A global climate model ensemble for downscaled monthly climate normals over North America

Use of downscaled global climate model projections is expanding rapidly as change vulnerability assessments and adaptation planning become mainstream in many sectors. Many impact analyses use at very high spatial resolution (~1 km) but low temporal (20- to 30-year normals). These applications have selection priorities that are distinct from resolution. Here, we select a 13-model ensemble an 8-model subset designed for robust change-factor downscaling monthly normals, describe their attributes North America. All models selected the Coupled Model Intercomparison Project Phase 6 (CMIP6) archives. The representative distribution equilibrium sensitivity, grid resolution, transient regional changes CMIP6 generation. consistent with IPCC's recent assessment likely range Earth's sensitivity. Our results emphasize several principles ensembles: (a) must be observationally constrained meaningful; (b) analysis multiple essential mean alone can misleading; (c) small (<8-member) ensembles should region-specific used caution; (d) higher not necessarily better; (e) simulations each model/scenario combination necessary represent precipitation uncertainty. Although focused our documentation on America, uses primarily criteria applicable normals other continents. Downscaled available ClimateNA (http://climatena.ca/). An accompanying web application (https://bcgov-env.shinyapps.io/cmip6-NA/) provides tools further visualization ensemble.