An unusual, large, latent-heat polynya opened and then closed by freezing and convergence north of Greenland's coast in late winter 2018. The closing presented a natural but well-constrained full-scale ice deformation experiment. We observed the closing of and deformation within the polynya with satellite synthetic-aperture radar (SAR) imagery and measured the accumulated effects of dynamic and thermodynamic ice growth with an airborne electromagnetic (AEM) ice thickness survey 1 month after the closing began. During that time, strong ice convergence decreased the area of the refrozen polynya by a factor of 2.5. The AEM survey showed mean and modal thicknesses of the 1-month-old ice of 1.96

Sea ice thickness is a key climate variable because it governs the mass, heat, and momentum exchange between the ocean and the atmosphere

The interplay of dynamics and thermodynamics results in large thickness variations, and ice thickness distributions (ITDs) are used to characterize them. Thermodynamic processes modify ice thickness slowly depending on the surface energy balance, and growth is limited to the equilibrium thickness

In contrast, deformation caused by differential ice motion leads to abrupt changes in ice thickness. Driven by winds, ocean currents, and tides and constrained by coasts and the internal stress of the ice pack, divergent motion creates areas of open water, e.g., leads and polynyas, and reduces thickness to zero. Convergent motion results in the closing of leads and then rafting and ridging of young and old ice. Ridging of thick ice forms pressure ridges that are many times thicker than the initial thickness

The ITD is a key parameter in the parameterization of many climate and weather-relevant processes. For example, effective heat transfer between the ocean and atmosphere is limited to thin ice. Hence, knowledge of the ITD is crucial for realistic short- and long-term model predictions of sea ice thickness and volume

Submarine and satellite-based observations have shown a substantial decline in sea ice thickness in the Arctic Ocean within the last 6 decades

However, the interdependency between sea ice thickness and enhanced sea ice dynamics is not yet well understood. Most apparently, the reduction in the material strength of the ice associated with its thinning is suspected to increase deformation

Ice thickness survey of the first-year ice (FYI) in the refrozen polynya off the coast of North Greenland in March 2018.

It remains challenging to quantify the net effects of changes in sea ice dynamics on sea ice thickness and volume change. The existing redistribution theory that links deformation and thickness change is not yet well constrained by observations

Here, we present a regional case study of sea ice deformation and its impacts on dynamic ice thickness change and redistribution using satellite synthetic-aperture radar (SAR) data and airborne electromagnetic (AEM) ice thickness observations. We have studied refreezing and convergence of ice that had formed in an unusual latent-heat polynya that occurred along the north coast of Greenland in the late winter of 2018 (

In this paper we provided a detailed analysis of deformation derived from SAR imagery and related it to the resulting ice thickness distributions obtained from the AEM surveys. We focused on three aspects: first, we related the large-scale area decrease of the refrozen polynya to the observed average thickness and showed that dynamic processes contributed about 50 % of the observed mean thickness. Second, we related the regional variability in mean thickness and the shape of the ITD to differences in regional deformation observed by SAR ice drift tracking. We established relationships between key properties of the ITD like mean thickness and

We based our work on AEM ice thickness measurements (Sect.

We quantified the thermodynamic growth from a model simulation (Sect.

We derived divergence and shear from SAR-derived sea ice motion fields (Sect.

We used a simple, volume-conserving ice thickness model to calculate ice thickness along the Lagrangian trajectories. We forced the model with the SAR-derived high-resolution deformation fields (Sect.

On 30 and 31 March 2018, the Alfred Wegener Institute's research aircraft

AEM thickness retrievals find the distance to the strongly conducting seawater under the ice. A laser altimeter provides the distance to the upper snow surface, and subtraction of these two distances gives the combined snow and ice thickness

To evaluate snow contribution to the observed total thickness, we analyzed snow thickness from Operation IceBridge (OIB) Sea Ice Freeboard, Snow Depth, and Thickness Quick Look data (for details, see “Data availability” at the end of the text). They surveyed the refrozen polynya on 22 March 2018. We note that OIB's observed modal snow thickness of 4 cm (mean 9 cm) agrees well with the expected accumulation between February and March from the Warren climatology

Meteorological observations at Villum Research Station (Station Nord; 81

Since our study focuses on the evolution of the ice that was formed and deformed during the closing of the polynya, we separated MYI from the newly formed FYI in the refrozen polynya. We used SAR images to identify the northern, outer boundary of the polynya visually. The boundary is clearly visible because of the strong radar backscatter contrast between the FYI (low backscatter) and MYI (high backscatter, Fig.

