The SMOS-SIT data provide essential information about sea ice that is complementary to observations of sea-ice concentration using higher-frequency passive microwave channels. To illustrate this, Fig. 1 shows SMOS-SIT sea-ice thickness together with observed sea-ice concentration from the OSTIA product for a day early in the freezing season and for a day late in the freezing season. Note that here and elsewhere, unless explicitly stated otherwise, sea-ice thickness denotes the equivalent sea-ice thickness, i.e. the sea-ice volume per area (see Notz et al., 2016 for further details).

Early in the freezing season, there are large areas of newly formed sea ice that is thin. Figure 1a shows that sea-ice thicknesses of 0.6–0.7 m dominate in the Beaufort and Chukchi seas, as well as in part of the central Arctic Ocean adjacent to them. In Baffin Bay, sea-ice thickness from SMOS-SIT is even thinner, at around 0.2–0.3 m. All these regions exhibit sea-ice concentrations of virtually 100 % (Fig. 1b), which demonstrates that sea-ice concentration observational products like OSTIA cannot be used to distinguish areas with thin new sea ice from areas of old thick sea ice; sea-ice thickness observational products like SMOS-SIT are needed to do that.

Sea-ice thickness in ORAS5 in early winter is comparable to that of SMOS-SIT (Fig. 1c). However, the model tends to simulate thicker ice on average. Note that the departures in Figure 1c, f are only shown for SMOS-SIT data points with a saturation ratio of less than 90 % and total retrieval uncertainty of less than 1 m (see Sect. 2.1 for definitions of these). Positive departures dominate, especially close to regions of thick ice. There are a few places in the Beaufort and the Siberian Shelf seas with negative departures, but in most of the thin-ice areas ORAS5 simulates ice around 0.4 m thicker than that retrieved by SMOS-SIT.

As the freezing season progresses, the ice edge moves further south, out of the Arctic Basin, and previously formed thin ice in the Arctic Basin becomes thicker. Polynyas and fracture zones begin to form. These refreeze very quickly, which is evident from the near-100 % sea-ice concentration but greatly reduced sea-ice thickness in these features. Figure 1d shows extensive refrozen polynyas in the Kara and Laptev seas, as well as a fracture zone covering the whole of the Beaufort Sea. In Baffin Bay, sea-ice thickness derived by SMOS-SIT is mostly below 0.3 m. Again, none of these features within the ice pack are picked up by the sea-ice OSTIA concentration product, which shows homogeneously high ice concentration throughout the ice pack (Fig. 1e).

The departures between ORAS5 and SMOS-SIT in late winter are large and positive throughout (Fig. 1f), with values of 1 m or more dominating. Most of this is due to the failure of the reanalysis to simulate relevant features like the refrozen coastal polynya in the Laptev Sea or refrozen fracture zones like the one visible in the SMOS-SIT data for the Beaufort Sea.

There are multiple plausible reasons for the poor representation of refrozen polynyas and fracture zones in the reanalysis: various deficiencies in the ocean and sea-ice models (e.g. too thick ice, inappropriate rheology, insufficient modelling of open-water ice growth, too strong upper-ocean stratification), the data assimilation methods (e.g. inappropriate background error covariance between ice concentration and ice volume) or deficiencies in the atmospheric forcing (e.g. too weak offshore winds). Further investigation of this reanalysis deficit is clearly needed, but for the most part this requires dedicated experimentation and is therefore out of the scope of this study. However, it can be said that there are conspicuous features in maps of sea-ice concentration increments (not shown), which directly affect ice thickness through implied ice volume increments as discussed by Tietsche et al. (2013). For the day in question, 15 April 2016, the sea-ice concentration increments are large and positive in the refrozen polynyas and fracture zones. This would suggests that the model dynamics tend to produce the features, but the assimilation increments suppress them in the reanalysis.

