There are currently many well-established climate attribution methodologies for single driver extremes, however, there is very little literature on more complex extremes such as compound, sequences and cascading hazard events. The main goal of the COMPASS project is to develop a harmonized methodological framework for climate and impact attribution of such complex extremes.
Complex extremes require analysis of multiple variables and hence are reliant on a methodology that can be adapted to take more than one variable into account and are able to integrate sufficiently large datasets to robustly capture enough of the multi-variate extremes. We assessed and identified several methods that have potential to meet our requirements, and we provide guidance on the methodologies that we think are likely to be easiest to adapt for complex extremes.

Key findings include:

• It is important to understand the attribution question i.e. ‘the framing’ the Use Case is seeking to answer before choosing the approach. Different approaches provide different climate information and can on occasion produce different results for the same event.
• Approaches range across a spectrum of conditionality from unconditional to highly conditioned (i.e. Storylines). Each approach provides useful information, which can be combined in a complementary fashion so as to provide a more complete picture of the influence of climate change on the event in question.
• Based off current literature we find that attribution methods using large ensembles are most suited for compound attribution, given the requirement for extra data in complex events that contain multiple extremes.

In addition to the recently published Deliverable 2.1 “Report on dataset on best-available methods and climate-datasets”, the team has also prepared the Milestone 3 “Report on methods assessment” that aims to explore the assessment of large ensembles for compound event attribution in more detail.

As an example, in Use Case 2b we have started to explore:

• Using a semi-conditional attribution approach to attribute the 2022 UK Drought Heatwave.
• Focusing on the large ensembles from HadGEM3-A, which include 525 ensemble members for both the current and a counterfactual climate.
• Carried out model validation tests on the variables of interest (precipitation and temperature) for England.
• Finding that the correlation between the variables are consistent between the observations and the model, making them suitable for this study.

In summary, there is an ever-growing increase in large ensembles available for the attribution of compound extremes, with the choice of ensemble and their suitability strongly dependent on the location of study and the variables of interest.

More information can be found in the following COMPASS project Deliverable: Cotterill et al., 2024: Report on dataset on best-available methods and climate datasets. Horizon Europe project COMPASS. Deliverable D2.1.