Geology Reference
In-Depth Information
ocean and therefore the precipitation. Therefore, snow-
fall tends to be low except near the ice edge. Antarctic,
on the other hand, is a continent surrounded by ocean
and, therefore, the moisture is readily available. The ice
cover that surrounds the land is exposed to precipita-
tion from the open ocean and therefore tends to be
covered by thicker snow. The weight of the snow may
push the ice below the sea level, resulting in the snow
being flooded by salty water.
The most commonly used approach for estimating
snow depth from passive microwave data are based on
correlating the observations (or derived parameters) with
in situ measurements of depth. Point measurements of
snow depth along transect lines are commonly conducted
in field experiments. However, the measurements may not
reflect the true average of the snow depth within the large
footprint of a satellite observation.
This approach was used by Markus and Cavalieri
[1998] where the gradient ratio from the SSM/I bright-
ness temperature from 19 and 37 GHz channels was
regressed against in situ measurements of ice thickness
from different expeditions in the Weddell, Bellingshausen,
and Amundsen Seas in the Antarctic from 1988 through
1994. The rationale behind the selection of these two
channels is their different scattering efficiencies from
snow particles. The algorithm utilizes the assumption
that the microwave emission increases in the presence of
snow and the increase is greater around 37 GHz than
19 GHz channels. It is applicable to dry snow conditions
only. The following equation that relates the snow depth
h (in centimeters) to the gradient ratio GR 37 V 19 V is used
[GR 37 V 19 V is calculated from equation (10.18)]:
where k 1 and k 2 are defined in terms of the brightness
temperature of the OW, ( T b ,OW ):
kT VT V
b
37
19
(10.105)
1
,
OW
b
,
OW
kT VT V
b
37
19
(10.106)
2
,
OW
b
,
OW
The correlation coefficient between h and GR 37 V 19 V is
found to be −0.77 from various ship cruises and 0.81 from
regional snow depth distributions [ Markus and Cavalieri ,
1998].
Markus and Cavalieri [1998] found that this algorithm,
on average, underestimated snow depths by 3.5 cm due
to primarily the areal integration of the SSM/I observa-
tion. The authors measured snow depth along with den-
sity and grain size from five sites in the Beaufort Sea,
Chukchi Sea, and Elson Lagoon. They found that the
snow in the Beaufort Sea site had larger grain size and less
density. However, this does not affect the correlation
between PSR snow depth and the in situ measurements.
However, the authors suggested that different sets of coef-
ficients were needed for snow over rough ice or MY ice.
The snow depth model presented by equation (10.103)
applies to dry snow only because under wet snow condi-
tions the emissivity from both 19 and 37 GHz become
equal (approaching 1), and therefore the gradient ratio
approaches zero regardless of the snow depth. Furthermore,
the model cannot be applied to metamorphosed snow
under melt/refreezing conditions because of the large grain
size in the snow cover, which reduces GR 37 V 19 V signifi-
cantly. The model does not give accurate results in the
case of snow‐free or a very thin snow layer on the ice
surface because the gradient ratio is almost zero from
both channels. Comiso et  al . [2003] developed a snow
depth retrieval algorithm from AMSR‐E based on the
model described above. Currently, this is the only opera-
tional snow thickness product from passive microwave
observations. It has been assessed to yield dry snow
thicknesses up to 50 cm (due to the limited penetration
depth) with about 5 cm accuracy for snow on smooth FY
ice [ Comiso et  al ., 2003]. The use of a lower frequency
channel (e.g., 6 GHz) to measure thick snow depth or a
high‐frequency channel (e.g., 89 GHz) to measure thin
snow has not been attempted.
The above‐described model is rather simple and does
not account for key factors that affect the snow structure
and therefore the emitted microwave radiation. Obviously,
this radiation cannot be a unique function of snow depth.
These factors can be grouped into two sets: the first
involves physical and geometrical characteristics of the
snowpack itself and the second involves properties of
the  underlying ice surface. Each set is addressed briefly
in the following.
hab
GR 37 19
(10.103)
VV
where a and b are coefficients derived from the linear
regression. From an experiment conducted in March 2003
over the Alaskan Arctic using the airborne Polarimetric
Scanning Radiometer (PSR) onboard NASA's P‐3 air-
plane, Markus et al . [2006] found a = 2.9 and b = − 782.4.
In this experiment, seven flights coincident with AMSR‐E
overpasses were conducted. Snow depth and ice thickness
were measured at about 5 m intervals along a 4.5 km
surface transect.
Due to the large footprint from the 19 and 37 GHz
channels (Table 7.4), the possibility of coexistence of ice
and water increases. Therefore, instead of using equation
(10.18), the gradient ratio is corrected for the ice concen-
tration C using to the following equation:
TVTVkC
TVTVkC
37
19
1
b
b
1
GR 37 19
(10.104)
VV
37
19
1
b
b
2
Search WWH ::




Custom Search