A Spatial Evaluation of Arctic Sea Ice in CMIP6 Historical Simulations
This study examined the historical simulations of Arctic sea ice extent, thickness, and volume in state-of-the-art Earth System and global climate models participating in Phase 6 of Coupled Model Intercomparison Project (CMIP6). Biases in the spatial pattern distribution of simulated sea ice were identified in order to understand regional limitations which may guide future model improvement.
Large model spread in the standard sea ice metrics of extent and volume is persistent in model intercomparison projects. Here, spatial analysis metrics are used to diagnose individual model limitations, which sea ice temporal analyses alone do not capture.
Standard analysis of simulated sea ice in CMIP6 models reveals that large model spread is a persistent characteristic carried over from previous intercomparison projects. In order to guide future model improvements, spatial analysis metrics of sea ice are used to expose common and individual regional biases in global models. In this study, a subset of CMIP6 models are compared against satellite observations and results from sea ice reanalyses, like the Regional Arctic System Model. We find that the CMIP6 multi-model mean captures the pan-Arctic sea ice cover mean states and declining trends for 1979-2014 remarkably well. However, individual models experience a wide range of uncertainty in the simulated spatial distribution of sea ice thickness and concentration. Many of these spatial biases may be related to the oceanic forcing, e.g., ocean heat transport, particularly in the marginal and shelf sea regions.