Uncertainty in modeled Arctic sea ice volume

Abstract. Sea ice is a key component of the Arctic climate system, and has impacts on global climate. Ice concentration, thickness, and volume are among the most important Arctic sea ice parameters. This study presents a new record of Arctic sea ice thickness and volume from 1984 to 2018 based on an existing satellite-derived ice age product. The relationship between ice age and ice thickness is first established for every month based on collocated ice age and ice thickness from submarine sonar data (1984–2000) and ICESat (2003–2008) and an empirical ice growth model. Based on this relationship, ice thickness is derived for the entire time period from the weekly ice age product, and the Arctic monthly sea ice volume is then calculated. The ice-age-based thickness and volume show good agreement in terms of bias and root-mean-square error with submarine, ICESat, and CryoSat-2 ice thickness, as well as ICESat and CryoSat-2 ice volume, in February–March and October–November. More detailed comparisons with independent data from Envisat for 2003 to 2010 and CryoSat-2 from CPOM, AWI, and NASA GSFC (Goddard Space Flight Center) for 2011 to 2018 show low bias in ice-age-based thickness. The ratios of the ice volume uncertainties to the mean range from 21 % to 29 %. Analysis of the derived data shows that the ice-age-based sea ice volume exhibits a decreasing trend of −411 km3 yr−1 from 1984 to 2018, stronger than the trends from other datasets. Of the factors affecting the sea ice volume trends, changes in sea ice thickness contribute more than changes in sea ice area, with a contribution of at least 80 % from changes in sea ice thickness from November to May and nearly 50 % in August and September, while less than 30 % is from changes in sea ice area in all months.

In this study, six Arctic sea ice thickness products are compared: the AVHRR Polar Pathfinder-extended (APP-x), ICESat, CryoSat-2, SMOS, NASA IceBridge aircraft flights, and the Pan-Arctic Ice Ocean Modeling and Assimilation System (PIOMAS). The satellite products are based on three different retrieval methods: an energy budget approach, measurements of ice freeboard, and the relationship between passive microwave brightness temperatures and thin ice thickness. Inter-comparisons are done for the periods of overlap from 2003 to 2013. Results show that ICESat sea ice is thicker than APP-x and PIOMAS overall, particularly along the north coast of Greenland and Canadian Archipelago. The relative differences of APP-x and PIOMAS with ICESat are −0.48 m and −0.31 m, respectively. APP-x underestimates thickness relative to CryoSat-2, with a mean difference of −0.19 m. The biases for APP-x, PIOMAS, and CryoSat-2 relative to IceBridge thicknesses are 0.18 m, 0.18 m, and 0.29 m. The mean difference between SMOS and CryoSat-2 for 0~1 m thick ice is 0.13 m in March and −0.24 m in October. All satellite-retrieved ice thickness products and PIOMAS overestimate the thickness of thin ice (1 m or less) compared to IceBridge for which SMOS has the smallest bias (0.26 m). The spatial correlation between the datasets indicates that APP-x and PIOMAS are the most similar, followed by APP-x and CryoSat-2.

Numerical sea ice forecasting products during the ninth Chinese National Arctic Research Expedition (CHINARE- Arctic) from Arctic Ice Ocean Prediction System (ArcIOPS) of National Marine Environmental Forecasting Center are evaluated against satellite-retrieved sea ice concentration data, in-situ sea ice thickness observations, and sea ice products from Pan-Arctic Ice Ocean Modeling and Assimilation System (PIOMAS). The results show that ArcIOPS forecasts reliable sea ice concentration and thickness evolution. Deviations of the 168 h sea ice concentration and thickness forecasts with respect to the observations are less than 0.2 and 0.36 m. Comparison between outputs of the latest version of ArcIOPS and that of its previous version shows that the latest version has a substantial improvement on sea ice concentration forecasts due to data assimilation of new observational component, the sea surface temperature. Meanwhile, the sea ice volume product of the latest version is more close to the PIOMAS product. In the future, with more and more kinds of observations to be assimilated, the high-resolution version of ArcIOPS will be put into operational running and benefit Chinese scientific and commercial activities in the Arctic Ocean.

