Loading...
Hurricane Landfalls: Where the Rubber Hits the Road
In the second of this two-part article, we dive into the world of landfalling storm behaviour: after all, landfalls drive losses. In the last post, we ended by asking whether the assumption we’d made that the percentage of “named storms” that make landfall in the Atlantic Basin was a static value.
Insights20 November 2025
Today, we investigate this with the help of Reask, leaders in climate-conditioned modelling of extreme weather, with whom we partnered in 2025 to help us understand the shift in global tropical cyclone risk in a changing climate.
How have landfalling storms changed in recent years? The chart below summarises some details of the previous 44 years of hurricane seasons, using the same metrics from our last Insights post, plotting the percentage of named storms by season that have made landfall as a hurricane. The data has only been split in two here to give a meaningful amount of historical data in each period.
If we contrast the solid blue line and the red line that represent the mean value of two 22-year periods from 1981 to 2002 and 2003 to 2024, respectively, we can see that the average percentage is fairly flat but has dropped slightly from 12.5% to 10.5% across the two 21-year periods where satellite data tracking storms has been fairly reliable.
We’ve also had the much documented, but relatively little-researched “hurricane drought” from 2006 to 2016, where no major hurricanes made landfall for an unprecedented 11 years. Assuming that this is nothing more than chance, we can get an alternative take on the period 2003-2024, minus this drought period: the mean in the dashed red line, giving a landfall percentage of 15.9% across this period.
But is this a wise thing to do? We’re dealing with very short datasets here, but it would be useful to understand whether landfalling activity remains flat across this period or whether there is any shift in the data.
This is an example of where our partnership with Reask comes to the fore. Their ensemble methodology of stochastic track development forced by the ERA5 reanalysis allows us to run each year multiple times to see whether their stochastically generated storm tracks show any changes in the landfalling behaviour.
We can contrast the behaviour of history (again, 1981 to 2024) in terms of basin-wide storm activity vs. landfalling hurricane activity from our one “take” on history with Reask’s 2500 simulations of each year, forced by historical atmospheric conditions.
Our brief comparison here shows how the historical dataset and Reask’s dataset tie together well: remembering that each of Reask’s datapoints here represent 2500 resamples of each year and we demonstrate the range of Reask’s simulation data with “whiskers” on each point that highlight a wide range of possible difference outcomes for each year.
As you might expect, as basin activity increases, landfalling activity increases, and this is the case for both our historical example and the Reask multiple simulations of history.
But how has the landfalling activity changed when we take into consideration the intriguing behaviour in the second half of the 1981-2024 period with the landfalling drought?
In the next chart, we show the data from the previous chart a little differently: a scatter plot showing the historical data from the previous chart split into two periods alongside, again, 2500 simulations of 1981 to 2024 from the Reask ERA5 stochastic model.
The x-axis shows the average landfalling percentage for 1981 to 2002 and the y-axis the same, but for 2003-2024. The green triangle is historical data; the red triangle history but with the “landfalling drought” removed. The blue dots are all the 2500 Reask simulations, and we’ve also included the mean Reask result across all the simulations in the white square.
The three lines on the chart point delineate whether for each example of history there is no change, 20% increase or 40% increase in the frequency of landfalling hurricanes between the two periods of 1981-2002 and 2003-2024.
There are three key things to point out on the plot:
- When you simulate “history” many times, because of the short nature of the dataset, there is a wide scatter of data in the simulations
- Maybe most interestingly, the Reask simulations largely point towards an increase in the percentage of named storms making land as hurricanes, but not exclusively.
- The red triangle that removes the hurricane drought appears a little more in tune with the many re-simulations of history; both fall into the “20-40% increase in percentage of tropical storms making landfall” lines superimposed on the plot
So, what could be happening here? For those regular users of catastrophe models, much time is spent trying to reconcile what’s happened in a limited period of history with what is being suggested from a model of synthetic events that are looking to provide extra information round the edges.
One theory could be that we’re simply seeing an increasing percentage of named storms going on to make landfall as hurricanes merely because warming seas are making storms stay longer at hurricane strength: an important and potentially overlooked consideration when considering tropical cyclone risk. This may not necessarily be evident from the historical data that we have: and we’re not aided by the anomalous “hurricane” drought from 2006-2016.
When history provides scattered data points that might show a weak trend, or hide a trend, this short example demonstrates how we can ally historical data with historical re-simulations from our partnership with Reask to help build confidence in understanding what might be happening in our historical hurricane record as we work towards the aim of pricing risk fairly for the present day.
Richard Dixon has 25 years of experience in the insurance industry building, researching and evaluating catastrophe models. He has a specific interest in understanding whether and how climate change is reshaping catastrophe risk, challenging conventional assumptions and developing new approaches to risk assessment. He has been a Visiting Research Fellow at Department of Meteorology at the University of Reading for the past 7 years and is a Fellow of the Royal Meteorological Society.
Hannah Croad is a catastrophe research analyst at OAK Global, holding a PhD in Arctic cyclones and an MMet in Meteorology from the University of Reading. She has a strong academic background in atmospheric science, with experience conducting research and publishing in scientific journals. Hannah’s work focuses on the intersection of meteorology and reinsurance, with a particular interest in understanding how climate change may be altering catastrophe risk.