Extreme events in biological, societal, and earth sciences: A systematic review of the literature

Abstract

The term “extreme event” is commonly used to describe high-impact, unanticipated natural events, like floods, tsunamis, earthquakes, and volcanic eruptions. It first appeared in the scientific literature in the 1950s and has since spread to disciplines as diverse as economics, psychology, medicine, and engineering. The term is increasingly being applied to the study of historical, prehistorical, and deep-time events across a broad range of scales, and it is widely acknowledged that such events have had profound impacts on the Earth’s biodiversity and cultures. Understandably, then, how people think about, define, and study extreme events varies considerably. With extreme events expected to become more frequent, longer lasting, and more intense in the coming decades as a result of global warming, the differing extreme event definitions—both across and within disciplines—is likely to lead to confusion among researchers and pose significant challenges for predicting and preparing for extreme events and their impacts on natural and social systems. With this in mind, we conducted a systematic quantitative review of 200 randomly selected, peer-reviewed “extreme event” research papers (sourced from Web of Science, accessed January 2020) from the biological, societal, and earth sciences literature with the aim of quantifying several pertinent features of the research sample. On the one hand, our analysis found a great deal of variability among extreme event papers with respect to research interests, themes, concepts, and definitions. On the other hand, we found a number of key similarities in how researchers think about and study extreme events. One similarity we encountered was that researchers tend to view extreme events within a particular temporal context and quite often in terms of rates of change. Another similarity we encountered was that researchers often think of and study extreme events in terms of risks, vulnerabilities, and impacts. The similarities identified here may be useful in developing a common and comprehensive definition of what constitutes an extreme event, and should allow for more comparative research into extreme events at all spatio-temporal scales which, we predict, will provide important new insights into the nature of extreme events.

The intervals between zero-crossings of non-Gaussian stable random processes

“…Figure 5 illustrates sections of realizations of the Y process used to generate a continuous Cauchy process Z = X/Y (the realization of the X process is not shown). The blue curve is for a Gaussian process with auto-correlation function given by exp(−(kt) 2 ) and this has five zerocrossings in the section shown. The blue curve in the lower panel shows the Cauchy process that results, and it can be seen that this realization has five pole-discontinuities corresponding to where Y vanishes and where the sign of Z changes on either side of the discontinuities.…”

“…This is because of the reciprocal weighting of X by Y which makes the intervals between successive crossings of Z greater, on average, than those of X. Multiple long realizations of Cauchy processes with n = 1024 elements are generated with X and Y having the same auto-correlation function ρ(t) = exp(−(kt) 2 ) with k values selected such that k = ln(2)/t 1/2 with t 1/2 = 2 2 , 2 3 , . .…”

“…Identifying the statistical properties for the frequency of specified events associated with a random process and the times between them, such as extremal events, is a canonical problem of statistical science with diverse applications to physical, engineering and financial phenomena, among others [1,2]. Understanding principal properties of these events, which are derived from an underlying random process, can be subsumed to analysing its zero-or level-crossings, or those of its derivative.…”

Predictability of decay events in transitional wall-bounded flows

“…First, they are linked to high amplitude phenomena, which dominate the behaviour of a dynamical system for a certain time span. Second, they take place at infrequent times [1]. They are mostly studied in the atmospheric sciences, where the predictability of events such as tornadoes or hurricanes is of obvious interest.…”

“…Extreme events are also found in simpler dynamical/complex systems [1]. A notable example is turbulence in shear flows, where extreme events have been linked with the self-sustaining process of turbulence in shear flows [3], or to sudden increases of the shear at the wall [4].…”