Geology Reference
In-Depth Information
meteorological conditions that change the composition
of the emitting layer. The key parameters that determine
the emissivity of this layer are those that control the
energy loss through absorption and scattering. This
includes ice salinity and snow wetness (which determines
the absorption) as well as snow metamorphism (which
determines the scattering loss). Effects of different forms
and measures of snow metamorphism on microwave
emission are discussed in section 7.7.3.2. Surface glaze
and layering decrease the horizontally polarized micro-
wave emission at low frequencies, hence increasing the
polarization difference [
Mätzler
, 1985]. This causes
underestimation of the ice concentration [
Comiso et al
.,
1997]. Under steep vertical temperature gradient within
the snow pack during polar winter (>25 °C per meter) the
sublimated vapor from the snow base condensates within
a short distance, forming a depth hoar layer with rela-
tively large grains of 1-10 mm in diameter. This process
is enhanced in the presence of dense ice layers within the
snow because the layers impede the upward flux of the
vapor. Snow grains scatter the energy and therefore
reduce the emissivity. If the algorithm incorporates the
brightness temperature, the resulting ice concentration
will be underestimated. The presence of the snow hoar
layer at any depth, and particularly near the surface,
tends to affect the gradient ratio and consequently the
algorithm that use this ratio (e.g., NT2).
Changes in atmospheric temperature are responsible
for significant changes in ice properties and more impor-
tantly snow forms and properties. However, the relations
are not easy to grasp. Even if the air temperature is used
as an input to an ice concentration algorithm, its exact
impact on the snow and therefore on the emitted radiation
and the calculated concentration will not be known. A
reasonable summary of snow on sea ice and its processes
in relation to weather events is presented in
Sturm and
Masson
[2010]. Most of the anomalies observed in the
results from ice concentration retrieval algorithms are
caused but processes in the snow. As soon as the snow
starts to fall on the ice surface, the microwave emission
starts to fluctuate and will continue until the snow settles.
This results in parallel fluctuations of the estimated ice
concentration (Figure 10.24). Snow accumulation on new
ice types will cause brine to be wicked up or the ice sur-
face to be depressed below the sea level. In both cases the
salinity of the snow will increase and its emissivity
decrease. This might lead to underestimation of ice
concentration.
Cycles of temperature rise near or above freezing
followed by a drop to deep freezing produce ice lenses
and layering within the snow rain-on-ice, followed by
freezing, lead to the same effect.
Sturm and Masson
[2010]
mentioned that ice layering comprises only 3% of the
snow cover on east Antarctic sea ice in winter. However,
these layers intercept rain water and snow melt from per-
colation further downward, causing the layers to thicken
(up to 30 mm) and spread out laterally.
Garrity
[1992]
studied the effect of these layers on microwave emission
and the calculated ice concentration and concluded
that they have a significant depolarization effect. In their
classification of snow on the ground,
Colbeck et al
. [1990]
identified saturated snow to have 15% of liquid water.
Any level above that turns the snow into what is defined
as slush. Ice concentration retrieval algorithms underesti-
mate the concentration in this case. Another meteorologi-
cal effect on the snow cover, caused by the wind, is known
as wind slab. Strong wind above 10 m/s produces highly
compacted snow layer (slab), which has fine grains (0.1-
0.5 mm). This hard snow (which can only be cut with a
saw) is used by Inuit to construct igloos. The effect of this
snow on the microwave emitted radiation is not known.
The above discussions signify the point that weather
effects on the snow is the major source of error in esti-
mating ice concentration from remote sensing data.
Warm temperature increases water contents in the snow,
which causes decrease in emissivity and brightness tem-
perature (see the example shown in Figure 7.46). This
decrease does not affect the total ice concentration but
causes the FY ice to be misclassified into MY ice.
A practical example that shows how meteorological
effects can change microwave emission from thin ice is
presented in Figure 10.18. The figure shows the distribu-
tion of data sampled every 30 min during the period
22-31 December 2001 from simulated sea ice grown in
outdoor tanks located in the National Research Council
in Ottawa, Canada. Ice grew from 3 to 90 mm during this
period. Ideally, all data should group into one point (the
ideal tie point) or at least form a narrow cluster with well‐
defined center. However, the data show wide scattering,
which defies any attempt to establish a single tie point
representative of this thin ice type. The spread of the data
points is not correlated with either ice thickness or sur-
face temperature (not shown in the figure) but was prob-
ably affected by snow/ice surface composition, which is
triggered by weather conditions. The polarization ratio
at 19 GHz is more sensitive to the weather factors than
the 85 GHz channel. The period during which the data in
the figure were acquired features one event of snowfall
(4 cm accumulation), slush and frozen slush at the ice sur-
face, and air temperature between −1.5 and −11.5 °C. It is
conceivable that the output from an ice concentration
algorithm that uses a single tie point for the data shown
in the figure will produce wrong results. Some algorithms
adjust the tie points based on daily average brightness
temperature data. This, however, would be more success-
ful in the case of the more stable surface of thick ice.
Emission models can be used to compute and analyze
the sensitivity of the retrieved ice concentration to the
