Assessing CMIP6 simulations of Arctic sea ice drift: Role of near-surface wind and surface ocean current in model performance

• Published in: Advances in Climate Change Research
• Main Authors: Xue Wang, Ran Lu, Shao-Yin Wang, Run-Tong Chen, Zhuo-Qi Chen, Feng-Ming Hui, Hua-Bing Huang, Xiao Cheng
• Format: Article in Journal/Newspaper
• Language: English
• Published: KeAi Communications Co., Ltd. 2023
• Subjects: Arctic sea ice drift; CMIP6; Model evaluation; Near-surface wind; Surface ocean current; Meteorology. Climatology; QC851-999; Social sciences (General); H1-99; Arctic; Sea ice
• Online Access: https://doi.org/10.1016/j.accre.2023.09.005; https://doaj.org/article/46e89d520e95416ead25f57c02e17e2f

Sea ice drift is a critical parameter for understanding the rapid changes in Arctic sea ice. Since the release of the Coupled Model Intercomparison Project Phase 6 (CMIP6), there has been a lack of quantitative analysis regarding CMIP6's simulation of Arctic sea ice drift. This study aims to assess the simulated Arctic sea ice drift from 1979 to 2014 by fifteen CMIP6 models against recent satellite retrievals, utilizing various quantitative indices. Additionally, the influence of near-surface wind and surface ocean current on model performance is further analyzed. The CMIP6 models capture several aspects of the observed Arctic sea ice drift climatology and variability. The seasonal patterns of sea ice drift speed in all models exhibit similarities with the observed data, and the models agree with the evaluation datasets, indicating that the seasonal evolution of sea ice drift corresponds to near-surface wind patterns. However, notable discrepancies are identified. All models overestimate sea ice drift speed, exceeding the observational data by 36%–97%. Fourteen out of fifteen models display larger seasonal variability (ranging from 0.74 to 1.28 km d−1) compared to the observed data (0.54 km d−1). Seven out of fifteen models exhibit a significant increasing trend in annual sea ice drift speed, similar to the observed trend of 0.58 km d−1 per decade, but with weaker trends (ranging from 0.11 to 0.33 km d−1 per decade). The remaining eight models reveal no statistically significant trend. The potential causes of such biases were further explored in this study. It suggests that the overestimation of sea ice drift speed in the models might be primarily attributed to the overestimation of near-surface wind speeds and their influence on sea ice drift speed. The models' overestimation of seasonal variability in near-surface wind speeds may account for the overestimation of seasonal variability in sea ice drift. The models' inability to represent the trend in sea ice drift speed may result from their failure to simulate ...