We used mean and modal thickness to characterize the ice thickness distributions, where the latter was calculated based on a bin width of 20 cm. We considered ice thinner than 10 cm open water. We characterized the tail of the ITD by the

We identified sections of level, undeformed ice along the profiles surveyed by

We aim at separating the dynamic and thermodynamic contributions to the overall thickness. For the thermodynamic growth, we carried out a dedicated thermodynamic model experiment of the refreezing polynya. Instead of applying a freezing-degree-day model like, e.g., that of

The magnitude of deformation is related to the area decrease of the closing polynya. Therefore, we identified the area of the FYI on near-daily Sentinel-1 SAR images from 25 February to 31 March 2018 (Figs.

As a first approximation we assumed ice volume conservation; i.e., the average dynamic ice thickness increase is proportional to the average area (

We computed ice drift fields with an ice tracking algorithm introduced by

We calculated sea ice deformation from the spatial derivatives of the gridded

We aimed at attributing differences in the regional thickness variability measured by the AEM surveys to differences in the deformation history of the respective ice parcels. To obtain information on the drift of those, we reconstructed Lagrangian ice drift trajectories of the surveyed ice parcels using a reverse timeline, as follows:

As starting point of the tracking, we down-sampled the spatial resolution of the thickness profiles surveyed on 30/31 March 2018, to 250 m. Occasional gaps in the thickness observations increased the distance between the starting points to up to 350 m. Next, we corrected the GPS data of the AEM measurements for the ice displacement that took place between the time of the AEM survey and the acquisition of the satellite images (maximum 6 h). In total, we initiated the tracking at 715 down-sampled points along the AEM profiles surveyed on 30/31 March (Fig.

We reconstruct the trajectory of each ice parcel by interpolating the regularly spaced velocity field to the end position at a given time step and adding the respective displacement to determine the end position for the next time step. As examples, four of the reconstructed trajectories are displayed as thin white lines in Fig.

For each time step, which was typically 1 d, we extracted divergence, shear, and total deformation from the deformation fields calculated based on the drift fields (Sect.

We performed the backward Lagrangian tracking from 30/31 until 1 March 2018. We chose 1 March 2018 as the last day of the backtracking because, before this date, the new ice in the polynya was not consolidated and did not reveal recognizable backscatter patterns for retrieving ice drift.

Uncertainties in the initial ice velocity fields propagate into the deformation estimates along the trajectories in three different ways.

The

Random errors in the velocity field introduce

The third source of errors for deformation calculations, the

Considering the errors mentioned above, we used the following approach for calculating possible thickness variations along a trajectory. We extracted divergence from the forward and backward deformation fields (see 2. above) of all grid cells in the circles of uncertainty (see 1. above) along the trajectories (Fig.

Sketch of the simple ice thickness model showing the vertical cross section of thickness change between time steps

Based on the basic principles of thermodynamic and dynamic ice thickness changes described by

Our model is based on the redistribution theory introduced by

Here, the thermodynamic growth or melt rate (

We made the following assumptions about the ice properties: first, we determined the thickness

Second, we approximated the thermodynamic ice growth

Based on Eq. (4) we obtain the mean thickness at each time by

To account for the uncertainties in the drift and deformation, we calculated for each of the 715 trajectories mean thickness and standard deviation as uncertainty from the 10 000 thickness estimates obtained from the 10 000 random combinations of divergence (see Sect.

In this section, we first quantify the large-scale dynamic thickness change that is linked to the decrease in the refrozen polynya area (Sect.

The AEM thickness surveys showed that after only 1 month of ice growth, the newly formed FYI had a mean thickness of 1.96

Ice thickness distributions (ITDs) displaying snow and ice thickness and observed by AEM on 30/31 March. ITD over the entire area of the refrozen polynya (black) and ITD of the level ice only (blue). Mean, mode, exponential fit, and FWHM are indicated for the former case. Ice thicker than 8 m was observed for less than 1 % of the refrozen polynya area.

The MITgcm thermodynamic model gives a thermodynamic ice thickness of 0.87

Figure

Dynamic and thermodynamic contributions to mean thickness from model and observations.

The shape of the ice thickness distribution showed signs of strong deformation (Sect.

After the polynya had reached its maximum extent on 25 February 2018

We computed a time series of mean ice thickness using Eq. (

The previous section was concerned with the large-scale, mean dynamic thickness change in the refrozen polynya. In the following, we examine local ice thickness and deformation variations, as well as the potential links between them.

