The problem of what to expect when you are expecting regional change -- Different evaluation strategies of regional prediction and projection performance in the NCPP framework

Ensembles of climate model experiments together with means and trends in instrumental records generally build the basis for evaluation of predictions and projections of regional climate change. Most are drawn from mean climatological changes and trends, and some describe how changes in modes of variability are simulated. But how good are these regional and/or mode changes from models? It is clear that at the regional scale also internal variability needs to be taken into account. This is the case for identifying the forced changes in the real world, but it is also critical when evaluating a model-based prediction or projection. The key question is about what part of the observed variability and change we can, and should, expect models to reproduce. This problem is not trivial and requires a host of conditional considerations covering different time scales. Next to the instrumental reference observations, even paleoclimatic information from well-dated and verified reconstructions can be of use for important elements of this evaluation, including a more complete representation of the full range of variability as well as potential information of systematic structural response in the climate system to radiative perturbations. This presentation provides an overview of how the National Climate Predictions and Projections platform is currently developing a catalog of strategies to evaluate performance in regional climate outlooks across seasonal to decadal and centennial time scales, and how new research can enrich and extend the tools for a scientifically sound evaluation of what to expect when one is expecting regional climate change.