Back-to-back superstorms

Update: regardless of what you think might be going on, there is no mathematical connection between this year's current # and intensity/categories of storms, versus any historic ocean temperatures.

No one likes to see consecutive, extreme natural disasters in the same part of the globe, and a lack of them in the past endows people with a false sense of how likely they truly are of occurring.  In this article, we’ll showcase various interesting statistics about the uncanny damage this years monster hurricanes are having on the U.S.  And while the economic toll may be record setting, particularly for the U.S. and even in inflation-adjusted terms, the loss of human life has been well below average for any major superstorm (Category 3+).


One in x-hundred year storm?
Seeing a few typical storms every hurricane season has led to a false sense of what we know about how any future extreme storm will unfold.  To cite Professor Taleb, “the empirical distribution is NOT empirical!”  Was Hurricane Harvey a once in 500-hundred year storm?  That’s unlikely and in some measures, it wasn’t the most extreme we’ve recorded (e.g., death toll, storm category, etc.)  Also, note it had the potential to be significantly worse if we consider than Houston is an economic hub and a top 5 most populated city.  Some lessons remain unlearned and reinforce horror, as noted in Zero Hedge the FEMA chief labels Hurricane Irma “a nuclear hurricane”.  No one really knows, including insurance companies in particular.

More Category 4+

Category 4 is the second worst hurricane reading.  And we are seeing perhaps more of them, including consecutively, in recent decades (though sample size again is low and per later in the article, there is no evidence to suggest climate change is behind this).  See this in the rightmost panel below.  At the same time, the time gap between severe hurricanes is curiously rising, as noted by the rising black-colored smooth curve in the rightmost panel (versus those in prior ½ century panels on the left). 

Several hundreds hurricanes were examined since mid-19th century.  And what’s remarkable therefore is that we have never seen many other situations such as Hurricanes Harvey and Irma before, how soon one Category 4 hurricane followed another in the same part of the world.  The only close example otherwise during the past ½ century were the Hurricanes Andrew and Iniki, both in late 1992, and on two different oceans!

Unexpectedly high catastrophe claims
While noting that this year’s hurricanes are not the worst we’ve seen in the U.S., they are the nastiest we’ve seen recently in Texas and Florida.  Catastrophe claims will easily be much higher than during normal seasons, and likely higher-per-person than what each state has seen in the past.  Fortunately in the U.S. the insurance company footprints tend to diversify their risk between these states, but not so for the reinsurers who have been highly impacted.

Lower than typical deaths

Of the eight costliest hurricanes to date (before this year), Hurricane Harvey’s death toll (currently at 70) is below the median.  And with “only” a handful of international deaths so far associated with Hurricane Irma as it pancaked a small number of Caribbean Islands (including Sir Richard Branson’s Necker Island), we might see similar low figures in the U.S.

Predicting the worst hurricanes and earthquakes
Of the worst-ever global risk events, we’ve seen more and the more common hurricanes and earthquakes among them!  Others shown below, are highly infrequent and can have very difficult to assess tail-risk. 

Where's the climate change increase?

As a final topic, the way one partitions data can give mixed signal on results.  Taking the past 150+ years of data, and slicing it in two equal time periods, we see the number of times states were impacted by major hurricanes in the first half is 49 and the number in the second half is 55.  The patterns shown below are similar.  Only time will tell if worsening temperatures and climate is sustained and leads to greater natural disasters.  Right now it would be false to jump to conclusions given the penchant scientists have toward model tinkering.

Finally, for more on the theory of multi-dimensional stochastics to understand a hurricane's geographic impact (or other advanced applications), see Chapter 4 of Statistics Topics.  And note that Barry Ritholtz has documented a list of charitable resources that serve to aid Hurricane Harvey victims.