In the Barents Sea there is good agreement between ORAS5 and SMOS-SIT, with a positive departure of 20 cm or less. Finally, Baffin Bay stands out as having extensive thin ice cover in SMOS-SIT but thick ice in ORAS5. The North Water Polynya at the northern end of Baffin Bay is captured both by SMOS-SIT and ORAS5.

Figure 2Normalised joint frequency distribution (scatter density) of co-located pairs of daily observed and analysed thin sea ice: (a) October to December 2011–2017, (b) February to April 2012–2017. The text insets in the lower-right corner give information on the pre-filtering of the data before the scatter density is produced: data points are only considered if the retrieval uncertainty is below 1 m (unc < 1 m), the sea-ice concentration from OSTIA is above 30 % (sic > 30 %) and the saturation ratio is below 90 % (srat < 90 %). The last line of the text inset gives the total number of data points for which the scatter density was calculated.

Download

The previous example maps show typical conditions in early and late winter, and typical departures between ORAS5 and SMOS-SIT. For a more quantitative assessment, we calculate departures for co-located daily sea-ice thickness in (a) the early-winter period 15 October to 15 December for the years 2011–2017, and (b) the late-winter period 15 February to 15 April for the years 2012–2017. We exclude data points where the SMOS-SIT retrieval is known to be unreliable: data points with a retrieval uncertainty of more than 1 m, a saturation ration of above 90 % or a sea-ice concentration below 30 % are not considered (see Sect. 2.1 for explanations of retrieval uncertainty and saturation ratio).

From these co-located pairs of observed and modelled daily sea-ice thicknesses we calculate the normalised bivariate joint frequency distribution, which we will call “scatter density” in the following for the sake of brevity. Scatter density plots give quite a complete picture of the departure statistics. For a good match, density should be high on the one-to-one line and low elsewhere. High density above the one-to-one line indicates a positive bias, high density below the one-to-one line indicates a negative bias. Conditional departure characteristics, e.g. for a certain range of observed values, can also easily be derived visually.

As can be seen from the scatter density in Fig. 2a, in early winter the agreement between SMOS-SIT and ORAS5 sea-ice thickness is quite promising as the distribution is close to the one-to-one line. However, the overestimation of sea-ice thickness by ORAS5 is confirmed but was already visually apparent from the maps in Fig. 1. For observed sea-ice thickness between 0 and 0.3 m, ORAS5 sea-ice thickness is about 0.3 m higher. The agreement becomes better for higher observed sea-ice thickness in the range 0.5–1 m. Note that the scatter density distribution has wide tails. For instance, for an ice thickness of 0.4 m in SMOS-SIT, ORAS5 values of up to 1.5 m exist. This is not so obvious in the scatter density but is clearly visible in the corresponding scatter plot, which tends to highlight outlier data points (not shown).

Part of the reason for ORAS5 having higher ice thickness than SMOS-SIT early in the freezing season is the simplified representation of thin ice in LIM2, the sea-ice component of ORAS5 (see Sect. 2.2): thermodynamic formation of new ice in LIM2 happens at a fixed actual (floe) thickness of 0.6 m, a value that has been chosen to approximate growth processes in the presence of thick sea ice (Hibler, 1979). Quite obviously, this is not a good representation of how sea ice forms from open water, which is the dominant regime at the ice edge early in the freezing season. As can be seen in Fig. A1c, the simplified LIM2 treatment of thin sea ice leads to an artificially high frequency in the occurrence of equivalent ice thicknesses around 0.6 m, because ice growth rates are artificially enhanced for equivalent ice thicknesses below that value (see Mellor and Kantha, 1989, Tietsche et al., 2017 and Shi and Lohmann, 2017 for further discussion on this).

