Thin-ice dynamics, polynya area and thermodynamic ice production in the Arctic for 2002/2003 to 2014/2015
Interannual average values for TIT ≤ 0.2 m are listed in
Table for each polynya region. They range between
6.8 cm (WNZ) and 16.3 cm (GLN), with an overall average of
about 12.7±0.6 cm. The underlying long-term time series of
average winter TIT within each polynya (not shown) reveal a tendency
towards decreasing thin-ice thicknesses in almost every region (e.g., up to
2.5 cm per decade in the Storfjorden polynya), with the only
exceptions being the CAA, GLN and NEW.
Monthly thin-ice frequencies, calculated per pixel as the fraction of days
with a TIT ≤ 0.2 m relative to the 13-year investigation
period, are presented in Fig. . Frequencies of larger than 0.5
are primarily found around the Canadian Arctic, first and foremost in the NOW
polynya and the eastern CAA. More specifically, coastal areas around Devon
Island and southwestern Ellesmere Island (Hells Gate, Cardigan
Strait) and larger areas at the
eastern exits of Lancaster Sound and Jones Sound are well visible and have
previously been related to tidal currents and slightly increased ocean heat
fluxes . Other areas with similar magnitudes
include the Storfjorden polynya and coastal areas (north)west of Novaya
Zemlya. In addition, elongated thin-ice areas along the Siberian shelf
(Laptev and Kara seas, frequencies around 0.05 to 0.35 each month) are well
delineated. Locations of frequent thin-ice occurrences in the Kara Sea are in
accordance with results from the study of . The northern
Barents Sea, Franz Josef Land and the Svalbard Archipelago also feature quite
high appearance rates of around 0.1 to 0.3. Contrary to earlier reports
, the Northeast Water (NEW) polynya in northeastern
Greenland (approx. 81∘ N, 13∘ W) neither shows any signs of
disappearance, nor is it limited to the spring to late autumn period. With
average frequencies of around 0.1 to 0.25 each month in winter, it is more
likely to be categorized as a regularly forming polynya. Comparatively low
frequencies below 0.15 (especially from January to March) are primarily found
in the Beaufort and Chukchi seas as well as in the East Siberian Sea. Vast
fast-ice areas, e.g., along the Siberian coast, can be detected from monthly
TIT frequencies, as these areas usually appear at fixed locations attached to
the shore and TIT frequencies tend towards zero as the ice quickly thickens
by congelation ice growth. Hence, our 13-year record of monthly
TIT-occurrence rates offers the potential to further develop optimized
automatic methods for a regular Arctic-wide mapping of monthly fast-ice
extents and could thereby complement currently existing approaches from
earlier studies e.g.,.
Compared to the study of , leads are only weakly visible
in these long-term averages (frequencies below 0.05–0.1). In
Fig. , leads are mainly located in the area of the Beaufort
Sea and north of Greenland (shear zones), which can be mainly attributed to
their relatively high spatial and temporal persistence. Frequent lead
occurrences in the East Siberian Sea found by , for
example, are not reflected in our study. In some regions, however, the
influence of (shelf) bathymetry and associated ocean currents on the spatial
distribution of polynya and lead occurrences is also visible in the TIT
frequencies derived here (e.g., eastern exit Vilkitsky
Strait, Hanna Shoal on the northern Chukchi Shelf, northern ESS).
Regional time series of the annual average polynya area
(POLA; TIT ≤ 0.2 m) in km2 from 2002/2003 to 2014/2015,
together with a seasonal comparison (November to December vs. January to
March) and a linear trend estimation. The estimated linear trend (in
km2 yr-1), its p value and the interannual average POLA (in
km2) are additionally listed in each subpanel. Please note the
varying scale on each y axis.
Average polynya area (POLA) in km2 in each polynya region
between 2002/2003 and 2014/2015 (SFR cloud-cover correction applied). Aside
from being based on the available winter period from November to March, it is
further separated between the early freezing season (November to December)
and the late freezing season (January to March). All values are derived from
daily MODIS TIT composites after application of the predefined polynya masks
(Fig. ). Trends are additionally given, where underlined, bold
and bold italic numbers denote statistical significance (two-sided t test) at
the 90, 95 and 99 % levels, respectively.