Arctic sea ice plays a critical role in modulating our global climate system and the exchange of heat fluxes in the polar region, but its impact on climate varies across different sea ice thickness (SIT) categories. Compared to sea ice cover, the performance of ice models in simulating SIT has been less evaluated, particularly in the sixth Coupled Model Intercomparison Project Phase (CMIP6). Here, we chose 12 CMIP6 models with the Community Ice Code model (CICE) components and compared their SIT simulations with the satellite observations and the Pan-Arctic Ice Ocean Modeling and Assimilation System (PIOMAS) model between 1980 and 2014. Our results show that the seasonal cycle of the PIOMAS SIT is consistent with satellite observations. Compared to the PIOMAS reanalysis, the multi-model ensemble mean (MME) well represents the sea ice extent in both the thin ice (<0.6 m) and thick ice (> 3.6 m). However, the MME SIT has larger biases in the Chukchi Sea, the Beaufort Sea, the central Arctic, and the Greenland Sea during winter and mainly in the central Arctic during summer. Both the MME and PIOMAS show decreasing trends in SIT over the entire Arctic Ocean in all seasons, but the interannual variability of SIT in MME is smaller than that in PIOMAS. Among the 12 CMIP6 models, the FIO-ESM-2.0 model shows the best simulation of the annual mean SIT, but the SAM0-UNICON and NESM3 models have the largest biases in the climatological mean SIT over the Arctic Ocean. We also demonstrate that the FIO-ESM-2.0 performs the best in the seasonal cycles of SIT. Our study suggests that more attention should be paid to the coupling of the CICE model with ocean and atmosphere models, which is vital to improving the SIT simulation in CMIP6 models and to better understanding the impact of Arctic sea ice on our climate system.

Using ICESat‐2 and CryoSat‐2 freeboards, we examine the variability of monthly Arctic sea ice snow depth, thickness and volume between October 2018 and April 2021. For the 3 years, satellite‐derived estimates captured a decrease in mean April snow depth (∼2.50 cm) and ice thickness (∼0.28 m) equivalent to an ice volume loss of ∼12.5%. Results show greater thinning of multiyear ice with an end‐of‐season thickness in 2021 that is lower by ∼16.1% (0.50 m), with negligible changes over first‐year ice. For the period, sea ice thickness estimates using snow depth from climatology result in thicker ice (by up to ∼0.22 m) with a smaller decrease in multiyear ice thickness (∼0.38 m). An 18‐year satellite record, since the launch of ICESat, points to a loss of ∼6,000 km3 or one‐third of the winter Arctic ice volume driven by decline in multiyear‐ice coverage in the multi‐decadal transition to a largely seasonal ice cover.

In a 600-year control run simulation of the Community Climate System Model, version 3 (CCSM3) of the National Center for Atmospheric Research, three events of extraordinary minimum September sea ice extent are identified, from which the system recovers within a few years. The control simulation is run in a stable climate mode using constant 1990 solar, trace gas, and aerosol forcing. The identified events are all of similar magnitude to the observed 2007 record low Arctic ice cover. In the first event (simulation year 451), the record low ice extent coincides with an extreme, cyclonic phase of atmospheric circulation and follows a 10-year period of steady decline in Northern Hemisphere sea ice volume that is associated with increased winter ice export. The ice extent recovered the following year when both the ice export and summer melt returned to more normal values. However, Arctic ice volume did not fully recover until 20 years after the event. In the second event (model year 490), the system was again preconditioned by a 10-year decline in Arctic sea ice volume because of larger than normal ice export. The event itself was associated with a coincident peak in early summer melt and ice export. During the recovery phase, the ice extent again rapidly returned to normal while the sea ice volume continued to decline for another 10 years, to its lowest value of the entire simulation. The recovery in Northern Hemisphere sea ice volume is associated with periods of reduced wintertime ice export through Fram Strait. The third event (model year 556―558) is characterized by a longer-lived anomaly (3 years). The system was not preconditioned by low ice volume, indicating that redistribution of sea ice (rather than ice mass loss) played an important role. Recovery for the third event is associated with a period of low ice export associated with a rapid return of an expanded Beaufort Gyre, which serves to recirculate sea ice within the Arctic Ocean. Notwithstanding known climate model deficiencies, the control simulation suggests that Arctic sea ice extent anomalies of the magnitude observed in the September 2007 minimum are plausible under 1990 forcing conditions. It is reasonable to conclude that more severe anomalies are achievable with more acute forcing conditions that are in excess of those of 1990.

In an ocean–sea ice model of the Arctic and the northern North Atlantic driven with 50-yr NCEP–NCAR reanalysis data, no appreciable trend in sea ice volume is found for the period 1948–98. However, rather long subperiods, for example, 1965–95, exhibit a large decline in Arctic sea ice volume. These results and the current data situation make connecting “global warming” to Arctic ice thinning very difficult because the large decadal and multidecadal variability masks any trend. Thermal and wind effects linearly contribute to the total sea ice volume variability. Wind stress forcing significantly contributes to the decadal variability in the Arctic ice volume, affecting both thermodynamic growth and the ice export rate. Ice export events are triggered by enhanced cyclonic wind stress over the eastern Arctic. However, large ice export events depend to a similar degree on the presence of thick ice that is generated in a preceding accumulation phase and do not depend on the local wind conditions aroun...