A second reason for higher ice thickness in ORAS5 than SMOS-SIT early in the freezing season is that SMOS-SIT retrieves sea-ice thickness under the assumption of 100 % sea-ice concentration. If the area captured by a SMOS pixel has only partial ice cover, the SMOS-SIT ice thickness retrieval is biased thin (Tian-Kunze et al., 2014). As can be seen from Fig. B1b, there is an almost perfect linear relationship between SMOS TBs and sea-ice concentration for intermediate sea-ice concentration values, which clearly indicates that geometrical averaging of open-water and sea-ice emissivity within a SMOS pixel plays a role. When excluding data points from Fig. 2a, where sea-ice concentration is below 95 % (not shown), the scatter density conditional on SMOS-SIT thickness being below 0.2 m is almost zero, which is a good indication that all thickness retrievals at least up to this thickness are likely to have a low bias due to neglecting the open-water contribution to L-band emissivity.

Figure 3Normalised joint frequency distribution (scatter density) of observed and analysed thin sea ice in late winter from February to April 2012–2017: (a) Barents and Kara seas (15–90^∘ E, 70–85^∘ N), (b) Laptev Sea (90–150^∘ E, 70–85^∘ N), and (c) Baffin Bay (75–53^∘ W, 65–80^∘ N). For an explanation of the text insets, see caption of Fig. 2.

Download

In late winter, ORAS5 has much higher sea-ice thickness than SMOS-SIT (Fig. 2b). Departures between 0.5 m and 1m are common throughout the SMOS-SIT thickness range of 0–1 m. There is a more linear shape in the scatter density distribution – this is promising in principle, but could result from errors compensating for each other in different regions, which would make the relationship less relevant. The scatter distribution is also much wider than for early winter, indicating larger and more uncertain reanalysis–observation differences. The larger discrepancy in late winter has several causes. Figure 1d–f illustrate the most obvious one: the ocean reanalysis does not simulate polynyas and fracture zones well. However, there are other causes, some of which are related to the properties of SMOS-SIT data. In the following section, we analyse the late-winter departures in more detail.

There is considerable regional dependence of the departures in late winter (February to April). Figure 3 shows the SMOS-SIT/ORAS5 scatter density as in Fig. 2b but for three key regions separately: the Barents and Kara seas, the Laptev Sea and Baffin Bay. For the Barents and Kara seas (Fig. 3a), the departure statistics are almost as good as for the pan-Arctic in early winter (Fig. 2a). We can conclude that this region has relatively good agreement between ORAS5 and SMOS-SIT sea-ice thickness throughout the winter. In the Laptev Sea (Fig. 3b), ORAS5 has no ice thickness below 1 m, whereas SMOS-SIT detects a lot of ice thinner than 1 m. There is a very low correlation between ORAS5 and SMOS-SIT ice thickness. This behaviour is consistent with our earlier assessment that refrozen polynyas occur frequently in the Laptev Sea in late winter and that they are detected by SMOS-SIT but are not well represented in ORAS5.

Finally, Fig. 3c shows the late-winter scatter density for Baffin Bay, which again has characteristics that are very different from the other two regions. In general, ORAS5 simulates much thicker ice than that retrieved by SMOS-SIT, but in contrast to the Laptev Sea case, there is a quite high rank correlation between SMOS-SIT and ORAS5; i.e. higher SMOS-SIT values are often associated with higher ORAS5 values but the correspondence is not necessary linear. This suggests systematic rather than random sources for the departures.

An interpretation of the results in Fig. 3 needs to start from the appreciation that the regions shown have quite different physical characteristics: in the Barents and Kara seas, sea ice is strongly affected by warm Atlantic water being advected towards and under the ice, which means the ice cover is constrained by SST. At the same time, prevailing winds modulate the location of the ice edge by transporting the ice. Both processes are expected to be reasonably well simulated by ORAS5, because winds are prescribed as forcing, and the SST are ingested from an observational product. For the observations, most of the calibration and validation campaigns for SMOS-SIT have been carried out in this area (Kaleschke et al., 2016). Thus, the Barents and Kara seas can be expected to be where the reanalysis–observation agreement is best.

