Sea
ice field report from KV Svalbard Cruise 19.-31. March 2007
Johan Wåhlin (Nansen
Environmental and Remote Sensing Centre)
and
Lars H. Smedsrud (Bjerknes Centre for Climate Research)
The sea ice program included taking cores and measuring temperature,
salinnity, and ice thickness, as well as snow depths and floe size.
Measurements were also planned to include new sea ice bulk salinity,
grease and congelation ice samples. The warm temperatures during the
cruise resulted in almost no mesurements of frazil ice and thin ice, so
the main results were the measurements of the fast ice in
Freemansundet, and of the floe from wich we were working east of Edge
Øya in the Barents Sea. At the same time as K/V Svalbard was
going on its cruise around Svalbard, ESA had their spring campaigns for
calibration/validation of instruments and simulation of future
spaceborne earth observation missions. [1] This made it interesting to
do in situ sea ice and snow measurements. In particular properties
important for SAR signature and ice thickness validation.
Sea
ice sections
On the ice stations in Freeman sundet and on the ice floe in Barents
sea a total of 5 transsections were made. In these sections holes where
drilled with an even spacing in which ice thickness, ice free board and
snow thickness was measured.
The Freeman sections were on level ice, about 40-50 cm thick, with only
small differences in ice thickness, the only exception is the hole at
1400 m in the WNW section where the measurements were done at the base
of a 3m high ridge. Apart from this particular hole, the depth of the
snow cover was also fairly constant. In the Barents sea the ice
thickness differed between the two sections, the southward section
having ice about 20 cm thicker than the eastern. The snow cover along
the south section was also thicker in general.
The point of measuring ice free board is to be able to use this to
estimate the ice thickness. From the equation of an object floating in
equilibrium the theoretical value of the ice draft can be derived.
Using the measured free board and snow thickness, and densities
proposed by Wadhams [2] of water = 1024 kg/m^3, ice = 915 kg/m^3, snow
= 330 kg/m^3, the theoretical ice draft is calculated plotted together
with the measurements from the cross sections in Figure 1 and Figure 2.
The average of the measured ice thickness and the theoretical ice
thickness of every section was also calculated, the results presented
in the table below:
---------------------------------------------------------
Average ice thickness | Measured [cm] | Calculated [cm]
---------------------------------------------------------
Freeman sundet
|
|
NNE section
| 45
| 27
WNW Section
| 56
| 59
Ridged
ice
| 465
| 270
---------------------------------------------------------
Barents sea
|
|
E
section
| 38
| 52
S
Section
| 61
| 74
---------------------------------------------------------
As can be seen in this table the calculations of ice thickness from the
known free board overestimates the ice thickness in two cases,
underestimates it in two and is very close to the actual thickness in
one case. The biggest error is for the section next to the ridge, but
here only a few holes were drilled in an area of heavily deformed ice.
A much longer section would be needed to give a fair picture of
the average ice thickness.
Looking closer at the sections made in Freeman sundet it can be seen
that the WNW section shows a good correlation between theoretical ice
thickness and measured thickness, in average they just differ a few cm.
In the figure (freem_sections.jpg) it can be seen that the curve for
the calculated ice draft matches the measured draft well except for the
ridge at 1400 m and at 200 m where a high negative free board is
present. In the NNW section the
correlation is not as good, the figure shows that the biggest errors in
calculated draft also here comes from holes with a big negative free
board. One explanation to this could be that ice with negative free
board is flooded with sea water that will be absorbed by the snow lying
on it making it a lot heavier and pressing the ice down even further. A
low free board implies either a heavy load or low buoyancy, so with a
real density much higher than the used snow density of 330kg/m^3 the
calculations would give a small ice draft. Snow density measurements in
the locations of the ice cores also gave densities from 240kg/m^3 in
the uppermost snow layer to 880kg/m^3 in the flooded lower layers with
typical values of 450kg/m^3 in between.
In the Barents sea, the theoretical ice draft of the eastward section
matches the measured thickness quite well except for areas with a thick
snow cover where the calculated ice draft was too big. This implies
that the snow density of 330kg/m^3 used in the calculations was
too high. No snow density measurements were made in Barents sea, but
the snow here was far lighter than any of the snow found in the Freeman
strait, where the lowest measured value was 245kg/m^3. In areas without
snow cover the calculated draft is smaller than the measured. This
could be because the areas with bare ice were not very big, so the
surrounding snow could contribute to the load, something
that the calculations do not consider. For the southward section this
relation between snow cover and error in calculated draft is not
present. This section however cuts through ice with more rafting and
ridging that possibly affects the free board more than the snow. This
is likely the case for the hole at 60m, where the draft calculated from
free board is much deeper than the measured draft. The surrounding
ridges are here likely to give buoyancy to the entire area, giving a
high free board also to thinner ice. The big calculated error at the
hole 100m is probably due to over ice which increased the free board
without contributing with much weight.
Apart from these errors the calculated ice draft follows the measured
draft quite well, but it also shows how sensitive the method is for
changes and errors in e.g. snow density.