November to March
November to December
January to March
Avg. POLA
Trend POLA
Avg. POLA
Trend POLA
Avg. POLA
Trend POLA
Region
(103km2)
(km2yr-1)
(103km2)
(km2yr-1)
(103km2)
(km2yr-1)
Beaufort Shelf (BSH)
2.7 ± 2.4
-13
5.5 ± 5.8
132
0.8 ± 1.1
-111
Canadian Arctic Archipelago (CAA)
24.9 ± 5.1
648
41.5 ± 8.3
958
13.7 ± 5.2
438
Cape Bathurst (CBP)
10.4 ± 7.6
205
20.6 ± 16.9
581
3.6 ± 2.0
-49
Chukchi Sea (CHU)
3.7 ± 2.2
-50
8.2 ± 5.7
-95
1.3 ± 1.1
-37
East Siberian Fast Ice (ESF)
1.1 ± 0.7
110
2.3 ± 1.3
200
0.4 ± 0.5
48
East Siberian Sea (ESS)
5.7 ± 2.1
175
10.3 ± 5.2
385
2.5 ± 1.1
33
Franz Josef Land (FJL)
12.9 ± 6.0
850
15.2 ± 9.7
1565
11.5 ± 4.9
380
Greenland North (GLN)
0.6 ± 0.5
28
0.8 ± 0.6
14
0.5 ± 0.4
37
Kara Sea (KAR)
40.2 ± 15.8
1839
64.9 ± 26.8
3209
23.5 ±15.5
904
Laptev Sea (LAP)
12.1 ± 4.2
845
17.0 ± 8.0
1559
8.8 ± 2.7
362
Northeast Water (NEW)
2.8 ± 0.8
-37
3.3 ± 1.0
-10
2.4 ± 0.9
-55
North Water (NOW)
30.3 ± 6.7
711
37.7 ± 7.1
1072
25.4 ± 7.8
464
Nares Strait–Lincoln Sea (NSL)
4.3 ± 2.3
-6
5.7 ± 2.6
18
3.4 ± 2.8
-22
Storfjorden (STO)
3.3 ± 1.0
139
3.9 ± 1.1
175
2.9 ± 1.3
115
Svalbard Archipelago (SVA+STO)
29.9 ± 4.9
488
32.8 ± 6.5
732
27.9 ± 5.3
327
Severnaya Zemlya North (SZN)
2.6 ± 1.5
263
3.4 ± 2.2
353
2.1 ± 1.2
203
Western Novaya Zemlya (WNZ)
42.3 ± 11.1
-588
40.4 ± 13.8
-1806
43.6 ± 13.0
230
Total
226.6 ± 36.1
5468
309.4 ± 62.6
8864
171.3 ± 32.6
3151
Regional time series of the annually accumulated ice
production (IP) in km3 from 2002/2003 to 2014/2015, together with a
seasonal comparison (November to December vs. January to March) and a linear
trend estimation. The estimated linear trend (in km3 yr-1), its
p value and the interannual average IP (in km3) are additionally
listed in each subpanel. Please note the varying scale on each y axis.
In Figs. and , the interannual
variability of the average POLA (in km2) and accumulated IP (in
km3) is presented for all examined polynya regions. In both figures,
the difference between the beginning (November to December) and end (January
to March) of the freezing (winter) period is additionally highlighted.