Abstract. As the summer Arctic sea ice extent has retreated, the marginal ice zone (MIZ) has been widening. The MIZ is defined as the region of the ice cover that is influenced by waves and for convenience here is defined as the region of the ice cover between sea ice concentrations (SIC) of 15 % to 80 %. The MIZ is projected to become a larger percentage of the summer ice cover, as the Arctic transitions to ice-free summers. Using numerical simulations, we explicitly compare, for the first time, individual processes of ice volume gain and loss in the ice pack (SIC > 80 %) to those in the MIZ to establish and contrast their relative importance and examine how these processes change as the summer MIZ fraction increases over time. We use an atmosphere-forced, physics-rich, sea-ice-mixed layer model based on CICE, that includes a joint prognostic floe size and ice thickness distribution (FSTD) model including brittle fracture and form drag. We demonstrate that this model is realistic using satellite observations of sea ice extent and PIOMAS (the Pan-Arctic Ice Ocean Modeling and Assimilation System) estimates of thickness. A comparable setup has also been compared to floe size distribution (FSD) observations in prior studies. The MIZ fraction of the July sea ice cover, when the MIZ is at its maximum extent, increases by a factor of 2 to 3, from 14 % (20 %) in the 1980s to 46 % (50 %) in the 2010s in NCEP (HadGEM2-ES) atmosphere-forced simulations. In a HadGEM2-ES forced projection, the July sea ice cover is almost entirely MIZ (93 %) in the 2040s. Basal melting accounts for the largest proportion of melt in regions of pack ice and MIZ for all time periods. During the historical period, top melt is the next largest melt term in pack ice, but in the MIZ, top melt and lateral melt are comparable. This is due to a relative increase of lateral melting and a relative reduction of top melting by a factor of 2 in the MIZ compared to the pack ice. The volume fluxes due to dynamic processes decrease due to the reduction in ice volume in both the MIZ and pack ice. For areas of sea ice that transition to being MIZ in summer, we find an earlier melt season: in the region that was pack ice in the 1980s and became MIZ in the 2010s, the peak in the total melt volume flux occurs 20(12) d earlier. This continues in the projection where melting in the region that becomes MIZ in the 2040s shifts 14 d earlier compared to the 2010s. Our analysis shows that a different balance of processes controls the volume budget of the MIZ versus the pack ice. We also find that the balance of processes is different for the MIZ in the 2040s compared to the 1980s, and conclude that we cannot understand the disposition between basal, lateral and top melt in a future Arctic solely based on increased MIZ fraction, since changes in surface energy balance remain a strong control on these behaviours.

To better simulate the seasonal evolution of sea ice in the Arctic, with particular attention to the marginal ice zone, a sea ice model of the distribution of ice thickness, floe size, and enthalpy was implemented into the Pan-arctic Ice–Ocean Modeling and Assimilation System (PIOMAS). Theories on floe size distribution (FSD) and ice thickness distribution (ITD) were coupled in order to explicitly simulate multicategory FSD and ITD distributions simultaneously. The expanded PIOMAS was then used to estimate the seasonal evolution of the Arctic FSD in 2014 when FSD observations are available for model calibration and validation. Results indicate that the simulated FSD, commonly described equivalently as cumulative floe number distribution (CFND), generally follows a power law across space and time and agrees with the CFND observations derived from TerraSAR-X satellite images. The simulated power-law exponents also correlate with those derived using MODIS images, with a low mean bias of –2%. In the marginal ice zone, the modeled CFND shows a large number of small floes in winter because of stronger winds acting on thin, weak first-year ice in the ice edge region. In mid-spring and summer, the CFND resembles an upper truncated power law, with the largest floes mostly broken into smaller ones; however, the number of small floes is lower than in winter because floes of small sizes or first-year ice are easily melted away. In the ice pack interior there are fewer floes in late fall and winter than in summer because many of the floes are “welded” together into larger floes in freezing conditions, leading to a relatively flat CFND with low power-law exponents. The simulated mean floe size averaged over all ice-covered areas shows a clear annual cycle, large in winter and smaller in summer. However, there is no obvious annual cycle of mean floe size averaged over the marginal ice zone. The incorporation of FSD into PIOMAS results in reduced ice thickness, mainly in the marginal ice zone, which improves the simulation of ice extent and yields an earlier ice retreat.

