Probabilistic representations of regional climate change in the CESM Large Ensemble

Because of the irreducible uncertainty of internal variability, regional climate change is inherently probabilistic on decadal timescales. This characteristic of regional climate change presents challenges both in the communication of climate change uncertainty and in the need for large initial-condition ensembles that accurately simulate observed natural variability across a range of timescales. In this study we use the new 30-member Community Earth System (CESM) Large Ensemble experiment for the period of 1920-2080 to explore the role of internal variability on probabilistic regional climate change on interdecadal timescales. Focusing on surface air temperature and precipitation, we present a means of expressing and visualizing probabilistic climate change projections as probabilities of threshold trend exceedance that vary with the trend length. We then evaluate the reliability of the CESM Large Ensemble for probabilistic climate change projections over the period of the observational record, which reflects the model's adequacy in simulating both the forced changes and internal variability. Finally, we compare the results of the CESM Large ensemble with that of a multi-model ensemble from the CMIP5 archive to highlight the roles of model and scenario uncertainty across a range of timescales.