ITDs of the four FYI zones of all three AEM lines on 30/31 March 2018. The ITDs differ in

Along all three ice thickness profiles (northern, central, and eastern) from the coast across the refrozen polynya, we found common patterns of thickness variability (Figs.

Ice thickness profile of the northern profile. The black and gray colors distinguish between level and deformed ice. Four subsets, representative of the four thickness zones, are displayed in

Figure

Properties of ITDs of different zones of the refrozen polynya and the result of the simple, volume-conserving thickness model (see Sect.

Trajectories, drift, and deformation during the three main deformation phases.

The dominant direction of the 715 reconstructed trajectories (see Sect.

In short, we were able to identify four zones across the FYI with clearly differently shaped ITDs and different deformation histories. Since thermodynamic growth was rather uniform, we conclude that the observed spatial thickness variability is fully linked to the deformation history of the ice. In the following section, we will further explore this link on a more quantitative basis.

In the previous section, we qualitatively described the relationship between the spatially varying deformation and ice thickness properties. Here, we provide quantitative relationships between divergence and total deformation on the one hand and different ITD properties on the other hand using linear regression (Fig.

Figure

Relationship between mean deformation and ITD key parameters in the four polynya zones. If applicable, standard deviations are displayed as error bars. Thickness and mean deformation are given for 1–31 March. We subtracted the thermodynamic thickness of 0.38 m that was reached on 1 March from the mean thickness on 30/31 March. Note that convergence is negative divergence.

In the two previous sections, we described the impact of the polynya-wide deformation and its local variations on the thickness distribution. We demonstrated that the area decrease of the closing polynya could directly be used to accurately predict the corresponding ice thickness increase (Sect.

We modeled thickness change along each of the 715 trajectories based on the modeled thermodynamic growth from the MITgcm run and the observed deformation between 1 and 31 March, as described in Sect.

In Fig.

Observed and modeled ITD with mean (dot), exponential fit to the tail of the distribution, and FWHM (horizontal bar).

Modeled and observed thickness profiles across the FYI from south (left) to north (right) of

The ITD of the modeled thicknesses along the 715 trajectories on 30/31 March is shown in Fig.

Lastly, we compared the modeled and observed thicknesses along the three AEM profiles (Fig.

One of our key results is that after only 1 month of thermodynamic and dynamic thickness growth, the ice thickness in the refrozen polynya had increased from 0.4 to 2 m. Sea ice deformation had contributed on average 50 % and locally up to 90 % to the mean ice thickness. This large contribution of sea ice dynamics is consistent with

Our results obtained on local scales of a refrozen polynya and over 1 month bridge the spatial and temporal gap between two recent, similar studies of ice deformation and thickness change: the short-term, local-scale study by

Our observations provide insights into two key aspects in modeling sea ice dynamics, namely, the mean dynamic thickness change and the effect of deformation on the shape of the ITD, whose accurate representation in models is the subject of current research

First, our results showed that mean dynamic thickness change can be approximated as a linear function of convergence (Fig.

The thermodynamic growth term (

The redistribution theory

Second, our results suggest that the

We test whether different ice thicknesses as suggested by

Based on the good linear fit (Fig.

We identified two processes that change the

Ridge formation models from

Rafting leads to a different

Lastly, we acknowledge other aspects: the creation of rubble fields, hummocks, or the ratio of shear to convergence could influence the

Based on a simple volume-conserving model, we derived thickness change along ice drift trajectories and calculated ITDs from the final thickness at the end of each trajectory.

The modeled ITD resembles the observed one in the typical, skewed shape with a dominant central mode and a long tail of thicker ice (Sect.

Our modeled ITD agrees well with the observations in the thinner thickness categories. However, it shows a second mode at 2.2–2.4 m (Fig.

Apart from those differences in the shape of the ITD, we have found that the modeled mean ice thicknesses were generally smaller than the observed ones (Table

First, our model does not account for the high macro-porosity of unconsolidated FYI ridge keels, which leads to an underestimation of the thickness. Numerous studies have shown that mean ridge porosities amount to 11 %–22 %

Second, in the simple, volume-conserving model, the thermodynamic growth was modeled based on the growth of an undeformed layer of ice, regardless of the actual mean thickness of each grid cell. Hence, the model overestimates thermodynamic growth in all cells that experienced strong convergence and were, therefore, thicker than the thermodynamic thickness. At the same time, our approach underestimates ice growth in all cells that experienced divergence because thermodynamic growth is stronger in leads than in adjacent consolidated ice. We carried out a sensitivity study to estimate the impact of unaccounted for new ice formation in leads. If there was divergence, we replaced the ice leaving the grid cell with new ice of a thickness that could form within 1 d. Integrated over 30 d and all profiles, this resulted in an additional 0.3 m of ice, i.e., a mean thickness of 2 m. Since the dominating deformation type in this study was convergence and shear, this effect is less important than in a different deformation regime. We suggest coupling the deformation history retrieved from SAR analysis with a fully developed sea ice model that considers those interdependencies in future work. For example, the single-column model Icepack includes full solutions for thermodynamic growth and melting, as well as mechanical redistribution due to ridging

Both those shortcomings can explain the observed differences in the mean thicknesses. However, there are additional reasons for deviations of observed and modeled thickness briefly discussed below.