Abstract. NASA's ICESat-2 mission has provided near-continuous, high-resolution estimates of sea ice freeboard across both hemispheres since data collection started in October 2018. This study provides an impact assessment of upgrades to both the ICESat-2 freeboard data (ATL10) and NASA Eulerian Snow On Sea Ice Model (NESOSIM) snow loading on estimates of winter Arctic sea ice thickness. Misclassified leads were removed from the freeboard algorithm in the third release (rel003) of ATL10, which generally results in an increase in freeboards compared to rel002 data. The thickness increases due to increased freeboards in ATL10 improved comparisons of Inner Arctic Ocean sea ice thickness with thickness estimates from ESA's CryoSat-2. The upgrade from NESOSIM v1.0 to v1.1 results in only small changes in snow depth and density which have a less significant impact on thickness compared to the rel002 to rel003 ATL10 freeboard changes. The updated monthly gridded thickness data are validated against ice draft measurements obtained by upward-looking sonar moorings deployed in the Beaufort Sea, showing strong agreement (r2 of 0.87, differences of 11 ± 20 cm). The seasonal cycle in winter monthly mean Arctic sea ice thickness shows good agreement with various CryoSat-2 products (and a merged ICESat-2–CryoSat-2 product) and PIOMAS (Pan-Arctic Ice-Ocean Modeling and Assimilation System). Finally, changes in Arctic sea ice conditions over the past three winter seasons of data collection (November 2018–April 2021) are presented and discussed, including a 50 cm decline in multiyear ice thickness and negligible interannual differences in first-year ice. Interannual changes in snow depth provide a notable impact on the thickness retrievals on regional and seasonal scales. Our monthly gridded thickness analysis is provided online in a Jupyter Book format to increase transparency and user engagement with our ICESat-2 winter Arctic sea ice thickness data.

PIOMAS-20C, an Arctic sea ice reconstruction for 1901–2010, is produced by forcing the Pan-Arctic Ice Ocean Modeling and Assimilation System (PIOMAS) with ERA-20C atmospheric data. ERA-20C performance over Arctic sea ice is assessed by comparisons with measurements and data from other reanalyses. ERA-20C performs similarly with respect to the annual cycle of downwelling radiation, air temperature, and wind speed compared to reanalyses with more extensive data assimilation such as ERA-Interim and MERRA. PIOMAS-20C sea ice thickness and volume are then compared with in situ and aircraft remote sensing observations for the period of ~1950–2010. Error statistics are similar to those for PIOMAS. We compare the magnitude and patterns of sea ice variability between the first half of the twentieth century (1901–40) and the more recent period (1980–2010), both marked by sea ice decline in the Arctic. The first period contains the so-called early-twentieth-century warming (ETCW; ~1920–40) during which the Atlantic sector saw a significant decline in sea ice volume, but the Pacific sector did not. The sea ice decline over the 1979–2010 period is pan-Arctic and 6 times larger than the net decline during the 1901–40 period. Sea ice volume trends reconstructed solely from surface temperature anomalies are smaller than PIOMAS-20C, suggesting that mechanisms other than warming, such as changes in ice motion and deformation, played a significant role in determining sea ice volume trends during both periods.

<p>Ocean bottom pressure (OBP) from the Gravity Recovery and Climate Experiment (GRACE) revealed Arctic Ocean circulation patterns and variability that were previously unknown (Morison et al., 2007; Morison et al., 2012; Peralta-Ferriz et al., 2014). OBP measurements from the GRACE Follow-On mission (GRACE-FO) are therefore increasingly important for monitoring Arctic Ocean variability, and critical for understanding and predicting the fate of the rapidly changing Arctic environment.</p> <p>In this work we use GRACE data from 2002 to 2017 jointly with a 10-year record of <em>in situ</em> OBP at the North Pole (2005-2015) complemented with <em>in situ</em> OBP in the Canada Basin (2015-2018), and wind reanalysis products, to create a proxy representation of the OBP anomalies that explains the largest possible fraction of the observed OBP variability in the Arctic Ocean and the Nordic Seas. We do this by performing a linear regression analysis, combined with maximum covariance analysis (MCA) – a technique that was tested prior to the decommission of GRACE and the launch of GRACE-FO (Peralta-Ferriz et al., 2016). Here, the first predictor time series is the <em>in situ</em> OBP record at the North Pole and Canada Basin; the second predictor time series is the expansion coefficients time series of the leading mode of MCA between the GRACE OBP coupled with the winds. We use this proxy OBP to merge GRACE with the first 2 years of available GRACE-FO OBP. We compare our merged OBP field with OBP output from the Pan-Arctic Ice Ocean Modeling and Assimilation System (PIOMAS). Preliminary results suggest a good agreement between the proxy and predicted OBP fields and both GRACE and GRACE-FO data, especially in the central Arctic, but also in the Nordic Seas. The OBP variations from the merged GRACE and GRACE-FO and from PIOMAS will be also explored.</p> <p><strong>