In the Laptev Sea, sea ice is still relatively well observed when it comes to SMOS-SIT validation, but it is more difficult to simulate in ORAS5. Because there is no ice edge in the Laptev Sea, SST information cannot be used to constrain the ice cover. Furthermore, as clearly visible in Fig. 1, extensive polynyas form there in February to April, mainly when offshore winds push back the ice from land or land-fast sea ice. These processes are not well simulated by the reanalysis, which tends to keep a compact thick sea-ice cover even in the presence of offshore winds. As a result, major departures can be expected.

Figure 4Time series of daily sea-ice thickness during the 2011–2012 winter at (a) a representative location in the Laptev Sea at 74.5^∘ N, 127^∘ E and (b) a representative location in Baffin Bay at 72^∘ N, 62^∘ W. Blue represents SMOS-SIT (full line) with added and subtracted uncertainty standard deviation (dotted lines). Red shows the five realisations of ORAS5. Black horizontal lines are the CryoSat2 average thickness for March–April provided by CPOM; black star is an EM-bird overfly for the Laptev Sea on 20 April 2012. The corresponding time series of sea-ice concentration are shown in Fig. 5b.

Download

In Baffin Bay, the occurrence of thinner ice of varying thickness is modelled and observed, but the modelled ice is roughly twice as thick. There is independent information that suggests that SMOS ice thickness has a low bias there (see Landy et al., 2017; Laxon et al., 2013; Tilling et al., 2015). CryoSat2 estimates (http://www.cpom.ucl.ac.uk/csopr/seaice.html, last access: 12 June 2018) indicate that between February and April, the ice in this region is typically 1.5 m thick. This is confirmed by independent expert judgement by ice analysts, who estimate that ice in this region and this season would typically be at least 1 m thick (Nick Hughes, personal communication, 2016).

To further illustrate and consolidate the findings from Fig. 3, we plot time series for two representative locations in the Laptev Sea and Baffin Bay in Fig. 4. Both show the typical behaviour of reanalysis–observation departures: SMOS-SIT and ORAS5 match well early in winter, but later on ORAS5 ice keeps thickening while SMOS-SIT thickness saturates, albeit with some strong fluctuations. We choose to present a full freezing season in winter 2011–2012, because this allows co-location with independent data in both locations. For the Laptev Sea (Fig. 4a), there was a campaign in April that measured the ice thickness using an EM-bird. This measurement method has demonstrated uncertainties of less than 0.1 m (Haas et al., 2009); hence we can use it as the ground truth to benchmark remote sensing observations and reanalysis results. The EM-bird measurement confirms that the ice was indeed only about 0.5 m thick, which indicates the presence of new thin ice in the Laptev Sea polynya (Tian-Kunze et al., 2014). The reanalysis is not able to simulate this. The CryoSat2 estimate for this location is around 1 m averaged over March and April, halfway between ORAS5 and SMOS-SIT.

For a representative location in Baffin Bay (Fig. 4b), there is reasonable match between reanalysis and observations until January. After that, the sea ice in ORAS5 keeps growing to reach thicknesses of 1.5–2 m in mid-April, whereas SMOS-SIT observations level off between 0.5 and 1 m until mid-April. The CryoSat2 estimate for this location, averaged over March–April 2012, is 1.8 m. This behaviour is generic: it occurs in all years and also when considering an area average over the west of Baffin Bay as defined by Landy et al. (2017)

Figure 5Time series for relevant SMOS-SIT and ORAS5 parameters for the Baffin Bay location (72^∘ N, 62^∘ W) for the full freezing season 2011–2012. Blue curves are SMOS-SIT parameters (except in panel b, where blue is observed ice concentration from OSTIA), and red curves are model parameters.

