Evaluation and projection of Arctic sea ice transport through Fram Strait with emphasis on atmospheric multimodal responses

Abstract

This study evaluates the historical (1988–2014) sea ice drift (SID), sea ice concentration (SIC), sea ice thickness (SIT), and associated sea ice volume transport (SIVT) from 12 CMIP6 models in Fram Strait based on multi-source reference datasets. In addition, a global ice-ocean coupled model based on the Finite Volume Community Ocean Model (Global-FVCOM), is included as a comparison. Both the simulated and reference SIVT show significant interannual variability, but inter-model spread exists. Compared to reference data, the CMIP6 models exhibit greater discrepancies than Global-FVCOM. The comprehensive assessment reveals notable performance differences among the models in simulating varying sea ice variables. Ensemble experiments, incorporating models with varying performance in simulating SIVT, demonstrate a decreasing trend in SIVT across all schemes under the SSP2–4.5 and SSP5–8.5 scenarios, with a more pronounced decline in the latter. Compared to the traditional multi-model mean scheme using all models, the scheme selecting six better-performing models through evaluation results of SIVT exhibits lower SIVT values and a slower decline. There are significant differences among the CMIP6 models regarding the dominant factors contributing to future SIVT variations. The largest number of models demonstrates a strong relationship between the dipole anomaly (DA) and sea ice transport, followed by those associated with the Arctic Oscillation (AO), while the fewest models show a significant connection to the Beaufort-Barents Oscillation (BBO). The choice of time period, whether by year or month, considerably influences the identification of dominant modes for sea ice transport.