Hurricane Ian inflicted massive destruction across many parts of Florida and the Carolinas. The power of nature dwarfs any destruction that humans can create. In spite of all the carnage that hurricanes produce, early warning and constantly improving forecasting facilitate timely responses that reduce risks and provide early warning to keep people out of harm’s way.
Hurricane season runs from June 1 through Nov. 30, representing one-half of the calendar year. The most active months historically have been September and October, when the most hurricanes typically have occurred.
Many people believe that hurricane frequency is increasing, and their virulence is intensifying. Does the data support this belief?
Full disclosure: I am neither a meteorologist nor a hurricane specialist. I am a data scientist. As such, I wanted to confirm this widely reported belief for myself and answer this question. To this end, I focused on the North Atlantic hurricane season.
Using data available from the Colorado State University Tropical Meteorology Project going back to 1852, I grouped the years up into 10 year periods, covering 1852-1861, 1862-1871, and so forth through 2012-2021. This smoothed out any anomalous years, providing 17 periods to compare and analyze. For each of these periods, the total number of named storms, hurricanes and major hurricanes were computed.
The 2002-2011 and 2012-2021 periods had the most named storms, totaling 159 and 168 respectively. The next highest count was 1952-1961 at 120. By comparison, 1942-1951 had 117 named storms and 1872-1881 had 74 named storms.
Focusing on hurricanes, these two most recent periods also had the largest number, although the difference with other periods was less pronounced. For example, 1962-1971 had 63 hurricanes, less than the 76 reported in 2002-2011 and the 73 reported in 2012-2021.
For major hurricanes, the same trends occurred, with 38 reported in 2002-2011 (the peak period) and 31 reported in 2012-2021. However, 1952-1961 had 27 major hurricanes, less than, but in the same range as these numbers.
What is more informative is the virulence of these storms, as measured by the accumulated cyclone energy. This is computed by summing the square of a storm’s maximum sustained winds over every six hours, divided by 10,000.
Note that virulence is different from damage, which is measured by the cost of the carnage from the named storm. A highly virulent storm that strikes only remote areas may be less damaging than a weaker storm that strikes a major urban area.
Averaging the accumulated cyclone energy per named storm provides a measure of the virulence of these storms over each period. This number has been remarkably consistent for several decades, with the average fluctuating between seven and 10 in every period since 1932.
Since 1932, the peak period was 1992-2001, with a 9.9 average. The two most recent periods were 8.57 in 2002-2011 and 7.47 in 2012-2021. By comparison, the 1892-1901 period average was 13.50, the largest such average. The 1982-1991 period average was the lowest at 7.19.
What do these observations mean about the future?
Not much. Data is like looking into a rearview mirror. It tells us where we have been, but not as clearly where we may be heading.
There are limitations with this analysis. Going back 50, 100 or 150 years, storm tracking was less dependable due to the limitations of technologies that were available then compared to today. Satellites make observations and measurements more precise, which means that the most recent data are more reliable than older data. For example, were storms missed in the late 19th century or early 20th century that today’s tracking technology would have detected and counted? Most certainly, yes.
Averages also obfuscate rare events. A cluster of highly virulent hurricanes during any given year can skew averages making them less informative.
How the years are grouped can also lead to higher counts in some periods and lower counts in adjacent periods.
What all this means is that the severity of hurricane seasons depends on what you measure and how you present it. The number of storms over the past two decades does appear higher than in previous time periods. The average virulence suggests that their severity has been surprisingly similar. For both these measures, tracking and measurement accuracy could account for some of this.
Natural disasters like hurricanes are a significant risk to lives and property, as Ian demonstrated. It is highly likely that sometime over the next month, yet another major hurricane will strike the United States mainland. New tropical threats are regularly being tracked. Such events are what define hurricane season.
Will 2022 be worse than previous years? We will know soon enough. Early predictions suggested that it would be. Until we reach Dec. 1, detection, preparation and action remain critical to keep people safe.
Sheldon H. Jacobson, Ph.D., is a professor in computer science at the University of Illinois Urbana-Champaign. A data scientist, he applies his expertise in data-driven risk-based decision-making to evaluate and inform public policy.