Evaluating CMIP6 model performance helps to improve the prediction of future changes in Arctic sea ice. We analyze the seasonal cycles, distribution, and evolution of sea ice in different regions from 1979 to 2014. We compare the output from selected CMIP6 models with reference data for sea ice motion. We also discuss the correlations between sea ice motion(SIM) and sea ice thickness (SIT) in reference data, and how CMIP6 models explain them. We select EC-Earth3, ACCESS-CM2, BCC-CSM2-MR, MPI-ESM1-2-HR, and NorESM2-LM for CMIP6 study. We compare outputs with reference data: Sea ice extent (SIE) from NSIDC; SIT from PIOMAS; and SIM from the IABP buoy data. Analytical techniques include Theil-Sen and Ordinary least squares (OLS) regression. Most selected CMIP6 models have seasonal cycles of SIM lagging behind IABP observations by 1-2 month and overestimate central Arctic SIM magnitude, with MPI-ESM1-2-HR having the highest discrepancy and NorESM2-LM lowest. The models show better simulation of SIM in the ice melting season than in the growing season. Models perform worse at capturing regional differences in SIM evolution and are overly conservative when simulating the increasing trend in ice motion, especially in coastal Arctic seas during summer. There is significant negative correlation between SIT and SIM in October.

Processes driving changes in sea-ice seasonality and sea-ice thickness were explored for a ‘warm-shelf’ region along the West Antarctic Peninsula using vertically coupled sea-ice-ocean thermodynamic simulations, with and without assimilated satellite sea-ice observations and moored ocean temperature observations. Simulations with assimilated sea-ice observations permitted investigation of surface [thermodynamic and dynamic (e.g., wind-driven)] processes affecting sea-ice thickness and seasonality. Assimilation of quasi-weekly variability in the depth and temperature of the deep warm pycnocline permitted examination of subsurface processes affecting sea-ice. Simulations using assimilated sea-ice observations (and implied motion) always produced greater surface heat fluxes and overall thinner sea ice. Assimilating seasonal and quasi-weekly variability in the depth and temperature of the pycnocline modified the start of the sea-ice season by −23 to +1 d, and also modified the sea ice thickness/seasonality to be thinner/shorter or thicker/longer at sub-seasonal and seasonal timescales, highlighting a mechanism where a shoaling pycnocline enhanced upward deep-water heat fluxes as transient surface-induced turbulence had a greater effect on a reduced mixed layer volume. The observed interplay of surface, subsurface, and sea-ice modulation of ocean-atmosphere heat transfer underscores the importance of representing the interaction between sea-ice concentration and upper ocean variability in climate projections.

Southern Ocean sea ice can exhibit widespread flooding and subsequent snow-ice formation, due to relatively thick snow covers compared to the total ice thickness. Considerable subkilometer scale variability in snow and ice thickness causes poorly constrained uncertainties in determining the amount of flooding that occurs. Using datasets of snow depth and ice thickness acquired in the Weddell Sea during austral winter 2013 (AWECS campaign) from three floes, we demonstrate large spatial variability of a factor 10 and 5 for snow and combined snow and ice thickness, respectively. The temporal evolution after the floe visit was recorded by automatic weather station and ice mass balance buoys. Using a physics-based, multi-layer snow/sea ice model in a one-dimensional and distributed mode to simulate the thermodynamic processes, we show that the distributed simulations, modeling flooding across the entire heterogeneous floe, produced vastly different amounts of flooding than one-dimensional single point simulations. Three times the flooding is produced in the one-dimensional simulation for the buoy location than distributed (floe-averaged) simulations. The latter is in close agreement with buoy observations. The results suggest that using point observations or one-dimensional simulations to extrapolate processes on the floe-scale can overestimate the amount of flooding and snow-ice formation.

