and more reliable with the years and technical developments. On the other hand, it is a question of showing that the correlation that everyone thinks is obvious between an active season and the risk for a given territory is not proven. And I will demonstrate it to you with illustrated examples and explain to you why it is not statistically founded.
The Caribbean arc is not the Atlantic basin
The first thing to absolutely integrate when reading these forecasts is that they speak of the Atlantic basin in its entirety, which includes the entire North Atlantic, the Caribbean Sea and the Gulf of Mexico. The West Indies, even if it is positioned in a privileged area for cyclonic development, remains only a small part. If we take the example of this first month of the 2020 season, we have 4 named systems and none concerned the arc. And this example is far from isolated.
Therefore, a very active season in the Atlantic basin does not necessarily imply a very active season in the West Indian region. The best example is that of 2005. It is the most active season since 1950 with no less than 29 cyclones .. we had to use the Greek alphabet to name them !!
And yet if you look at the map you will realize that only one cyclone (Emily) has passed over the extreme south of the tropical storm arc.
In 1967, the second most active season, 2 systems (out of 26) passed over the West Indies, one in depression and the other in storm.
We have to go back to 1995, the third most active season with 3 systems, to see 19 cyclones pass with Luis near the northern islands in Cat.3, Marilyn on Dominica in Cat.4 and Iris on Guadeloupe in storm. 1 cyclones is roughly the gross seasonal average over the West Indian region.
We can therefore see that over the 3 most active seasons, which counted a total of 74 systems, 5 passed through the West Indies, of which only 2 were in hurricanes (1 Cat. 1 and 1 Cat. 4). The global over-activity in the Atlantic is not a sign of activity on the West Indies and even less on a given island.
The West Indies is great
A second thing to integrate is that at the scale of an island, the West Indies is large! If we take Martinique for example, we have a length of the Atlantic coast which represents less than 10% of the length (approximate) of the arc. And Martinique is the second largest island in the arc behind Guadeloupe. So even if 2 systems were to cross the West Indies (which is rare, the average is less than 5) the risk for a specific island would still be low. For the northern islands, the quantitative risk is therefore even lower. On the other hand, on the north of the arc, the risk of qualified intensity is higher.
I would come back in another article to the notion of the linear exposure of each island to cyclonic risk, which with its latitude is the determining element of statistical risk.
The statistics speak for themselves
If we take the figures for 70 years, the proportion of systems on the Atlantic having touched or approached an island in the West Indies is ultimately low.
And if we bring that back to the forecast for a season, there is no longer any relevant statistical impact.
Consider a very active planned season, with 22 named systems. The statistical risk for Martinique would be 22 × 2% or 0,44%. Consider a planned season of little activity, with 12 named systems. The statistical risk for Martinique would be 12 * 2% or 0,24%. The difference between 0,44% and 0,24% is totally marginal : it's just called luck or bad luck. Nominal deviations of this level are absolutely not representative for determining a risk!
So, at the level of an island, the risk is roughly equivalent and proven… whether the season is active or not!
What conclusions can be drawn ?
The first is the one I remember ALL THE TIME: whatever the forecast, preparation and Watching must be strictly the same. This is the key point of prevention. Whether the season is active or not, the statistical risk is almost the same for a given small island. It is therefore not necessary to stress in the event of an active planned season and it is not necessary to be lax in the event of a less active planned season.
The second is that these forecasts are tools intended to give an overall view of risk. But even more a vision of risk for the United States. Because that is what it is all about: the risk for the largest territory concerned. Almost 25% of cyclones in the Atlantic basin affect a coast of the United States. In this case, the difference between an active and less active season becomes relevant and the forecast makes it possible to anticipate the means to be implemented.
The third is to always relate to the raw data and the facts. A forecast is a mathematical choice based on variable initialization (forcing). And depending on this initialization, we will have results that are different or adapted to certain objectives, which means that this forecast is intimately linked to the vision of things and to the needs of the person who publishes it. In this case those who publish these forecasts have an objective linked to the risk on the USA and not that for an island of 80km long. And I hate myself for giving this forecast information when it comes out because it generates mostly totally irrational behaviors and analysis (and researching the media buzz doesn't help temper).
We live in a hurricane zone and the danger is there whatever the activity of the season. This is the only thing that matters and that should guide our prevention choices for this risk during the hurricane season, and especially not reading a forecast for which we are statistically irrelevant!
Find all the hurricane statistics on the dedicated pages: Cyclonic statistics of the Atlantic basin, hurricane statistics of the West Indies et island hurricane statistics.
