EVA, Statistics That Help Understand Extreme Events and Climate Change
A devastating flood, a long period of drought or a day with temperatures above 40 degrees Celsius can be considered as extreme values of the empirical probability distribution of collected data, such as daily rainfall, the number of consecutive days without precipitation or the temperatures in a city.
Some studies point out that these types of events are relatively more frequent in recent years. This finding provides further evidence that a process of climate change is taking place. The observations generated by these events are extremely large compared with most of the data recorded in the past, and their impact on the environment and society can be terribly negative. This makes the study of these phenomena fall into a field of Statistics called extreme value analysis. It should therefore come as no surprise that the guest speakers at EVA 2023, the 13th Extreme Value Analysis Conference (Bocconi University, June 26-30, 2023) are not statisticians but specialists in climate change.
"As statisticians involved in extreme value analysis," explains Simone Padoan, Associate Professor in the Department of Decision Sciences and conference organizer, "we are able to produce complex models and sophisticated statistical analyses, but we want to understand the needs of those people who use these results so that we can maximize the impact of our work. That's why Jaroslav Mysiak from the Euro-Mediterranean Center on Climate Change, and Valerio Lucarini from the University of Reading have been invited."
At least 260 scientific presentations are expected at the conference, with the participation of about 300 mathematicians and statisticians actively researching in this area. Though extreme value analysis may look like a narrow field, its devotees are numerous and its applications manifold. The conference, now held every two years, is at its 13th edition and celebrates the 40th year since its founding (the first, in 1983, was held in Portugal). The scientific output is so rich that it has led to the publication of a specialized journal, Extremes, on whose editorial board Padoan sits.
"The area is intellectually fascinating because modeling very rare events is dauntingly complex," Padoan says, "and even when you get to a model, the scarcity of data, due precisely to the rarity of the event, also requires considerable statistical processing effort in order to produce accurate estimates."
Although the application at the center of attention this year is climate change, there are several phenomena that can present extreme values with significant consequences, to say the least. "First of all," Padoan says, "finance. A stockmarket crash is the typical example of an extreme value that is hard, but certainly useful, to predict."