Download

When judging the compatibility of observational and model-based estimates of sea-ice thickness, their uncertainties should be taken into account. The available uncertainty estimates are indicated in Fig. 4 in the form of five perturbed ensemble members of the ORAS5 reanalysis and in the form of lower and upper bounds of the SMOS-SIT uncertainty estimate provided with the data set. The estimated ORAS5 uncertainty is very small – well below 0.3 m most of the time. It is almost certainly too small, as it does not account for structural uncertainty in the model and data assimilation methods. By contrast, the SMOS-SIT uncertainty range (see Sect. 2.1) is very variable and often very large. Sometimes it covers the whole range of fathomable values and sometimes it is small, but independent evidence suggests that the truth lies far outside the uncertainty range provided. An example of the former case is the SMOS-SIT ice thickness in the Laptev Sea (Fig. 4a) in February: the retrieved value is 1.2 m, but the uncertainty ranges from 0 m to more than 2 m. An example for the latter case is the SMOS-SIT ice thickness in Baffin Bay in April 2012 (Fig. 4b): the retrieved value is 0.5 m with an uncertainty estimate of only 0.1 m. As argued before, the true sea-ice thickness was very likely much higher than that.

Given that ORAS5, CryoSat2 and expert judgement agree that sea ice in Baffin Bay at this time of year should be considerably thicker than SMOS-derived thicknesses, we tentatively suggest that there is a problem with the retrieval assumption of SMOS-SIT in this region. From Fig. 5a, e it can be seen that the slight decrease in SMOS TBs from February onwards is interpreted as a strong decrease in retrieved sea-ice thickness in SMOS-SIT, in disagreement with the ORAS5 reanalysis and independent observations.

A sea-ice concentration slightly below 100 % (Fig. 5b) might also play a role: for the high SMOS TBs typical of late-winter conditions in Baffin Bay, even a few percent of open water within the SMOS footprint will lower TBs significantly by geometrical averaging (note that differences of open-water fraction of a few percent are difficult if not impossible to observe reliably; Ivanova et al., 2015). The assumption of 100 % sea-ice cover made by SMOS-SIT will then lead to a thickness retrieval that has a low bias. The scatter density of OSTIA sea-ice concentration versus SMOS-SIT sea-ice thickness for the west of Baffin Bay in late winter (not shown) shows moderate correlation between the two; i.e. there was open water present and it was usually associated with lower ice thicknesses in the SMOS-SIT retrievals. This is an indication – but not proof – that SMOS-SIT might systematically underestimate ice thickness in Baffin Bay because of non-negligible amounts of open water.

Figure 6Monthly November means of the pan-Arctic area covered by ice thicker than given thresholds in SMOS-SIT (a) and ORAS5 (b).

Download

Sea-ice surface temperature (Fig. 5d) is almost always colder in ORAS5 than in SMOS-SIT. This is consistent with the ice being thicker in ORAS5 than in SMOS-SIT: the thicker the ice, the smaller the surface heating by conductive heat flux from the relatively warm ocean water below the ice to the relatively cold surface of the ice. However, different near-surface temperatures in the two reanalyses (JRA-55 and ERA-Interim,) might also play a role (see Bauer et al., 2016), because they will have a direct impact on the implied sea-ice bulk temperature. Note that there is an apparent artefact in the ice surface temperature in the SMOS-SIT product: it has a constant value of around −4 ^∘C for extended periods in November and December. Differences in snow thicknesses (Fig. 5c) mirror differences in the ice thickness, because SMOS-SIT assumes an empirical piecewise linear relationship between the two (Tian-Kunze et al., 2014). Furthermore, sensitivity studies by Maaß (2013) suggest that the decrease in TB could be the result of the sea ice becoming fresher at a different rate than assumed by the empirical rate assumed by SMOS-SIT. Testing all these hypothesis in detail is beyond the scope of this paper, because neither does SMOS-SIT deliver the assumed sea-ice salinity as part of the data product, nor does the ORAS5 sea-ice model have a good treatment of ice salinity. Further investigation should be undertaken, and we suggest that the assumed sea-ice salinity be made part of the SMOS-SIT data product.

