• Richard Dixon

Three Thoughts on the 2020 Atlantic Hurricane Season

Plenty has been written about the forthcoming hurricane season and its potential as indicated by the now myriad hurricane season forecasts and the possibly complex intertwining with the COVID-19 situation. I won't regurgitate the seasonal forecast information and it's all available here, however I wanted to highlight three things that I've been chewing on.

1) What if it turns out to be another 2005-style season?

2005 was a blockbuster season. Warm seas and a train of storm after storm guided towards land by high pressure near Bermuda. The fact that our seas are warming should mean that 2005 type seasons will still be quite a shock to the system, but really shouldn't be a total surprise.

There is still a lot of debate in the science community on the uptick in sea surface temperatures (SSTs) and whether this might be having an influence on the strength of our tropical cyclones. Recent work from James Kossin and co-workers has looked at the global uptick in numbers of major hurricanes worldwide and the chart on the right here according to the paper has 98% statistical significance. Admittedly, not everyone is happy with this work but the data provided gives us an interesting talking point and I far as I'm concerned, the debate will continue until we've got adequate hurricane-resolving climate models to help us understand this uptick better: we certainly won't get very far very quickly using new historical data.

The way I see it from a catastrophe modelling perspective is that there are some years when things "click" and the larger-scale synoptic setup ends up pointing storms toward the US East Coast. All things being equal, if you have a warmer sea in one of these years, it suggests that more of those storms approaching have the potential to become more intense, so our "tail years" in the EP curve could, quietly, be rising. There are always other factors [vertical shear, El Nino, Saharan dust] always at play in any one hurricane season of course.

A 2005-style season might add a bit more fuel to the fire that tail risk is changing.

2) What if it turns out to be another 2010-style season?

Right, let's look at it from a different standpoint. Remember 2010? The Colorado State University (CSU) hurricane forecast in June correctly called for an above-average 10 hurricanes - we actually got 12. And yet - from the light-coloured dot on the chart below (labelled 10 for 2010), none of these made a US mainland landfall, despite that season having the 2nd most hurricanes in the basin since 1970.

Graph showing (x-axis) the yearly ACE Index versus (y-axis) the number of hurricanes in the basin for each year from 1970 to 2019. Colour of dots indicates the number of US mainland hurricane landfalls for that year. Dates abbreviated by last two numbers of year.

The 2010 season - and the surrounding seasons around then have always fascinated me. The major landfalling hurricane drought flagged up some of the information coming out of climate model output that suggests a shift to lower landfalling activity via a shift in the broad-scale motion of the tracks in the Atlantic. It's also been reported to a certain extent in historical data, which might suggest it's happening already.

The image below (apologies for the quality - it's as good as it was in the original document), taken from this paper, shows the 100-year percentage wind gust change and storm surge height change at the coast in 6000 years of simulations from 1951-2011 versus 2051-2111 under the 2-degree warming scenario. It shows a broad increase of windspeeds in the mid-Atlantic and a broad decrease close to the US mainland, which to me signifies a shift to more tracks staying out to sea: not that this is necessarily a good thing for Caribbean. And remember - this is only one model.

This image has always stuck with me as it mimics - to a certain extent - what we saw from 2006-2016 where tracks stayed more out to sea.

Was 2006-2016 a glimpse of the future - or indeed our present-day risk?

3) What's Happening to the AMO Index?

The Atlantic Multi-decadal Index tracks the see-sawing over a 20-30 time period of ocean basin temperatures in the Atlantic. Some of us will have used them in cat modelling to delineate the landfalling basin activity into warm and cold years, something I wrote about a while ago on this blog looking into the uncertainties around using this data.

If you do a simple search online you'll find two readily-available AMO indices, one from NOAA (I've used the unsmoothed data on their website) and another from Colorado State University (CSU) - the coloured axis labels correspond to the colour of the lines.

The two indices typically follow each other through the phases of positive and negative AMO index: you can see the post-1995 warm period pretty much in both, but since about 2014, the CSU index has gone negative - and remained negative. This year, the difference between the two indices is larger than it ever has been. I'll admit here to being less au fait with ocean dynamics, but if we are using these indices to try and understand the see-sawing of the ocean's warmth, understanding which index to use would be helpful.

For reference: the CSU index uses "1) Sea surface temperatures (SSTs) in the far North Atlantic from 50-60°N, 50-10°W and 2) sea level pressures (SLPs) in the North Atlantic from 0-50°N, 70-10°W". The NOAA index uses just SSTs, and these are from 0-70°N, rather than the 50-60°N values used by CSU. The difference shown above is related to the averaging region, but this divergence in values is - let's say - unhelpful if we're trying to understand whether we're heading into another cooler phase of the AMO (if indeed it exists).

As well as the uncertainty shown above, another issue for me here is - if we are heading into a cooler phase of the AMO - whether some of the cooling gets lost to a certain extent in the globally warming sea trend?