• Richard Dixon

A Change in Atlantic Hurricane Potential?

If you notice how far east in the Atlantic Basin Ophelia was in 2018, how far east Hurricane Florence was earlier that season for a 110-knot system, how intense and late in the season hurricane Eta was before it made landfall and have an inquiring mind, these last few hurricane seasons have been fascinating.

In the light of these storms seemingly pushing the boundaries of strength and location in recent years, the question I wanted to ask is:

Are conditions for supporting hurricanes becoming more frequent and widespread?

And let's make it very clear first up however, that I'm definitely not attempting to answer here:

a) Is the formation of tropical storms becoming more frequent?

b) Are hurricane landfalls becoming more frequent?

So, what sort of environmental conditions are conducive to supporting a healthy hurricane? Well I'm going to look at three here:

a) Sea-surface temperatures > 26.5c

b) Relatively high humidity in the low-to-middle levels of the atmosphere*

c) Low vertical wind shear (increase of wind with height in the atmosphere)*

For this, I've used the ECMWF ERA5 reanalysis of the atmosphere and ocean to answer the question of where the three conditions above are all met at the same time in the atmosphere. The ERA5 data was recently extended back to 1950 - which for me is an important extension to their dataset as it covers two periods of "warm seas" associated with warm phases of the Atlantic Multidecadal Oscillation (AMO), that is if you believe such a thing exists.

The chart below shows, for various regions across the Atlantic shown in the bottom-right plot, the sum of daily area in each box that meets all three of the criteria, summed across the entire year.

The first thing to point out is that for the Main Development Region of the Atlantic (the purple area and graph), this yearly time-integrated area that can support hurricanes isn't increasing year on year. However, if we look at the area in the other boxes that are primarily closer to landfall in the Atlantic, the area that can support these regions is growing over time. This, I think, is pretty interesting. If I interpret what we're seeing here, I reckon:

If a hurricane forms in, or and moves toward, the western Atlantic, we're now seeing an increased chance that the oceanic and atmospheric environment can support it.

Let's delve into the data a little more. One thing I'm quite wary of is that as much as we're in a warm sea period now, we also were in one during the 1950s. So, instead of showing the above data with line graphs, I've taken a look into comparing the present day and last positive AMO period by contrasting the frequency that our three criteria are met in the 2010s versus the 1950s.

The chart below shows (left) the change in the number of days where the potential for supporting hurricanes has been met and on the right, this described as a percentage change.

To pick a couple of regions on this map:

a) For a large region off the coast of Florida just north of the Caribbean, we're seeing an increase of 14-28 days where this criteria is met - or an increase of somewhere between 50-100 percent.

b) On the north and east borders of this region, we're seeing >100% increases in frequency of where hurricanes can be supported. Understandably interesting are the 100-200+% increases in frequency of occurrence just off the coast of the mid-Atlantic. Is the risk in the NE US seaboard quietly increasing markedly because hurricanes can be supported further north? Hard to tell, but it's definitely an eye-opener.

There is also the discussion around whether warmer seas mean a longer hurricane season. So here's the above map, but restricted to the months December to May, outside of the June-to-November confines of the hurricane season:

We can see how outside of the hurricane season we're seeing an increase in days supportive of hurricane activity in the southern North Atlantic. The suggestion that we're seeing a longer hurricane season because we're seeing more systems in May or December is often met with the rebuttal that we're observing them more easily through satellites and more earth observations, which makes sense. However, this data also suggests that we are seeing better environments for supporting tropical systems beyond the ends of the traditional (Jun 1st - Nov 30th) hurricane season.

You can also notice the decrease in favourable conditions to the south of the Caribbean outside the season: I've not dug into this yet, but am intrigued nonetheless.

So - essentially a lot of what I'm showing here is suggestive that the environment in which our hurricanes can grow is changing. The current warm phase of the Atlantic appears to be different from the last warm phase. The area and time over which hurricanes can be supported is growing.

But....we can go one step further here. The work of Bister and Emanuel allows us to take the properties of the column of the air of the atmosphere to understand the potential maximum hurricane intensity of a storm that could be supported by the atmosphere at that place and time.

What I've shown here is the change in frequency of hurricane potential intensity (PI) when there is low shear and high relative humidity, per two of the restrictions above. I'm not using the SST criterion here.

One thing that is fairly obvious is that in the western Atlantic, there is an increase in the frequency of storms between 1950 and 2010. As we up the potential intensity threshold that we are using, the area of the Atlantic that can support these intensities is smaller - however the frequency at which these higher intensities can be supported increases markedly.

We can demonstrate this by picking a point on the map - and here I'm choosing a location just east of Florida in the Bahamas at 26N, 76W, to look at the shift in potential hurricane intensity [remember again - this is just potential based on the background atmospheric conditions].

The chart on the right here shows how the return period of certain strengths of hurricane potential intensity shifts between the 1950s and 2010s. You can see how the return period of the more common-or-garden strengths aren't shifting much, however when you move to the rarer strengths, the atmosphere is becoming more supportive of these more frequently.

These type of small shifts in short-RP events with larger shifts in longer-RP events I spoke about at the ISCM climate meeting re: wildfire potential (2nd-to-last slide), and we're seeing the same here, it would seem.

This does suggest that older hurricane data is less relevant to the present day as the conditions for hurricane development were rather different and we likely have to take, if possible, a present-day view of risk to reflect the present-day hurricane potential.

This, obviously, is not easy with short historical records to represent the present-day, but there are glimmers of hope in how we build models and think about the fact that the underlying risk may be shifting - and here I'll point you to the hurricane work of the researchers at Columbia University and the hurricane model they are building that has recently been supported by Aon and at Reask in association with Twelve Capital - both of whom are helping us think about the shifting landscape of risk in novel but robust ways.

Let me end by climbing upon my soapbox (again):

I'll admit to have been a bit of a cynic on climate impacts on present-day hurricane risk as the numbers in terms of losses don't necessarily show it. But hopefully you've been seeing on this blog how I've been doing a bit of research** to educate myself (and, with a bit of luck, you) as to what might be happening by looking well beyond the loss or landfalling numbers.

The climate topic has a lot of noisy people on either side of the debate and I've tried to keep my fingers in my ears to inform myself in as unbiased way as possible: my recommendation is that you try and do similar !

* For those interested in the thresholds I've used, I've looked at average humidity between 850 and 550hPa > 60% and wind shear < 20 knots calculated between 200 and 850hPa.

** This research is not peer reviewed. I think it's important to note this.