Brage NP

Gabrielsen, Geir Wing

Jørgensen, Even H.

Evenset, Anita

Kallenborn, Roland

English

Report from the AMAP Conference and Workshop Impacts of POPs and Mercury on Arctic Environments and Humans

Hop, Haakon

Falk-Petersen, Stig

Spatial and temporal variability of the ice-ocean system in the Marginal Ice Zone of the Barents Sea

Brodersen, Christopher

Jaklin, Gunn Sissel

Lønnum, Lasse

Orheim, Olav

Ramseng, Arnhild

Norwegian

75 år for Norge: Norsk Polarinstitutt jubilerer

Anker-Nilssen, Tycho

Bakken, Vidar

Strøm, Hallvard

Golovkin, Alexander N.

Bianki, Vitali V.

Tatarinkova, Iverta P.

Sostojanie populjacij morskich ptic, gnezdjascichsja v regione Barenceva morja

Fuglei, Eva

Øritsland, Nils Are

Energy cost of running in an Arctic Fox, Alopex Lagopus

Shapiro, Inna

Colony, Roger

Vinje, Torgny

5Shapiro et al.: Polar Research 22(1), 5–10
April sea ice extent in the Barents Sea, 1850–2001
Inna Shapiro, Roger Colony & Torgny Vinje
Three observational data sets are used to construct a continuous record 
(1850–2001) of April ice edge position in the Barents Sea: two sets of 
Norwegian ice charts (one from 1850 to 1949 and the other from 1966 
to 2001) and Soviet aircraft reconnaissance ice extent charts from 1950 
to 1966. The 152-year April ice extent series is subdivided into three 
sub-periods: 1850–1899, 1900–1949 and 1950–2001. For each of these 
study sub-periods, a mean April ice edge and a set of anomalies (differ-
ences in position between a given April and the mean April ice edge) are 
computed. The calculations show the mean ice edge position retreated 
north-eastward over the 152-year period, with the greater retreat seen in 
the changes from the 1850–1899 sub-period to the 1900–1949 sub-period. 
The distribution of the standard deviation of the ice edge anomaly over 
the linear distance along the mean ice edge shows no substantial differ-
ence between any of the three periods of the study. Within each study 
period, the maximum variation is observed in the sector bounded by the 
25° E and 49° E meridians, which covers the main pathway of the warmer 
water fl ow from the Norwegian Sea.
I. Shapiro, R. Colony & T. Vinje, International Arctic Research Center, University of Alaska Fairbanks, 930 
Koyukuk Drive, Box 757335, Fairbanks, AK 99775, USA, inna@iarc.uaf.edu.
In nature, ice edge position is sometimes well-
defi ned and sometimes ill-defi ned. After persist-
ent on-ice wind, a few tens of metres can separate 
compact ice cover from the open sea. During ice 
melt and with off-ice wind, loose ice can be scat-
tered over several hundred kilometres. The sci-
entifi c literature often contributes to confusion 
regarding a defi ned ice edge, and for many prob-
lems we are better served by reference to a mar-
ginal ice zone. For climate research, there may be 
merit in regarding ice edge position as a primary 
variable and defi ning appropriate statistics.
In this study, we defi ne ice edge as the iso pleth 
of 30 - 40 % ice concentration. On one side of 
the boundary less than 30 - 40 % of the sea sur-
face is covered by ice; on the other side more 
than 30 - 40 % of the sea is covered. The 30 % ice 
coverage has historical relevance: it is the maxi-
mum ice cover in which sailing vessels can easily 
manoeuvre.
Using a combination of ship, aircraft and satel-
lite observations, we estimate the mid-April ice 
edge position for the period 1850–2001. The ice 
edge position is then regarded as a random vari-
able, and we are provided with 152 realizations. 
Defi nitions of the mean and variance are devel-
oped and sample statistics are estimated.
Data
Our study uses three data sets to construct a con-
tinuous observational record of April ice edges in 
the Barents Sea from 1850 to the present. Nor-
wegian ice charts comprise two of the data sets: 
the fi rst from 1850 to 1949, and the second from 
1966 to 2001. Soviet reconnaissance aircraft col-
lected Barents Sea ice edge data for the interven-
ing years, 1950–1966.
The study’s geographic domain is limited to the 
Barents Sea, a region of the Arctic Ocean extend-
ing from Svalbard to Novaya Zemlya and south-
6 April sea ice extent in the Barents Sea
ward to the Kola Peninsula, roughly bounded by 
5° to 60° E in longitude and 67° to 80° N in lati-
tude. Ice edge positions are defi ned by a set of 
geographic coordinates, latitudes and longitudes, 
specifying the observed points along the ice edge. 
Data coverage, especially in the early part of the 
study period, is insuffi cient for daily or weekly 
analysis. The twice-monthly data are suffi cient 
for analysis of the mid-month ice edge position. 
Furthermore, we select the April mean ice edge 