Concerning POLA, it shows that the largest average winter extents are found
in the NOW, WNZ and KAR areas. The study of
demonstrated that the large POLA values in the NOW region are part of a
(non-significant) long-term increase of average polynya extents between 1978
and 2015. In case of polynyas in the proximity of Novaya Zemlya, our average
winter value for POLA of around 42×103 km2 is fairly close
to the respective value from (49×103 km2), despite the circumstances that their study covered an
extended winter period from September to May and featured a different mask
area, which stretched over some part of the western Kara Sea. As mentioned
earlier, presented POLA values for the Kara Sea. His
retrievals are based on approximately the same reference area
(Fig. ), which in this case allows for a fair comparison to
the numbers presented here. It shows that the average POLA in the late
freezing season reveals similar magnitudes in recent years. During the period
from 1979 to 2004, the average POLA (in : January to April)
ranged between 1 and 5×104 km2 (except for 1995, which was
around 6×104 km2), which is close to the range of the
results presented here for January to March (Fig. ;
Table ). Although the estimated positive trend in POLA
remains non-significant for the Kara Sea as in , the
magnitude of the trend in the late freezing period (January to March, around
9000 km2 decade-1) seems to have increased from
2400 km2 decade-1 to around
9000 km2 decade-1 over the last 13 years. The interannual POLA
variability in all regions is generally pronounced, but it is especially
large for smaller polynyas and/or thin-ice regions such as the NSL, NEW and ESS. Concerning
seasonal differences, it appears that some regions (e.g., NEW, GLN, LAP, SZN)
have the tendency towards larger thin-ice areas during the freeze-up period
since the time period of approximately 2006/2007 to 2007/2008. About 8 to 10
polynya regions show distinct positive trends of up to 18 390 km2
per decade (KAR), with only the LAP, ESF and SZN regions being significant
(two-sided t test) with p≤0.05. Interestingly, subregions located in
the proximity of the Beaufort Gyre (BSH and CBP) indicate very large thin-ice
areas between November and December 2007, shortly after the second lowest
September sea-ice extent since 1979 (approx. 4.7 million km2;
). This did not appear in a similar way in 2012
(record low of approx. 3.4 million km2). A detailed investigation
shows that the freeze-up in the Beaufort Sea area was much slower in 2007 and
extended until mid-December, while in 2012 the same area was ice-covered by
10 November. The study of linked this significant
delay in ice growth to upward mixing processes of ocean heat in the Canada
Basin, originating from the release of stored solar heat input following
summer 2007. This resulted in large areas with very thin ice (around
170 000 km2) from November to December and consequently allowed for
huge amounts of latent and sensible heat to be released from the ocean,
leading to extraordinary high IP values in these areas
(Fig. ).
Average accumulated ice production (IP) in km3 in each
polynya region between 2002/2003 and 2014/2015 (SFR cloud-cover correction
applied). Aside from being based on the available winter period from November to
March, it is further separated between the early freezing season (November to
December) and the late freezing season (January to March). All values are
derived from daily MODIS TIT composites after application of the predefined
polynya masks (Fig. ). Trends are additionally given, where
underlined, bold and bold italic numbers denote statistical significance
(two-sided t test) at the 90, 95 and 99 % levels, respectively.
November to March
November to December
January to March
Acc. IP
Trend IP
Acc. IP
Trend IP
Acc. IP
Trend IP
Region
(km3)
(km3yr-1)
(km3)
(km3yr-1)
(km3)
(km3yr-1)
Beaufort Shelf (BSH)
23 ± 23
-0.5
19 ± 23
0.0
5 ± 5
-0.4
Canadian Arctic Archipelago (CAA)
215 ± 43
5.4
136 ± 23
2.5
79 ± 31
2.9
Cape Bathurst (CBP)
78 ± 54
1.0
60 ± 48
1.1
18 ± 10
-0.1
Chukchi Sea (CHU)
27 ± 16
-0.5
20 ± 14
-0.3
7 ± 6
-0.2
East Siberian Fast Ice (ESF)
9 ± 5
0.8
7 ± 3
0.6
2 ± 2
0.3
East Siberian Sea (ESS)
41 ± 13
1.5
28 ± 13
1.3
13 ± 5
0.2
Franz Josef Land (FJL)
99 ± 41
5.6
42 ± 24
4.1
57 ± 22
1.5
Greenland North (GLN)
5 ± 4
0.2
3 ± 2
0.0
3 ± 2
0.2
Kara Sea (KAR)
277 ± 111
12.6
174 ± 76
9.0
104 ± 56
3.6
Laptev Sea (LAP)
96 ± 33
6.8
51 ± 24
4.8
45 ± 14
2.0
Northeast Water (NEW)
22 ± 7
-0.4
10 ± 3
0.0
12 ± 4
-0.4
North Water (NOW)
277 ± 67
6.0
135 ± 27
3.4
145 ± 45
2.6
Nares Strait–Lincoln Sea (NSL)
39 ± 22
0.1
20 ± 10
0.2
20 ± 16
-0.1
Storfjorden (STO)
21 ± 6
0.9
10 ± 4
0.5
11 ± 4
0.4
Svalbard Archipelago (SVA+STO)
214 ± 33
0.8
91 ± 24
1.6
123 ± 24
-0.8
Severnaya Zemlya North (SZN)
22 ± 11
2.0
10 ± 6
0.9
12 ± 6
1.1
Western Novaya Zemlya (WNZ)
367 ± 124
-7.0
136 ± 56
-6.8
231 ± 88
-0.2
Total
1811 ± 293
34.5
940 ± 178
22.4
871 ± 175
12.1
Average (2002/2003 to 2014/2015) accumulated ice production
(m per winter) during winter (November to March) in the Arctic, north
of 68∘ N. The margins of applied polynya masks
(Fig. ) are shown in black dashed lines.
