Abstract The polar regions are undergoing profound changes, but climate models still struggle to accurately represent key polar climate processes. Exploring how increased horizontal resolution affects model biases in the polar regions and lower latitudes, we performed a coordinated set of global model simulations for the recent past with grid refinement in the polar regions, with the finest resolution implemented poleward of 60°. We employed three state‐of‐the‐art models: the Community Earth System Model, the Icosahedral Nonhydrostatic model, and the Model for Prediction Across Scales. The result is the first coordinated multimodel ensemble with variable‐resolution models with refinement over the polar regions. This work documents the common experiment protocol and presents the first results, focusing on temperature, zonal wind, and turbulent surface fluxes. The experiments were configured with fully coupled atmosphere and land components and prescribed sea‐surface temperatures and sea‐ice concentrations. Forcings for greenhouse gases and aerosols were based on the historical experiment from the sixth phase of the Coupled Model Intercomparison Project. We find that the influence of the refinement, assessed by comparing simulations with refined and uniform resolution, can vary considerably across models, hemispheres, and seasons. The most consistent improvements occur in wintertime upper‐tropospheric high‐latitude temperature, leading to improved Equator‐to‐Pole temperature contrasts and zonal wind in both hemispheres. Though all models underestimate the Arctic winter near‐surface temperature compared with the fifth generation of the European Center for Medium‐Rrange Forecasts Reanalysis, validation against Arctic and Antarctic in‐situ measurements reveals modest improvements in the refined experiments, with the measured values being better represented in the Arctic compared with the Antarctic and in summer compared with winter.
Graff et al. (Mon,) studied this question.