This talk has been automatically captioned. You can remove these by pressing CC on the video toolbar.

Name: Svenja Szemkus

Talk Title: Extremal dependence as given by the tail pairwise dependence matrix in precipitation and temperature data

Abstract: A better understanding of the dynamics and impacts of extreme weather events and their changes due to climate change is the subject of the ClimXtreme project (climxtreme.net) funded by the German Federal Ministry of Education and Research. The CoDEx project is investigating how data compression techniques can contribute to a better description and understanding of extremes. Various unsupervised learning approaches, such as clustering or principal component analysis, focusing on extremes have been developed recently. We use principal component analysis to study the spatial (co-)occurrence during extreme weather events such as heavy precipitation, heat waves or droughts. The focus on extreme events is done by using the tail pairwise dependence matrix (TPDM), proposed by Cooley and Thibaud (2019) as an analogue to the covariance matrix for extremes. Since the simultaneous occurrence of precipitation deficits and high temperature played an important role, especially in heat waves, we explore how Cooley and Thibaut's concept can be used in this regard. We propose an estimation of the TPDM based on pairwise dependencies of two variables. A singular value decomposition gives us insight into the spatial co-occurrence of extreme spatial patterns, contributing to the understanding of so-called compound events. We use daily precipitation and temperature data both at observation stations and in regional reanalyses in Germany and Europe, extract spatial patterns, and investigate historical events using this method.

 This talk is a contributed talk at EVA 2021.