Despite the uncertainties on a local scale discussed in the previous sections, there is good agreement in the large-scale distribution of thin (< 1 m) sea ice and its interannual variability. Figure 6 shows time series of the area covered by sea ice with thickness above various thresholds in November from 2011 to 2017. The uppermost curve is the area of sea ice with at least 0.1 m thickness. The 0.1 m curve corresponds quite well to the NSIDC sea-ice extent if the observational gap around the North Pole is taken into account. The lowermost curve is the area of sea ice with at least 0.9 m thickness.

The overall magnitude, variability and trend of the area for the various ice thickness thresholds generally agree quite well between ORAS5 and SMOS-SIT. The extreme summer minimum in 2012 is visible as the sea-ice area reduces in November for all thickness classes. In 2013, there was a marked recovery. Since then, there has been a downward trend in all classes, with a small uptick in November 2017. Importantly, this indicates that the well-established summer sea-ice decline in recent years has started to affect the wintertime state. These signals of interannual variability are in good agreement with ice volume estimates derived from CryoSat2 radar altimetry (Tilling et al., 2015).

It is important to recall that, in the thickness range 0.9 m and above, SMOS-SIT relies heavily on auxiliary fields to retrieve the sea-ice thickness from SMOS brightness temperature. To produce Fig. 6 it was necessary to consider all SMOS-SIT data points, even those with high uncertainty and/or saturation ratio close to 100 %. As shown in Appendix D, the resulting maps and scatter densities are not realistic, and one should be cautious when interpreting the lowermost curve in Fig. 6a. Nevertheless, it is encouraging to see that overall the same interannual variability and trends of thin sea-ice area are derived from ORAS5 and SMOS-SIT.

Interannual variability and trends in sea ice in the Arctic do not occur in synchrony in different regions. Figure 6 shows November conditions, when sea ice is present not only in the central Arctic Ocean, but also in the adjacent seas, in the Canadian Archipelago, Baffin Bay, Labrador Sea and Hudson Bay. All these regions are exposed to regional climate variability and change that is not necessarily aligned: the Barents, Kara and Laptev seas are heavily influenced by the North Atlantic inflow. In the East Siberian, Chukchi and Beaufort seas the role of the North Atlantic diminishes, and other processes related to the Siberian High and Pacific climate are more important.

In the East Siberian, Chukchi and Beaufort seas (Fig. 7a, b), interannual variability of area cover is higher for thicker ice than it is for thinner ice. This feature is detected by both SMOS-SIT and ORAS5; it is more pronounced in ORAS5, where the area covered by ice thicker than 0.7 m more than doubled between 2012 and 2013 and then decreased in each subsequent year to reach the same level as 2012 in 2016.

Figure 7Monthly November means of the regional area covered by ice thicker than given thresholds in SMOS-SIT (left) and ORAS5 (right). The boundaries of the longitude–latitude boxes are 0–150^∘ E, 70–90^∘ N for panels (a) and (b); 150^∘ E–120^∘ W, 70–90^∘ N for panels (c) and (d); and 120–70^∘ W, 55–83^∘ N for panels (e) and (f).

Download

The Barents, Kara and Laptev seas (Fig. 7c, d) also exhibit a strongly reduced area coverage in 2012 for all thickness categories. However, ice cover continued to increase until 2014, by which time the area covered was almost twice as high as 2012 in some categories. The unusually high area cover in 2014 might at least in parts be due to an unusual circulation in autumn 2014: anomalously high pressure over Scandinavia combined with low pressure over Siberia in September–November led to anomalous high northerly components in the winds in these seas, which would have both encouraged thermodynamic ice growth and spreading of the ice by advection.

Another interesting feature in the Barents, Kara and Laptev seas is the increasing area of ice thicker than 0.9 m simulated by ORAS5. The year-to-year changes in thicker ice area as seen by SMOS-SIT are very different, but we would advise caution when interpreting the SMOS-SIT time series for these thicker ice categories for the reasons detailed in Appendix D.