The inaugural data from the first systematic program of sea-ice observations in Kotzebue Sound, Alaska, in 2018 coincided with the first winter in living memory when the Sound was not choked with ice. The following winter of 2018–19 was even warmer and characterized by even less ice. Here we discuss the mass balance of landfast ice near Kotzebue (Qikiqtaġruk) during these two anomalously warm winters. We use in situ observations and a 1-D thermodynamic model to address three research questions developed in partnership with an Indigenous Advisory Council. In doing so, we improve our understanding of connections between landfast ice mass balance, marine mammals and subsistence hunting. Specifically, we show: (i) ice growth stopped unusually early due to strong vertical ocean heat flux, which also likely contributed to early start to bearded seal hunting; (ii) unusually thin ice contributed to widespread surface flooding. The associated snow ice formation partly offset the reduced ice growth, but the flooding likely had a negative impact on ringed seal habitat; (iii) sea ice near Kotzebue during the winters of 2017–18 and 2018–19 was likely the thinnest since at least 1945, driven by a combination of warm air temperatures and a persistent ocean heat flux.

Clouds play an important role in the Arctic surface radiative budget, impacting the seasonal evolution of Arctic sea-ice cover. We explore the large-scale impacts of springtime and early summer (March through July) cloud and radiative fluxes on sea ice by comparing these fluxes to seasonal ice volume losses over the central Arctic basin, calculated for available observational years 2004–2007 (ICESat) and 2011–2017 (CryoSat-2). We also supplement observation data with sea-ice volume computed from the Pan-Arctic Ice–Ocean Modeling and Assimilation System (PIOMAS) during summer months. We find that the volume of sea ice lost over the melt season is most closely related to observed downwelling longwave radiation in March and early summer (June and July) longwave cloud radiative forcing, which together explain a large fraction of interannual variability in seasonal sea-ice volume loss (R2 = 0.71, p = 0.007). We show that downwelling longwave fluxes likely impact the timing of melt onset near the sea-ice edge, and can limit the magnitude of ice thickening from March to April. Radiative fluxes in June and July are likely critical to seasonal volume loss because modeled data show the greatest ice volume reductions occur during these months.

In the marginal ice zone, surface waves drive motion of sea ice floes. The motion of floes relative to each other can cause periodic collisions, and drives the formation of pancake sea ice. Additionally, the motion of floes relative to the water results in turbulence generation at the interface between the ice and ocean below. These are important processes for the formation and growth of pancakes, and likely contribute to wave energy loss. Models and laboratory studies have been used to describe these motions, but there have been no in situ observations of relative ice velocities in a natural wave field. Here, we use shipboard stereo video to measure wave motion and relative motion of pancake floes simultaneously. The relative velocities of pancake floes are typically small compared to wave orbital motion (i.e. floes mostly follow the wave orbits). We find that relative velocities are well-captured by existing phase-resolved models, and are only somewhat over-estimated by using bulk wave parameters. Under the conditions observed, estimates of wave energy loss from ice–ocean turbulence are much larger than from pancake collisions. Increased relative pancake floe velocities in steeper wave fields may then result in more wave attenuation by increasing ice–ocean shear.

In an effort to improve the reliability of Arctic sea-ice predictions, an ensemble-based Arctic Ice Ocean Prediction System (ArcIOPS) has been developed to meet operational demands. The system is based on a regional Arctic configuration of the Massachusetts Institute of Technology general circulation model. A localized error subspace transform ensemble Kalman filter is used to assimilate the weekly merged CryoSat-2 and Soil Moisture and Ocean Salinity sea-ice thickness data together with the daily Advanced Microwave Scanning Radiometer 2 (AMSR2) sea-ice concentration data. The weather forecasts from the Global Forecast System of the National Centers for Environmental Prediction drive the sea ice–ocean coupled model. The ensemble mean sea-ice forecasts were used to facilitate the Chinese National Arctic Research Expedition in summer 2017. The forecasted sea-ice concentration is evaluated against AMSR2 and Special Sensor Microwave Imager/Sounder sea-ice concentration data. The forecasted sea-ice thickness is compared to the in-situ observations and the Pan-Arctic Ice-Ocean Modeling and Assimilation System. These comparisons show the promising potential of ArcIOPS for operational Arctic sea-ice forecasts. Nevertheless, the forecast bias in the Beaufort Sea calls for a delicate parameter calibration and a better design of the assimilation system.