Abstract. Compound events (CE), characterized by the combination of climate phenomena that are not necessarily extreme individually, can result in severe impacts when they occur concurrently or sequentially. Understanding past and potential future changes in their occurrence is thus crucial. The present study investigates historical changes in the probability of hot and dry compound events over Europe and North Africa, using ERA5 reanalyses spanning the 1950–2023 period. Two key questions are addressed: (1) Where and when did the probability of these events emerge from natural variability, and what is the spatial extent of this emergence? This is explored through the analysis of “time” and “periods” of emergence, noted ToE and PoE, defined as the year from which and the moments during which changes in compound event probabilities exceed natural variability. The new concept of PoE allows for more in-depth signal analysis. (2) What drives the emergence? More specifically, what are the relative contributions of changes in marginal distributions versus in the dependence structure to the change of compound events probability? The signal is modelled with bivariate copula, allowing for the decomposition of these contributions. A focus on the dependence component is explored to quantify its effect on the signal's emergence. The results reveal clear spatial patterns in terms of emergence and contributions. Five areas are studied in greater depth, selected for their contrasted signal behaviors. In some regions, the frequency of hot and dry events increased, mainly due to a change in the marginals. However, other regions see a decrease of CE probabilities, mainly driven by a change in the drought index. Although the dependence component is rarely the main contributor to PoE, it remains necessary to detect signal's emergence. Without considering the dependence component, the date of ToE and the duration of PoE can be overestimated as well as underestimated (even more than 20 years) depending on the area. These findings provide new insights into the drivers of CE probability changes and open avenues for advancing attribution studies, ultimately improving assessments of risks associated with past and future climate change.
Schmutz et al. (2026) studied this question.