position for study of interannual ice edge varia-
bility, because the March/April ice edge extent 
has been viewed traditionally as the annual maxi-
mum ice edge extent (Vinje 1985).
Norwegian charts, 1850–1949
The Norwegian Polar Research Institute (NPI) 
organized and charted comprehensive observa-
tions of ice edge position gathered from annual 
March-to-September sealing and hunting expe-
ditions in the Nordic seas. Starting in the early 
1850s, ships’ logbooks formed the record of the 
ice edge positions, and made limited descriptions 
of sea ice type near the ice edge. The ice edge 
data and ice description were processed digitally 
at NPI. Figure 1 illustrates data from a typical 10-
year record from 1880 to 1889.
For the 1850–1949 period, the ice extent obser-
vations for each year are combined into a bi-
weekly ice map series, and each series is typi-
cally assigned the date of the last day within the 
corresponding fortnight. Figure 2 illustrates two 
such composite ice maps for 1912: one is dated 
8 April and the other 22 April. In both cases the 
plot shows two ice extent lines, running relatively 
close to each other. The inner line bounds rough-
ly the 7/10 ice concentration, and the seaward 
boundary corresponds to an approximate 3/10 ice 
concentration. The ice edge shown on this fi gure 
is a combination of several ice extent lines, each 
corresponding to a different ice extent observa-
tional record. These lines sometimes overlap, and 
sometimes have small gaps of several tens of kil-
ometres between them. In our calculations, we 
use the seaward ice boundary as the April ice 
extent, combine the lines from different records 
into a single ice extent line and, for anomaly cal-
Fig. 1. Spring to autumn monthly ice extent for 1880–1889, reconstructed from observations of ice edge position made by sealing 
and whale-hunting expeditions.
7Shapiro et al.: Polar Research 22(1), 5–10
culations, we linearly interpolate the small gaps 
between lines from different records.
Soviet aircraft reconnaissance, 1950–1966
The ice-covered seas played a strategic role for 
Soviet national security during the Cold War, and 
Soviet Air Force aircraft charted the ice edge on 
a regular basis. The Soviet aircraft reconnais-
sance ice edge data typically consist of a single 
ice extent line. Figure 3 is an example of a com-
posite ice chart for 1962 and is a typical record of 
Soviet Air Force ice edge observations contained 
in the Environmental Working Group joint U.S.–
Russian sea ice atlas (Arctic Climatology Project 
2000).
Norwegian ice charts, 1966–2001
Modern ice charts provide high resolution data 
for ice edge position and sea ice type. Such charts 
have been compiled routinely by the Norwe-
gian Meteorological Institute since 1966, the US 
Navy/NOAA Joint Ice Center (now the Nation-
al Ice Center) since 1972, and the UK Meteoro-
logical Offi ce. These three ice map series were 
intercompared for quality control and to obtain 
a consistent modifi ed series for the NPI digitiza-
tion work.
High quality, all-weather satellite data have 
been available since 1979. Today’s ice charts are 
produced using data from ships and aircraft, as 
well as the 1.5 km and 25 km resolution satellite 
data. Figure 4, an ice map for 18 April 1988, is 
Fig. 2. April 1912 ice edge position—an example of the ice 
extent measurements based on ships’ logbook records and 
reconstructed digitally at the Norwegian Polar Institute.
Fig. 3. 1962 monthly ice extent from Soviet aircraft observa-
tions.
Fig. 4. Ice chart for 18 April 
1988—an example of a modern 
Norwegian ice chart using high 
resolution satellite data as well 
as data from ships and aircraft. 
Ice concentrations are as follows: 
very close drift ice (9/10 - 10/10); 
close drift ice (7/10 - 9/10); open 
drift ice (4/10 - 7/10); very open 
drift ice (1/10 - 4/10); and open 
water (0/10 - 1/10).
8 April sea ice extent in the Barents Sea
to the 90° E meridian. It is parameterized by arc 
length distance, s:
x = x (s) = [x (s), y (s)]                (1).
Here, s is the distance (km) measured along the 
ice edge from the west coast of Svalbard to the 
Kola Peninsula.
We defi ne the annual anomaly of ice edge loca-
tion, ϕ (s), as the perpendicular distance from the 
observed ice edge to the mean at sample points 
along the mean ice edge on 25 km spacing. By 
convention, positive anomalies are associated 
with increases in ice extent, while negative anom-
alies are associated with decreases in ice extent. 
The scalar ice edge anomaly is easily recognized 
when the observed ice edge and the mean ice edge 
are roughly parallel. For cases where the perpen-
dicular from the observed ice edge to the mean 