Regarding IP, many of the features described above are also visible in the
regional time series of Fig. . Contrary to
, the majority of polynya regions show overall positive
(up to 126 km3 per decade (KAR)) or no trends in winter ice
production, and only four regions indicate a slight yet insignificant
decrease over the last 13 years (BSH, CHU, NEW, WNZ). Complete overviews of
calculated average POLA and IP values per region, together with their
respective trends, are given in Tables
and , respectively. These overviews highlight the huge effect that seasonal
differences (November to December vs. January to March) have on
calculated average values and trends for the complete winter period from
November to March. Consequently, the numbers discussed here should be
regarded as winter integrals with potentially inherent effects originating
from the timing of freeze-up onset. In cases such as the Kara Sea,
Franz Josef Land, the Chukchi Sea and the Canadian Arctic Archipelago, for example, large
thin-ice and potential open-water areas during the early freezing period in
November and December imprint on the total winter averages as well as derived
trends of POLA and IP, especially from 2007/2008 onwards. While the majority
of polynyas also feature positive trends in the late freezing season from
January to March, these trends are for the most part insignificant.
The average total ice production in Arctic polynyas sums up to
1811 ± 293 km3 per winter. Thus, it lies in between previously
determined average values of 2940 ± 373 km3
(; 1992/1993 to 2007/2008) and 1178±65 km3
(; 2002/2003 to 2010/2011) per winter. We expect that the
MODIS-derived quantities offer a valuable increase in both spatial and
quantitative accuracy due to the use of high-resolution and gap-filled daily
fields of thin-ice thicknesses. A shortening of the averaging interval for the
period 2002/2003 to 2010/2011 (as in , but not accounting
for differences in covered winter period) marginally reduces the average
total ice production derived here by about 1–2 %. In order to assess
apparent differences between our data set and the passive
microwave data set from , a more direct comparison based on
identical reference areas and the same winter period would be necessary.
(a) Decadal trends (m per decade) of winter
(November to March) ice production in the Arctic, north of 68∘ N.
Trends are calculated by applying a linear regression to the annual
accumulated IP per pixel for the period 2002/2003 to 2014/2015. Areas with
statistical significance (based on a two-sided t test) at the 95 and
99 % level are depicted in (b). The margins of applied polynya
masks (Fig. ) are shown as black dashed lines.
A spatial overview of the average (2002/2003 to 2014/2015) accumulated ice
production per winter (November to March) is presented in
Fig. . Similar to Fig. , the NOW polynya
stands out at first glance due to its high average ice production of up to
14 m per winter. However, smaller polynyas in the Canadian Arctic
(around Devon Island) feature comparatively high values for ice production.
