|Title||Interpreting results from the NARCCAP and NA-CORDEX ensembles in the context of uncertainty in regional climate change projections|
|Publication Type||Journal Article|
|Year of Publication||2018|
|Journal||Bulletin of the American Meteorological Society|
|Keywords||interpretation, NA-CORDEX, NARCCAP, Regional climate change Projections|
Ensembles of dynamically downscaled simulations provide valuable information on regional climate change projections, but their interpretation remains challenging due to complexities in the experimental design.
Two ensembles of dynamically downscaled climate simulations for North America - NARCCAP and NA-CORDEX - are analyzed to assess the impact of using a small set of global (GCMs) and regional climate models (RCMs) on representing uncertainty in regional projections. Selecting GCMs for downscaling based on their equilibrium climate sensitivities is a reasonable strategy, but there are regions where the uncertainty is not fully captured. For instance, the six NA-CORDEX GCMs fail to span the full CMIP5 ranges in summer temperature projections in the western and winter precipitation projections in the eastern U.S. Similarly, the four NARCCAP GCMs are overall poor at spanning the full CMIP3 ranges in seasonal temperatures. For the Southeast, the NA-CORDEX GCMs capture the uncertainty in summer but not in winter projections, highlighting one consequence of downscaling a subset of GCMs. Ranges produced by the RCMs are often wider than their driving GCMs but are sensitive to the experimental design. For example, the downscaled projections of summer precipitation are of opposite polarity in two RCM ensembles in some regions. Additionally, the ability of the RCMs to simulate observed temperature trends is affected by the internal variability characteristics of both the RCMs and driving GCMs, and is not systematically related to their historical performance. This has implications for adequately sampling the impact of internal variability on regional trends and for using model performance to identify credible projections. These findings suggest that a multi-model perspective on uncertainties in regional projections is integral to the interpretation of RCM results.
|Short Title||Bull. Amer. Meteor. Soc.|