intersects the observed ice edge at more than one 
point (e.g. where the observed ice edge contains 
a peninsula or salient extending from the gener-
al trend of the ice edge), the anomaly is defi ned 
as the minimum value of this perpendicular dis-
tance. Figures 5 and 6 illustrate an example, the 
April anomaly for 1924. In this example, the dis-
tance ϕ (s) is the shortest distance from the 
observation to the mean and defi nes the posi-
tion along the mean. Two sample points have 
been selected to illustrate the anomaly distance, 
the fi rst point, m, at s = 350 km and the second 
point, k, at s = 1150 km. The graph of anomaly 
by distance for the April 1924 case is shown in 
Fig. 6, and the sample points of Fig. 5 are labelled 
together with their corresponding anomalies.
Fig. 5. Mean April ice edge, 1900–1949 (red dotted line) and 
observed ice edge, April 1924 (blue line).
an example of a modern ice chart. Ice types are 
described as one of six categories in a continu-
um (ice concentration in parentheses): open water 
(0/10 - 1/10); very open drift ice (1/10 - 4/10); open 
drift ice (4/10 - 7/10); close drift ice (7/10 - 9/10), 
very close drift ice (9/10 - 10/10); and fast ice. A 
description of the NPI ice charts series is given in 
Vinje (2001).
Working with contemporary Norwegian ice 
charts for the 1966–2001 period, we used the 
boundary between open drift ice and very open 
drift ice—a 4/10 ice concentration line—as the 
April ice extent line.
Statistical technique
Our study’s underlying premise is that direct 
observations of sea ice extent are suitable for 
quantitative study and standard statistical tools 
may be used to test plausible hypotheses. We treat 
each annual April ice edge position as an inde-
pendent sample estimating the true mean ice edge 
for the month. The total 152-year observational 
record is subdivided into three sub-periods, each 
spanning approximately 50 years: 1850–1899, 
1900–1949 and 1950–2001. We defi ne anomalies 
as differences in position between a given April 
and the ensemble mean of all April ice edges over 
the corresponding study sub-period, and we com-
pute fi rst- and second-moment statistics for the 
set of observed April ice edges.
Observed ice edge as a random line
For each year of the 152-year observational 
record, the April ice edge is described by the avail-
able set of observed latitudes and longitudes. In 
some years, there is more than one set of April ice 
edge observations. In these cases, the mean April 
ice edge for the year is selected as the complete 
set of ice edge observations in April closest to 15 
April. In cases where two sets of ice edge obser-
vations are equally close to 15 April, the later set 
is selected.
Postulated mean, parametric representation, 
and defi nition of perpendiculars
The mean ice edge position is a vector function in 
a rectangular coordinate system with the origin 
at the North Pole, x axis corresponding to the 
Greenwich meridian, and y axis corresponding 
_       _
9Shapiro et al.: Polar Research 22(1), 5–10
The anomaly is regarded as a random variable. 
The anomaly, ϕ (s), has the advantage of being a 
scalar function of the scalar distance. The mean 
monthly ice edge over the study period must 
therefore satisfy the condition:
E{ ϕ (s) } = 0                           (2).
April mean ice edge
Iteration of mean and importance of the 
smooth mean
April ice edges over the three study sub-peri-
ods (1850–1899, 1900–1949 and 1950–2001) are 
plotted for purposes of estimating a provisional 
mean. Based on the provisional mean, the set of 
anomalies for each study sub-period is estim ated. 
Subject to the condition of Eq. (2), the mean is 
re-estimated and the new set of anomalies is re-
calculated. This approach combines the advan-
tage of a reasonable representation of the mean 
ice position with preservation of the statistical 
integrity and robustness of the data.
The smoothing technique uses a fi rst-guess 
April mean ice edge for each study sub-period. 
After anomalies are computed, the condition 
of Eq. (2) is imposed, and the mean ice edge is 
adjusted to satisfy the condition. Several itera-
tions of this process are needed typically to arrive 
at an ice edge estimate that meets the zero-anom-
aly condition. A Savitzki-Golay smoothing fi lter 
(also called digital smoothing polynomial fi lter or 
least-squares smoothing fi lter) is used to make the 
ice edge relatively smooth (Orfanidis 1996).
Additionally, several trials are performed with 
different fi rst-guess ice edges to test the robust-
ness of the technique. It is found that the resulting 
mean ice edge solutions are nearly identical after 
