Is newer better? How can we tell if CORDEX6 improves our understanding of regional climate change?

The development of new generations of climate projections is a core part of climate science.  But while new datasets are always welcome, transitioning from one dataset to another raises many questions amongst the users of climate service: Is the new dataset different? Is it better? And how do we know? Many European climate services are currently facing these questions as the next generation of Euro-CORDEX data nears publication and uptake into climate services. Here we demonstrate a quantitative approach for assessing the differences between different generations of climate projections in a way that is relevant to the users of a climate service. We build our analysis on the use of global warming levels, an approach that can correct for the particularly problematic difference between the RCP and SSP emissions-scenarios used in the two ensembles. We provide a practical illustration of the method using approximately 40 annual and seasonal bias-adjusted indicators of climate change from Klimaatlas, the Danish Climate Atlas, calculated for 98 Danish municipalities in a standardised manner for both CORDEX5 and CORDEX6. We apply a Bayesian hierarchical testing framework to detect changes in both the response of an indicator and the associated uncertainty. We highlight cases where CORDEX6 reduces uncertainty, but also where it is indistinguishable from CORDEX5. Finally, we discuss how each of these cases can be handled in a climate service, and the way that the two ensembles will be merged in a future release of Klimaatlas to produce a climate service using both datasets to their fullest extent.