Most other areas in the Arctic produce on average between 1 and 3 m of
ice per winter, with a few noticeable exceptions like Franz Josef Land (about
4–5 m per winter), the northern tip of Novaya Zemlya (5–7 m
per winter) and some coastal areas in the Kara Sea (1–4 m per
winter). While the core areas of high ice production show a high resemblance
to with marginal differences in absolute numbers, MODIS
is capable of providing enhanced spatial detail. This is especially valuable
concerning the narrow thin-ice areas along the coast and fast-ice edges in
the eastern part of the Arctic (Kara Sea, Laptev Sea, East Siberian Sea), as
these areas are not resolved by the coarse-resolution passive microwave data
(6.25 km; ). This striking advantage is also
reflected in the comparatively narrow fjords and bays or sounds around Greenland
and the Canadian Archipelago, where a high ice production of up to 3 m per
winter is found. While these observations, mostly related to differences in
spatial resolution, could explain the discrepancy described above in average
accumulated numbers to some extent (compare ), the net
effect of a lower grid size cannot be quantified here.
Spatial trends between the winter seasons 2002/2003 and 2014/2015 (November
to March) can be calculated by applying a linear regression to the annual
accumulated IP per pixel. The resulting map is shown in
Fig. a. Aside from many interesting small-scale patterns, two
main conclusions can be drawn from this spatial overview: (1) while the
trends identified in the western Arctic show no consistent pattern, large
areas of the eastern Arctic are characterized by significant (two-sided
t test; significance levels indicated in Fig. b) positive
trends that can exceed 2 m per decade and (2) we observe opposing
negative–positive IP trends along the coasts of the Laptev and Kara seas,
which could be due to changes in fast-ice extent over the 13-year period.
Decreasing fast-ice extents and durations in the eastern Arctic between 1976
and 2007 were recently described by . In addition,
analyzed the fast ice in the southeastern Laptev Sea
in more detail (1999 to 2013). While their study showed that the winter
maximum fast-ice extent (around March to April) as well as the shape and location of
the fast-ice edge did not vary significantly over the regarded time period,
they likewise presented an overall decrease in the fast-ice season
(-2.8 dyr-1) due to later
formation and earlier break-up. These described changes regarding the timing
of fast-ice formation in early winter could explain the observed structures
of positive–negative trends in the proximity of fast-ice areas.
The geographical location of the Laptev Sea in the eastern Arctic.
The applied polynya mask is marked in red, enclosing the locations of typical
polynya formation along the coast and fast-ice edge (dashed white line;
position derived from long-term thin-ice frequencies in March;
Fig. ). Flux gates from the study by at
the northern (NB) and eastern (EB) boundary of the Laptev Sea are shown in
the inset map (cyan solid lines). Bathymetric data from
(IBCAO v3.0).
In order to put these observations into context, we suppose that this
characteristic pattern of opposing trends in the western and eastern Arctic
as well as the apparently fast-ice-related structures in the Laptev Sea and
Kara Sea could be connected to a fall freeze-up that has generally appeared
later in recent years, which itself is thought
to result from a complex mixture and/or interplay of year-round
steadily increasing (2 m) air temperatures e.g.,, distinct
large-scale atmospheric patterns e.g., and the overall
downward trend of total sea-ice extent and volume in the Arctic
e.g.,. The latter implies a tendency
towards a more fragile, and thus mobile, sea-ice cover in the Arctic, with a
potentially increased sensitivity for external forcing mechanisms (i.e.,
strong winds and/or ocean currents) that are responsible for thin-ice
formation in polynyas and leads. Since the Laptev Sea is one of the main regions with highly
pronounced and significant positive trends in both POLA and IP throughout the
complete winter period, the following section will take a closer look at
polynya dynamics in the region.
Regional focus – Laptev Sea
One main advantage of the high-resolution MODIS data is the ability to
perform detailed investigations on a regional scale across the Arctic. The
grid spacing of 2 km allows for the detection of relatively fine,
delineated polynya structures and for more accurate statements about areas of
high ice production than were possible in previous studies.
The Laptev Sea was previously described as a key region to investigate
climatic changes in the Arctic shelf seas , as it is one of
the major source areas for sea-ice export into the Transpolar Drift system
. As can be seen in Fig. , the
Laptev Sea is located between the Severnaya Zemlya at the western boundary,
the Lena River delta at the southern edge and the New Siberian Islands in the east (approximately 70–80∘ N,
100–140∘ E). The water-mass composition in the Laptev Sea is
temporarily quite variable, as there is a huge freshwater inflow during the
summer and autumn period (around 750 km3 per year;
) and strong ice formation accompanied by brine
rejection in polynyas during winter . These processes
significantly alter the stratification of the upper ocean layers as well as
the salinity levels in the annual cycle. These and other recurring features
of the sea-ice and ocean environments were recently illustrated and
updated by .