The daily imaging of the polynya by SAR images cannot account for deformation caused by tides. Tides and inertial motion can cause recurrent opening and closing with associated sub-daily new ice formation and subsequent deformation. These processes can contribute 10 %–20 % of the Arctic-wide seasonal ice growth

Single early deformation processes before 1 March might already have created an inhomogeneous ice thickness field in contrast to our assumed, initial, uniform thickness. Since we did not observe a decrease in the total polynya area between 25 February and 1 March, ice thickness variations in this period could only be explained by localized effects.

Even the consideration of the uncertainties in the deformation parameters and in the positions of trajectories cannot explain all deviations between modeled and observed thickness (Fig.

We did not consider additional opening and closing of ice due to shear on subgrid scales that can be observed in similar situations

An unusual latent heat polynya with a size of

This study provides evidence of the high relevance of deformation dynamics in creating and maintaining a thick ice cover. In the refrozen polynya, sea ice deformation contributed on average 50 % and locally up to 90 % to the mean thickness. Within 1 month, the dynamic processes re-established an ice cover with a mean thickness of 1.96 m, almost as thick as the surrounding multi-year ice, which had a mean thickness of 2.1 m (results not shown here).

In the view of a changing Arctic with increasing fractions of thin ice, increased ice drift speed, and a higher frequency of deformation events, accurate representation of sea ice deformation in models is crucial for predictions of future sea ice thickness and extent. Our observations reveal new insights into the link between deformation and the redistribution of ice, which determines the shape of the ice thickness distribution (ITD). We provide quantitative evidence that the deformation magnitude impacts the

We developed a simple volume-conserving model to derive dynamic thickness change from deformation fields with a spatial resolution of 1.4 km obtained from SAR satellite imagery. The modeled mean thicknesses were smaller than AEM thickness observations, but they agree within the limits of the main uncertainties due to ridge porosity and omitted new ice formation in leads formed by divergence.

The volume-conserving model allowed us to reconstruct an ITD that resembled the ITD obtained from the AEM thickness observations. They both have the typical skewed shape with a dominant central mode and a long tail of thicker ice. However, we note that without a redistribution scheme, the thickest ice of the ITD cannot be realistically modeled.

For future work, we suggest coupling the deformation history retrieved from SAR analysis with a fully developed sea ice model that takes drift and deformation as forcing and calculates dynamics and thermodynamics for several thickness categories, e.g., Icepack

Sentinel-1 scenes are available from the Copernicus Open Access Hub (

Video supplement 1 (available at

The supplement related to this article is available online at:

LvA carried out the analysis, processed the deformation data, and wrote the manuscript. All authors contributed to the discussion and provided input during the concept phase and the writing process.

Christian Haas is a member of the editorial board of

We thank Martin Losch for fruitful discussions on the content of this paper and for helping set up the MITgcm model runs. Thomas Hollands supported us with the tracking algorithm. Nils Hutter provided initial oceanographic data sets for the model runs. Stefan Hendricks and Jan Rohde were involved in the field campaign and processed the AEM data. Thomas Krumpen and Florent Birrien provided guidance on the thermodynamic growth in the polynya. Conor Murray gave valuable feedback on copy-editing.
This work contains modified Copernicus Sentinel data (2019), and snow thickness data used in this study were acquired by NASA's Operation IceBridge. We acknowledge EUMETSAT Ocean and Sea Ice Satellite Application Facility (OSI SAF;

This research was funded by the Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research (AWI), through its research program “Polar regions And Coasts in the changing Earth System”, Topic 1.4 “Arctic sea ice and its interaction with ocean and ecosystems”, and by the Deutsche Forschungsgemeinschaft (DFG) through the International Research Training Group (IRTG) ArcTrain (GRK1904). The article processing charges for this open-access publication were covered by the Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research (AWI).

This paper was edited by Jennifer Hutchings and reviewed by Amélie Bouchat and one anonymous referee.