Finally, in Canadian waters, Baffin Bay and the Labrador Sea (Fig. 7e, f), no decrease in ice area for any category is detected, neither by SMOS-SIT nor by ORAS5. Relative year-to-year variations in ice area also tend to be much smaller than in the other two areas.

The consistency in the time series presented in this section demonstrates that large-scale variability and trend of thin sea ice early in the freezing season can be monitored by both SMOS-SIT and ORAS5 with relative confidence. Both products indicate that year-to-year variability in the pan-Arctic area of thin sea ice is currently strong enough to mask any expected negative trend, and that different regions show distinct – even opposed – variability and trends. These can be related to specific regional anomalies in atmospheric circulation and surface conditions for any given year.

In light of the previously discussed shortcomings and uncertainties both in the current version of the SMOS-SIT data and the current version of the ocean reanalysis, we suggest proceeding with caution. It is clear that there is a generic trend for analysed sea ice to be thicker than what is retrieved from SMOS. Indications are that both problems in the model and in the observations contribute to this.

For the model, an overly simplistic treatment of open-water sea-ice growth (see Sects. 2.2 and 3 and Smedsrud and Martin, 2015) leads to overestimation of ice thicknesses during the freeze-up season (October–December). Later in winter, the reanalysis is mostly incapable of simulating the polynyas and fracture zones present in the interior of the ice pack.

For the observations, low sensitivity of the SMOS brightness temperatures for ice thicknesses larger than 0.5 m is compensated for in the SMOS-SIT retrieval algorithm by heavily relying on auxiliary fields from external sources, such as 2 m temperature and winds, sea-ice salinity and snow thickness on sea ice. These have considerable and poorly quantified uncertainties associated with them (e.g. Bauer et al., 2016), which leads to the uncertainty in the retrieved ice thickness. For ice thicknesses below 0.5 m, the assumption of 100 % sea-ice concentration becomes questionable.

The previous example illustrates that reanalysis–observation departures have several distinct root causes, and future data assimilation studies using SMOS should treat each of the following scenarios differently:

1. The model over- or underestimates large-scale ice thickness in the areas of first-year ice. Overestimation is typical in October–December in the Arctic Shelf seas. Sea-ice thickness as derived by SMOS is within the range of the unconstrained sea-ice model, so that data assimilation will unequivocally provide a better estimate of the truth than the model or observations alone.

2. SMOS-SIT systematically underestimates ice thickness. We argue that this typically occurs in Baffin Bay and Labrador Sea during late winter. Assimilating SMOS-SIT data here would deteriorate the simulated state. We would argue that the quality of the observational product in this region needs to be improved before it is used for data assimilation.

3. SMOS-SIT detects the presence of thin ice in fracture zones and polynyas, but there are fundamental structural deficits in the reanalysis (see discussion in Sect. 3) that prevent it from simulating them. Here, SMOS-SIT can contribute to model validation and improvement. Assimilating SMOS-SIT data would lead to a better state estimate but would force the model outside the range of states it would normally occupy. Assimilation is probably beneficial for arriving at better state estimates and initial conditions, but an investigation is needed to ensure no undesired unphysical side effects are triggered during the assimilation.

With further progress in the retrieval algorithms and the modelling for thin sea ice, the distinction between the above three departure scenarios might become obsolete, and unqualified use of the data for model validation and data assimilation will become possible without the need for manual intervention and interpretation. Until then, we suggest using SMOS-SIT data as a means of detecting the presence of thin sea ice and designing data assimilation studies with the above three departure scenarios in mind.

The SMOS-SIT data are publicly available from https://icdc.cen.uni-hamburg.de/1/daten/cryosphere/l3c-smos-sit.html (Tian-Kunze et al., 2016). The ORAS5 data will be publicly available from the Copernicus Marine Environment Monitoring Service under http://marine.copernicus.eu/ (last access: 12 June 2018).

The authors declare that they have no conflict of interest.