Investigating heatwaves/droughts and convective precipitation extremes using compact descriptions of spatio-temporal fields

The detection of local climate change signals, in particular those related to extreme events, is challenging due to the large internal variability of the climate system. The BMBF-funded project ClimXtreme Module B-CoDEx focuses on improving the signal-to-noise ratio of climate change signals in extreme weather events using innovative data compression methods.

This study uses principal component analysis (PCA) for spatial extremes (Cooley and Thibaud, 2019) to analyse heatwaves and droughts over the northern hemisphere. An extremal pattern index (EPI) is introduced as an integrative measure of the intensity and spatial extent of an extreme heat anomaly. Its bivariate extension is used to account for simultaneous spatial extremes in two variables. EPI provides us with a compact description of heatwaves. We see, for example, that preceding precipitation deficits significantly influence the development of heatwaves, and that heat waves often coincide with instantaneous short-term droughts. 

To investigate extreme hourly precipitation, a scale-dependent decomposition using the dual-tree wavelet transform is proposed, as described e.g. in Buschow and Friederichs (2021). For this study, we rely on reanalysis data (COSMO-REA6, CERRA) over Germany. A comparison of the two datasets regarding their representation of large- and small-scale events shows significant differences, especially for the small-scale events. Furthermore, we apply established methods to perform a scale-dependent detection of extreme precipitation and to reveal trends that are hidden by variability on other scales.