the conditions of Eq. (2) are imposed, regardless 
of which ice edge position is used as a fi rst guess. 
The mean ice edge positions for each study sub-
period are shown in Fig. 7. We note here that the 
mean ice edge position appears to have retreated 
north-eastward over the 152-year overall period, 
and that the retreat is greater from the 1850–
1899 sub-period compared to the 1900–1949 
sub-period, then from the 1900–1949 sub-period 
compared to the 1950–2001 sub-period.
Variance of the ice edge
We defi ne the variance of the ice edge position:
σ 2 (s) ≡ var { ϕ (s) } = E {ϕ2 (s) }, or
σ 2 = — ∫ var { ϕ (s) } ds                   
(3)
where L is the path length along the ice edge (in 
km). The spatially homogeneous variances are 
σ 2 = 2.48 × 104 km2 (1850–1899), σ 2 = 1.43 × 104 
km2 (1900–1949) and σ 2 = 1.3 × 104 km2 (1950–
2001).
Figure 8 shows the standard deviation (the 
square root of the variance, computed in Eq. [3]) 
of the ice edge anomaly from the mean ice edge 
for each of the three sub-periods of the study. 
Figure 8 suggests little variation in the standard 
deviation of the anomaly between different study 
Fig. 6. Ice edge anomaly, April 1924. Fig. 7. Mean April ice extent: 1850–1899, 1900–1949 and 
1950–2001.
1
L
10 April sea ice extent in the Barents Sea
sub-periods, indicating that although the mean 
ice edge changes location from one study period 
to another, the variance of the ice edge within 
each study sub-period is essentially unchanged. 
Within each study sub-period, the least varia-
tion is in the beginning segments (near Svalbard) 
and the ending segments (near the Kola Penin-
sula), with maximum variation between 800 and 
1600 km along the ice edge. This zone of maxi-
mum variance corresponds to the sector bounded 
by the 25° and the 49° E meridians, and covers the 
main pathway of the throughfl ow of the warmer 
Atlantic water from the Norwegian Sea (Loeng 
et al. 1997).
Conclusions
The analysis of the continuous observational 
record of April ice edge in the Barents Sea shows 
the retreat in the mean ice edge position from 
1850 to present. The 152-year record is comprised 
of three study periods, each approximately 50 
years long, and comparison between these peri-
ods indicate largest retreat in ice edge from the 
pre-industrial (1850–1899) era to the fi rst half of 
the 20th century. Ice in the Barents Sea continues 
to retreat during the last half of the 20th century.
We conclude that the distribution of the stand-
ard deviation of the ice edge anomaly from the 
mean ice edge over the linear distance along the 
mean ice edge shows no substantial difference 
between any of the three periods of the study, 
while the largest variability of standard deviation 
within each study period corresponds to the area 
of warm Atlantic water infl ow to the Barents Sea 
from the Norwegian Sea.
Acknowledgments.—This work was supported by the Frontier 
Research System for Global Change. The authors thank H. 
Goodwin for help with the data set, T. Shy for helpful sugges-
tions and two reviewers for their constructive comments.
References
Arctic Climatology Project 2000: Environmental Working 
Group joint U.S.–Russian sea ice atlas. F. Tanis & V. Smo-
lyanitsky (eds.). CD-ROM. Ann Arbor, MI: Environmen-
tal Research Institute of Michigan in association with the 
National Snow and Ice Data Center.
Loeng, H., Ozhigin, V. & Adlandsvik, B. 1997: Water fl uxes 
through the Barents Sea. ICES J. Mar. Sci. 54, 310–317.
Orfanidis, S. J. 1996: Introduction to signal processing. Eng-
lewood Cliffs, NJ: Prentice-Hall.
Vinje, T. 1985: Drift, composition, morphology and distri-
bution of the sea ice fi elds in the Barents Sea. The physi-
cal environment, western Barents Sea. Nor. Polarinst. Skr. 
179 C. Oslo: Norwegian Polar Institute.
Vinje, T. 2001: Anomalies and trends of sea-ice extent and 
atmospheric circulation in the Nordic seas during the 
period 1864–1998. J. Clim. 14, 255–267.
Fig. 8. Standard deviation of the ice edge anomaly: (a) April 
1850–1899; (b) April 1900–1949; and (c) April 1950–2001.
(a)
(b)
(c)


April sea ice extent in the Barents Sea, 1850-2001

Divine, Dmitry V.

Korsnes, Reinert

Makshtas, Alexander P.

Variability and climate sensitivity of fast ice extent in the north-eastern Kara Sea

Ahlstrøm, Øystein

Mydland, Liv Torunn

Comparative nutrient digestibility of arctic foxes (Alopex lagopus) on Svalbard and farm-raised blue foxes (Alopex lagopus)

Urset, Anne

The Place names of Svalbard

Zyryanov, Denis

Haarpaintner, Jörg

Numerical modelling of the Storfjorden (Svalbard) polynya development due to wind stress : role of the sea ice rheology and damping forces

Not a member yet