During the freezing period (from roughly October to June), fast ice forms along
the coastlines of the Laptev Sea, which usually reaches its maximum areal
extent by April. The approximate location of the fast-ice edge at the end of
March is depicted in Fig. . For drifting sea ice, the
fast-ice edge forms an advanced coastline with heavy ridging occurring along
this edge during onshore wind events . The combination
of this fast-ice edge and offshore components of the mean wind patterns
enable the formation of several flaw-lead polynyas across the Laptev Sea,
which can reach widths of up to 200 km
.
The daily polynya area (TIT ≤ 0.2 m) in the Laptev
Sea region for the winter seasons 2002/2003 to 2014/2015. Values are
calculated within the margins of the applied polynya mask
(Fig. ) and saturated at a level of 6×104 km2 for a better discrimination of lower values.
When comparing previous studies dealing with ice production rates in the
Laptev Sea
,
it becomes clear that there are large differences depending on the applied
methods and various data sets. In these studies, values for the accumulated
ice production during an average winter season (“extended” winter period
from November to April) range between 55 km3 for an
approach using microwave and thermal infrared remote sensing data in
combination with atmospheric reanalysis data and 258 km3
when using a simple relationship between wind direction,
wind speed, and polynya area. Estimated
average values (September to May) from (152 km3)
and (77 km3) range in between. Although derived
for different time periods and slightly varying reference areas, these large
discrepancies highlight the relevance of applying improved, high-resolution
approaches to quantify sea-ice production.
Overview of winter (November to March) accumulated ice
production (m per winter) in the Laptev Sea region between 2002/2003
and 2014/2015.
In order to give an overview of the long-term development of thin-ice areas
(TIT ≤ 0.2 m) in the Laptev Sea, the daily POLA is presented
in Fig. . It is evident that the largest areas of
thin-ice appear on average in November and more recently also in December
(compare Table ). A tendency towards an increased duration
of these polynya events can be observed. In the winter seasons 2008/2009 and
2009/2010, large POLA exceeding 50 000 km2 were also observed in
January, and another major polynya event can be noted for mid-February 2015. A
pronounced seasonal variation is visible for the winter seasons 2004/2005,
2005/2006 and from 2010/2011 onwards, while the other years show less polynya
activity (more lengthy periods with a closed polynya; white color in
Fig. ) and overall smaller polynya extents in February
and March.
Figure shows an annual comparison (2002/2003 to 2014/2015)
of accumulated (November to March) ice production (in m per winter)
for the Laptev Sea. The highest ice production rates of sometimes more than
4 m per winter occur predominantly in the proximity of the Taymyr
Peninsula and Severnaya Zemlya (western Laptev Sea), as well as along the
southern fast-ice edge (mainly coastward of the regions with high ice
production). However, ice production in the eastern Laptev Sea (west and
north of the New Siberian Islands) shows a greater interannual variability.
Furthermore, it is striking that the position of the fast-ice edge in
Fig. is highly variable over the 13-year record (as in
Sect. 4.1, Fig. a). However, it has to be noted that certain
bands of higher ice production, especially in the southeastern Laptev Sea,
reflect the winter evolution of fast ice
compare and are hence primarily related to the
early winter period from November to December. Another interesting
observation can be made in the Vilkitsky Strait, which is located in the
western Laptev Sea south of Severnaya Zemlya (Fig. ).
