• Back


A Look At Hurricane Clustering

The official 2020 Atlantic hurricane season is slowly coming to an end, albeit without seemingly wanting to halt storms formation. While it is too early for any quantitative assessment of the impact on the (re)insurance and capital markets industry of this season (putting aside the complication of the SARS-CoV2 pandemic), it is apparent that this season has broken, or at least matched, several historic records with many drawing parallels with 2005, the most extreme season on record, the year Katrina struck.

Storm Name

Landfall date

Approximate Landfall location 

Estimated maximum 1-min sustained winds [mph]

Tropical Storm Cristobal 

June 07

Between Mississippi River and Grand Isle, LA


Hurricane Hanna

July 25

Padre Island, TX


Hurricane Laura

August 27 

Cameron, LA


Tropical Storm Marco 

August 24 

Mouth of Mississippi, LA 


Hurricane Sally

Sep 15 

Gulf Shores, AL


Tropical Storm Beta

Sep 21

Matagorda Peninsula, TX 


Hurricane Delta

October 09

Creole, LA


But one question in particular stands-out. Even the casual observer would have not failed to notice that there was a preponderance of hurricane landfalls along the US Gulf Coast, centered around Louisiana. A number of these events took almost the same path. A question for many insurance professionals and regular citizens alike: is this observation an unusual one, or is 2020 a harbinger of future hurricane landfalls?

2020 Continental U.S. Named Storms Landfall through November 8 2020
2020 Continental U.S. Named Storms Landfall through November 8 2020

Fifty Years of Clustering Data

A paper by Kossin et al from 2010 examined the climate effects on hurricane tracks. Using clustering techniques, the authors split the historical hurricane tracks from 1950 to 2007 into 4 distinct groups. A group they denoted as Cluster 4 were storm forming in either the Atlantic just east of the Caribbean, or in the Caribbean itself, tracking into the Gulf, and making a landfall (if they didn’t dissipate) along the US Gulf Coast. Amongst the characteristics of this group was a:

  • high likelihood of being a major hurricane;
  • very high likelihood of making landfall; and,
  • higher likelihood of having a higher intensity at landfall

In a 2018 paper to the Casualty Actuaries Europe meeting (PDF), Dimitris Papachristou of the Bank of England PRA extended upon the work of Kossin et al to ask if hurricanes come in clusters. Papachristou found that:

  • Cluster 4 storm numbers and volatility has been increasing over time
  • The likelihood of 2 or more Cluster 4 major hurricane making landfall in year is 1-in-7 modelled (or 1-in-11 actual)
  • The likelihood of 3 or more Cluster 4 major hurricane making landfall in year is 1-in-30 modelled (or 1-in-68 actual)

2020 Hurricane Season: The New Normal?

In a nutshell, the work referenced above shows that the events we saw in 2020 are not actually unusual - we would expect 2 major landfalling hurricanes along the Gulf coast with a 10% probability every year.

The next question to ask is, why? What causes hurricanes to cluster so? The summer months over North America are, from a weather standpoint, relatively quiescent. Weather patterns over the US are dominated by high pressure systems which seemingly set in place. The jet stream is typically at higher latitudes so isn’t there to ‘push’ weather around. We see, for example, the Bermuda High - a high pressure system over Bermuda which locks in place for most of the summer months. The Bermuda High can have a major controlling effect on whether hurricanes forming in the deep tropics hit the US East Coast. So, if we have a relatively stable atmosphere in which weather systems take the same path, and we have hurricanes forming in roughly the same place, it’s not unreasonable to expect that what hurricanes form will go the same way.

So, 2020, whilst seemingly extreme for the residents of Louisiana especially, such behavior is not unusual and is in-fact explicable. However, the findings that the number of storms forming and moving into the Gulf is increasing, is a major worry for residents of the Gulf coast and the insurers that support them.  This also indicates that there are likely to be continued years of clustering, driving volatile return profiles and capital needs for the reinsurance and retro markets. In many ways, this validates the periodic “Class of 20XX” phenomena for reinsurance startups which is being witnessed again this year.

More Posts

Card image cap
The Data Analytics Report: Q&A with Alastair Speare-Cole

Published Sep 07, 2021

Card image cap
Creating value through insurance data infrastructure

Published Aug 05, 2021

Card image cap
From Data to Intelligence: Systems Alchemy for the Insurance Sector

Published Aug 05, 2021

Card image cap
Part 1: How to Not Get Lost in Translation

Published Jul 23, 2021