Some of the most impactful climate and weather events result from compounding drivers. To robustly assess the current and future risk from such compound events, a better understanding of the associated sources of uncertainty is needed. Internal variability confounds detection and attribution of human-induced climate change and imposes irreducible limits on the accuracy of climate projections. Response uncertainty can lead to divergent projections for many societally important quantities such as precipitation. Combined with unknown future greenhouse gas emissions, these uncertainties can result in a socio-economically paralyzing range of future storylines. Climate model large ensembles are uniquely positioned to assess these uncertainties and are rightfully gaining popularity in compound event research, but they need to be accompanied by rigorous model validation and robust observational constraints to reach their full potential in terms of usefulness for practitioners. This perspective discusses these opportunities and challenges at the example of water resources and provides an outlook on application-oriented compound event research with large ensembles.