The distribution of thin-ice areas contributing significantly to the total
sea-ice production in that area seems to shift westwards towards the Kara Sea
in several years (2005/2006 to 2012/2013 and 2014/2015). In some cases, the
shape of these areas resembles an arch-type pattern of an ice bridge mechanism, a
feature that commonly appears in the Nares Strait between Ellesmere
Island and Greenland, for example .
discovered that most of the ice being incorporated in the
Transpolar Drift originates from the western and central part of the Laptev
Sea. Furthermore, it is indicated that the contribution from polynyas, while
being generally small, is limited to events in the proximity of the Laptev Sea
boundaries. As noted before, the northwestern Laptev Sea shows by far the
largest contribution to the total winter ice production in the Laptev Sea
polynyas, which implies a potential significant influence on the interannual
variability of ice export during winter. In order to check this
hypothesis, we compare annual accumulated IP values to independently derived
ice-area export (IAE) values (both presented as anomalies and normalized with
their standard deviation) in Fig. for 2002/2003 to
2014/2015. IAE values are taken from the updated time series of
, where they were calculated as the integral of the
product between the eastward and northward component of the ice drift
velocity and ice concentration at the northern boundary (NB) and eastern
boundary (EB) of the Laptev Sea, respectively. Similar to a high agreement
between polynya area and across-boundary ice export ,
there is also a significant correlation between calculated ice production and
the areal ice export (r=0.69 with p=0.009).
Normalized anomalies of accumulated winter ice production (IP of
the present study, dashed line) and accumulated ice area export (IAE, solid
line) for the winter seasons from 2002/2003 to 2014/2015. IAE data are based on an
updated time series from .
The spatial overview of annual ice production (Fig. ) is
supplemented by the previously shown time series of the average winter POLA
and accumulated IP per winter (Figs.
and , respectively). Both time series of POLA and IP in
the Laptev Sea show an overall positive trend (significant with p≤0.01), which can for the most part be traced back to larger thin-ice areas
during the freeze-up period in November and December (as described above,
Fig. ). This is underlined by Tables
and , which both reveal largest average values of POLA and IP
and most significant trends during that period of winter. The average ice
production from November to March in the Laptev Sea is estimated with about
96±33 km3 (2002/2003 to 2014/2015), with a positive trend of
6.8 km3 per year. Compared to other Arctic polynyas (see
Table ), this corresponds to a share of about 5 % of the
total ice production in polynya regions.
As the relative strength of the Transpolar Drift is dependent on atmospheric
dynamics, it has previously been linked to atmospheric indices like the
Arctic Oscillation (AO) index . For the period from 1982 to
2009, the study by presented indicators for a
net strengthening of both the Transpolar Drift and the Beaufort Gyre as well
as a general increase in the Arctic ice drift speed, which is presumably
related to a decreasing fraction of thick multiyear (MY) ice. As mentioned
before (Sect. 4.1), the latter is thought to be connected to an increased
fragility and mobility of the Arctic sea-ice cover, which may have implications for pan-Arctic polynya and lead dynamics.
According to , a positive winter AO promotes both an
increased ice transport out of the Arctic Ocean through the Fram Strait and an
increased ice transport away from the Siberian coastal areas, thereby leaving
open water and thin ice that foster new ice formation. Hence, positive trends
in both POLA and IP not only fit well to the previously estimated positive
trend in IP from but also to the positive trend of 0.85×105 km2 per decade in the Laptev Sea ice area flux
. Other linkages and dependencies with the Arctic sea-ice
extent in September (annual minimum), the timing of the freeze onset and
further connections to large-scale atmospheric circulation patterns are very
likely and have been proposed by various previous studies
e.g.,.
In particular, a significant lengthening of the melt season in recent years, and
hence a later freeze-up in autumn, already seems to imprint on the derived
POLA (i.e., thin-ice area) and IP estimates in the early winter period
. In that context, increasing
atmosphere and ocean temperatures in autumn and winter were recently
reported by . These increasing temperatures comprise the potential to alter or shift
vertical temperature gradients with consequences for the surface energy
balance and ultimately IP. Furthermore, a shortened fast-ice duration and
enhanced variability of the fast-ice edge in early winter presumably influences the location of flaw leads and
consequently high ice production and brine release. Admittedly, all these
(potential) interconnections are rather complex and require more
detailed investigations that go beyond the scope of the present study. In the
context of other reported changes during the spring and summer period
, it may emerge that the overall setup for
atmosphere–ice–ocean interactions in the Laptev Sea is gradually changing
towards a new state.