Emergence of compound events: quantifying the importance of marginal and dependence properties changes

Many climate-related disasters often result from a combination of several climate drivers, also referred to as "compound events''. By interacting with each other, these hazards can lead to huge environmental and societal impacts, at a scale potentially far greater than any of these climate drivers could have caused separately. Marginal and dependence properties of climate drivers, as well as their changes over time, are key statistical properties influencing the probabilities of compound events. A better understanding of how the statistical properties of variables leading to compound events evolve and contribute to the change of their occurences is a crucial step towards risk assessments. Here, based on copula theory, we develop a new methodology to quantify the contribution of marginal and dependence properties to the overall probability of compound events. For illustration purposes, the methodology is applied to analyse changes of probability for compound precipitation and wind extremes, and their potential time of emergence, in a 13-member multi-model ensemble (CMIP6) over the region of Brittany (France). Results show that compound precipitation and wind extremes probabilities from CMIP6 ensembles mostly increase for the end of the 21st century. Yet, the contribution of marginal and dependence properties to these changes of probabilities can be very different from one model to another, reflecting a large uncertainty in climate modelling. These results highlight the importance of both marginal and dependence properties changes for future risk assessments due to compound events, and the need to understand the differences' sources of statistical properties between climate models.