Extreme Events: Developing a Research Agenda for the 21st Century

Workshop Background

Extreme Events: The Definition Thing

A snail was robbed by a turtle. When the snail reported the robbery to the police, he was asked if he could describe the turtle's appearance. "I didn't get a good look at him," the snail said. "I mean, it all happened so quickly."

You will be participating in a workshop on "extreme events." If the workshop is to be successful, we will need quickly to get beyond "the definition thing." To help us do so, let's imagine instead that the workshop is about "tall things," rather than "extreme events." In this case, we would want to define what we mean by "tall things." After all, the adjective "tall" and the noun "things" do not combine to suggest a very rigid category; indeed, they evoke the sense of a Rorschach test.

A conversation among tall things experts might go like this:

Workshop Overview

Integrated Summary

Workshop Participants

Other Workshop Attendees

Workshop Agenda

Workshop Logistics

Contributed Papers

Breakout Groups

Appendix: Research Topics

Workshop Proposal

Workshop Background

One workshop participant might begin by saying that the Empire State Building, Mt. Everest, and the California Redwood are all, obviously, "tall things." But, another participant suggests, what about the Denver Post Building in downtown Denver and Mt. Rogers in Virginia and the Douglas Fir, aren't they tall, too? Well shoot, a third person says, we might as well just claim that the Washington Monument, Scottsbluff, Nebraska, and an apple tree are all tall things. To which a now–exasperated listener might reply that it's beginning to sound as if all buildings, hills, and trees are tall things. A person in the back, thus far quiet, now pipes up in a smart–aleck fashion that everyone is missing the point – we have forgotten Kareem Abdul Jabbar, Paul Bunyon, and even Bill Bradley. The earnest person who began the discussion might at this juncture let out a sigh, and suggest that we are making a mockery of the exercise, as tall is obviously a relative term ...

Ah-ha, the moderator now says. We have reached an important point in the discussion. We must ask: "Tall with respect to what?" The Empire State Building might be considered tall with respect to the earth-bound perspective of most people (and most buildings) but it is fairly insignificant when compared to most mountains. And Paul Bunyon is not real, so is he really a "thing?" The story tells us that were he real, he indeed would be quite tall with respect to most ordinary real people, but does this count?

In short, the moderator summarizes, "Context of both tallness and thingness matters for our definition."

And so it is with "extreme events." The term "extreme" is a relative and contextual modifier. A 3-inch rainfall in New York City is enough to bring the city to a standstill. A 3-inch rainfall in Miami might not be worth notice.

A statistician might jump in here and say, "We statisticians have solved this definitional problem. An extreme event is one that occurs at the tails of a particular distribution of events." But the emergency manager from a coastal community might take issue, by arguing, "Well, that is all well and good, but for our community ANY landfalling hurricane is an extreme event, even if it is an average storm!" The statistician would respond quickly, to preserve the paradigm and save face, "Yes, but in this case, the relevant events are not all hurricanes, but the distribution of daily weather experienced by your community."

The moderator speaks. "So what we seem to be saying, then, is that not only are extremes defined contextually, but so are the event categories which can give rise to extremes."

"Well I suppose so," says the statistician.

An extreme event can be defined as follows:

An occurrence that with respect to some class of related occurrences, is either notable, rare, unique, profound, or otherwise significant in terms of its impacts, effects, or outcomes.

For example, we might say that a hurricane is an extreme event ...

  • with respect to other hurricanes when it reaches category 3 strength;
  • with respect to typical wind conditions most anywhere on the planet;
  • with respect to coastal communities when it is forecast to make landfall with winds greater than 35 knots;
  • with respect to mangrove forests when it topples a certain percentage of stands;
  • with respect to the federal government when a governor requests a presidential disaster declaration;
  • with respect to the insurance industry when claims exceed industry reserves;
  • with respect to the construction industry when the number of homes destroyed lead to demand for rebuilding greater than local industry capacity.

These various definitions of a hurricane as an extreme event are not mutually exclusive. Indeed, hurricane Andrew was an extreme event according to all of these definitions. But many other storms fit only one or a few of these definitions.

The definition of "extreme event" requires that two questions be answered:

1. What class of events are considered as the "universe" of relevant occurrences?


2. What criteria are used to define "extremeness"?

This contextual definition, and these questions, give rise to researchable questions having to do with a) the relevant class of events, and b) the thresholds of extremeness. For example, returning to hurricanes, we might as: How rare was the hurricane Andrew disaster?

To answer this question we might begin by looking at rainfall events of the magnitude of Andrew. But this would not be enough – we would also want to know something about the degree of societal exposure in regions prone to such deluges. We might begin to address this question by looking backwards and asking how many such events (or combination of events) have we seen in the past? What about the future? Can we project present trends? Can we say something about the uncertainty in the projections? Can we say something about the degree of societal response required to reduce the uncertainty associated with the impacts of such an event?

The moral of this story is that for our extreme events workshop we might justifiably and profitably avoid the question of "what is an extreme event?" (answer: it depends). Any successful effort to conceptualize "extreme events" as a researchable issue will rest on an explicit awareness of the context–dependence of both "extremeness" and "eventness."

[ Top of Page ]