Large-scale changes in Arctic sea ice thickness, volume and multiyear sea ice (MYI) coverage with available measurements from submarine sonars, satellite altimeters (ICESat and CryoSat-2), and satellite scatterometers are summarized. The submarine record spans the period between 1958 and 2000, the satellite altimeter records between 2003 and 2018, and the scatterometer records between 1999 and 2017. Regional changes in ice thickness (since 1958) and within the data release area of the Arctic Ocean, previously reported by Kwok and Rothrock (2009 Geophys. Res. Lett. 36 L15501), have been updated to include the 8 years of CryoSat-2 (CS-2) retrievals. Between the pre-1990 submarine period (1958–1976) and the CS-2 period (2011–2018) the average thickness near the end of the melt season, in six regions, decreased by 2.0 m or some 66% over six decades. Within the data release area (∼38% of the Arctic Ocean) of submarine ice draft, the thinning of ∼1.75 m in winter since 1980 (maximum thickness of 3.64 m in the regression analysis) has not changed significantly; the mean thickness over the CS-2 period is ∼2 m. The 15 year satellite record depicts losses in sea ice volume at 2870 km3/decade and 5130 km3/decade in winter (February–March) and fall (October–November), respectively: more moderate trends compared to the sharp decreases over the ICESat period, where the losses were weighted by record-setting melt in 2007. Over the scatterometer record (1999–2017), the Arctic has lost more than 2 × 106 km2 of MYI—a decrease of more than 50%; MYI now covers less than one-third of the Arctic Ocean. Independent MYI coverage and volume records co-vary in time, the MYI area anomalies explain ∼85% of the variance in the anomalies in Arctic sea ice volume. If losses of MYI continue, Arctic thickness/volume will be controlled by seasonal ice, suggesting that the thickness/volume trends will be more moderate (as seen here) but more sensitive to climate forcing.

Because of limited high-quality satellite and in situ observations, less attention has been given to the trends in Arctic sea ice thickness and therefore sea ice volume than to the trends in sea ice extent. This study evaluates the spatial and temporal variability in Arctic sea ice thickness using the Pan-Arctic Ice Ocean Modeling and Assimilation System (PIOMAS). Additionally, the Community Earth System Model Large Ensemble Project (LENS) is used to quantify the forced response and internal variability in the model. A dipole spatial pattern of sea ice thickness variability is shown in both PIOMAS and LENS with opposite signs of polarity between the East Siberian Sea and near the Fram Strait. As future sea ice thins, this dipole structure of variability is reduced, and the largest interannual variability is found only along the northern Greenland coastline. Under a high-emissions scenario (RCP8.5) projection, average September sea ice thickness falls below 0.5 m by the end of the twenty-first century. However, a regional analysis shows internal variability contributes to an uncertainty of 10 to 20 years for the timing of the first September sea ice thickness less than 0.5 m in the marginal seas.

Sea ice thickness is an important component of the Arctic environment, bearing crucial significance in investigations pertaining to global climate and environmental changes. This study employs data from the HaiYang-2B satellite altimeter (HY-2B ALT) for the estimation of Arctic Sea ice thickness from November 2021 to April 2022. The HY-2B penetration coefficient is calculated for the first time to correct the freeboard in areas with sea ice concentration greater than 90%. The estimation accuracy is improved by enhancing the data on sea ice density, seawater density, snow depth, and snow density. The research analyzed the effects of snow depth and penetration coefficient on sea ice thickness results. The results of sea ice type classification were compared with OSI-SAF ice products, and the sea ice thickness estimation results were compared with four satellite ice thickness products (CryoSat-2 and SMOS (CS-SMOS), Centre for Polar Observation and Modelling Data (CPOM), CryoSat-2 (CS-2), and Pan-Arctic Ice-Ocean Modeling and Assimilation System (PIOMAS)) as well as two validation ice thickness data sets (Operation IceBridge (OIB) and ICEBird). The accuracy of sea ice classification exceeds 92%, which is in good agreement with ice type product data. The RMSD of sea ice thickness estimation is 0.56 m for CS-SMOS, 0.68 m for CPOM, 0.47 m for CS-2, 0.69 m for PIOMAS, and 0.79 m for validation data.