Sea
Ice temperature
The temperature profiles were measured in different ways in the two ice
stations. In Freeman sundet the core was lifted out of the ice and the
temperature was measured in small holes that were drilled with 5 cm
spacing from bottom to top of the core. On the Barents sea station the
temperature was instead measured by drilling a few centimetres into the
ice, putting the thermometer into the hole, measuring temperature and
how deep the hole was and then drill another couple of centimetres for
the next measurement.
This second method proved not to be optimal. In the upper part of the
ice sheet it worked well, but some way down into the ice, water started
to flow into the hole. This is because the sea ice is permeable, so
when a hole is drilled the water level in this hole start to adjust to
sea level. The water flowing into the hole originates from brine higher
in the ice flowing down and from brine below the hole getting pushed up
by the sea water. If the hole is deep enough, also sea water will enter
the hole. [3] The measurements from the point where water started
flowing in are in other words affected by the temperature regime both
above and below the level the thermometer is placed.
In Freeman sundet the air temperature was about -4 degrees Celsius 10
cm above snow surface. The temperature profiles of all cores and snow
profiles are very similar. A peak minimum is reached about 5 cm below
the snow surface, from here the temperature increases evenly to the
bottom of the ice sheet where it reaches the freezing temperature of
the water as shown in Figure
3. The minimum temperature inside the snow cover implies
that the air temperature has been colder before and that the warmer air
not yet have had time to warm the entire snow cover.
The temperature at 10 cm above snow surface in the Barents sea was
about -8 degrees Celsius. Ice core 1 and 2 were taken in areas with
very little or no snow so the air-ice boundary was as cold as the air.
From here is was a strong gradient the first 10 cm to thereafter stay
at almost constant temperature for core 1 and a small gradient for core
2, see Figure 4. Ice core 3 was
taken right next to core 4, so the temperature measurements were made
in common for both and presented in the core 4 plot in the same figure.
Here the snow cover insulated the ice surface giving it a higher
temperature. In this measurement a temperature gradient is present down
to about 35 cm from where the temperature is almost constant. Since the
measurements in the lower sections of the ice were corrupted, nothing
should be interpreted from the almost constant temperature present
here, but the strong temperature gradient in the upper layer indicates,
opposite to the Freeman strait, that a cooling of the air has taken
place.
Sea
Ice salinity
The conductivity, and hence the salinity, of the melted samples was
measured with a Mettler Toledo conductivity sensor. The ice core
samples to be melted for salinity measurements were put in plastic
boxes, however there were not enough boxes for all samples so some had
to go into plastic bags. For the Barents sea some of these bags were
leaking, so the samples were lost.
In Freeman sundet the salinity curves shows a clear C--shape with
the highest salinities found in the upper and lower part of the core,
as can be seen in Figure 5.
The curves all look similar, except for the high salinity at the top of
core 2 and the low salinity at the bottom of core 4. The high
slainities found at the surface, especially in core 2 is likely to be
from sea water flooding the ice due to the general negative freeboard
along the sections.
The Barents sea ice cores do not have the C-shape of Freeman sundet.
Here the maximum salinity is found 15-25 cm from the upper surface of
the ice to decrease further down. In core 4 a small peak in salinity
can be found at about 60 cm from the ice surface. This could be a case
of rafting, since core 3 that was taken only a metre away had a
thickness half of core 4, and with a similar peak in salinity as is
seen in core 4 at 70 cm (Figure
6). What talks against this would be that the core seemed to be
solid between 40 and 60 cm.
Frazil
ice and thin ice
Grease ice and thin ice measurements were taken during 23. March when
KV Svalbard did a transect from the ice station in Freemansundet out
towards deeper waters in Storfjorden. 13 samples were taken during CTD
stations, some only thickness measurements. The sea ice cover in
Storfjorden consisted of small floes 2-3 m across in places, sometimes
much larger, intersped with a lot of grease and patches of rafted ice
thinner than 10 cm thick. An image of the thin ice cover on March 23
2007 is included below.
The mean grease ice thickness was 27.3 cm (11 samples), and the mean
thin ice thickness was 17.25 cm (4 samples). The mean salinity of the
thin solid ice was 16.2 psu, while that of the grease ice had
26.0 psu. All frazil ice samples were sieved using a plastic sieve.
Given the small number of samples and that the values are comparable to
those found in [4] no further prosessing of the samples are planned.
References;
[1] Wooding M. and Pearson T. ,ESA Spring 2007 Campaigns,
CryoVEx/IceSAR/BioSAR
-Operations plan, European Space agency, 2007
[2] Wadhams P. et al. ,Relationship between sea ice freeboard and draft
in the Arctic basin,
and implications for ice thickness monitoring, JGR-Oceans, 97, C12
20,325-20,334, 1992
[3] Notz D. ,Personal communication, Max Planck Institute for
Meteorology, Hamburg, Germany
[4] Smedsud and Skogseth; Field
measurements of Arctic grease ice properties and processes
Cold Regions Science and Technology (2006), Volume 44,
Issue 3, April,
171-183, doi:10.1016/j.coldregions.2005.11.002.