246
A. Rango, A.E. Walker and B.E. Goodison
from 19,37, and 85 GHz channels for snow cover mapping; the 85GHz data were of
use in mapping very shallow snowcovers «5cm depth). They use a decision tree for
snow classification, including the separation of snow/no snow and the calculation of
depth. On a global scale, Grody and Basist (1996) use a decision tree to produce an
objective algorithm to monitor the global distribution of snowcover, which includes
steps to separate snow cover from precipitation, cold deserts and frozen ground. With
time, algorithms can be expected to become more sophisticated as they incorporate
filters to eliminate or minimize errors or biases.
Several investigators have attempted to use SAR data to map snow cover area and
to infer the snow water equivalent. In the snow cover aspects, Nagler and Rott (1997)
indicate that the European Remote Sensing (ERS) satellite SAR (C-band) cannot
distinguish dry snow and snow free areas. When the snow becomes wet, the backscattering coefficient is significantly reduced and the wet snow area can be detected.
In order to derive a snow map for mountain regions, both ascending and descending
orbit images must be compared with reference images acquired from the same positions. This technique was used to successfully derive snow cover for input to the
Snowmelt Runoff Model (SRM) for two drainage basins in the Austrian Alps (Nagler
and Rott, 1997). Haefner and Piesberger (1997) also used ERS SAR data to map wet
snow cover in the Swiss Alps. Shi and Dozier (1998) are developing multibandmultipolarization methods that have some promise for obtaining snow water equivalent with SAR data, however, the required satellite sensors are not likely to be available in the near future.
Ice and Glaciers. Microwave data are used to derive other cryospheric information.
Although the resolution of passive microwave satellite data prevents its use in deriving information on mountain glaciers and river ice, satellite SAR data has been shown
to be useful (Rott and Nagler, 1993; Leconte and Klassen, 1991; Rott and Matzler,
1987). Glacier snowline mapping is an important input in hydrological models and
in the computation of glacier mass balance. Adam et ai. (1997) used ERS-l C-band,
VV SAR data to map the glacier snowline within 50-75m of ground based measurements. The technique could separate wet, melting snow from glacier ice and bedrock,
but was not applicable when the snow was dry, since dry snow was transparent to Cband SAR. Some research has been done on freeze-thaw applications in permafrost
areas (England, 1990; Zuerndorfer et aI., 1989) but other sensors with higher resolution (e.g. Landsat) maybe more suited to mapping permafrost areas (Leverington and
Duguay, 1997; Duguay and Lewowicz, 1995).
The spatial and temporal patterns oflake ice freeze-up and break-up can be monitored using passive microwave measurements over large lakes because of the large
difference between the microwave emissivity of the lake ice and fresh water. Due to
the coarse resolution of satellite passive microwave radiometers, the technique is
limited to large lakes (> 1 00km 2 in size). The use of passive microwave remote
sensing for lake ice thickness determination is limited to longer wavelengths (e.g., 15GHz frequencies), which can be sensed through the entire ice thickness, as opposed
to shorter wavelengths (e.g., 37GHz) which are emitted from the near surface and are
sensitive to the overlying snow cover (Hall and Martinec, 1985). The potential
A. Rango, A.E. Walker and B.E. Goodison
from 19,37, and 85 GHz channels for snow cover mapping; the 85GHz data were of
use in mapping very shallow snowcovers «5cm depth). They use a decision tree for
snow classification, including the separation of snow/no snow and the calculation of
depth. On a global scale, Grody and Basist (1996) use a decision tree to produce an
objective algorithm to monitor the global distribution of snowcover, which includes
steps to separate snow cover from precipitation, cold deserts and frozen ground. With
time, algorithms can be expected to become more sophisticated as they incorporate
filters to eliminate or minimize errors or biases.
Several investigators have attempted to use SAR data to map snow cover area and
to infer the snow water equivalent. In the snow cover aspects, Nagler and Rott (1997)
indicate that the European Remote Sensing (ERS) satellite SAR (C-band) cannot
distinguish dry snow and snow free areas. When the snow becomes wet, the backscattering coefficient is significantly reduced and the wet snow area can be detected.
In order to derive a snow map for mountain regions, both ascending and descending
orbit images must be compared with reference images acquired from the same positions. This technique was used to successfully derive snow cover for input to the
Snowmelt Runoff Model (SRM) for two drainage basins in the Austrian Alps (Nagler
and Rott, 1997). Haefner and Piesberger (1997) also used ERS SAR data to map wet
snow cover in the Swiss Alps. Shi and Dozier (1998) are developing multibandmultipolarization methods that have some promise for obtaining snow water equivalent with SAR data, however, the required satellite sensors are not likely to be available in the near future.
Ice and Glaciers. Microwave data are used to derive other cryospheric information.
Although the resolution of passive microwave satellite data prevents its use in deriving information on mountain glaciers and river ice, satellite SAR data has been shown
to be useful (Rott and Nagler, 1993; Leconte and Klassen, 1991; Rott and Matzler,
1987). Glacier snowline mapping is an important input in hydrological models and
in the computation of glacier mass balance. Adam et ai. (1997) used ERS-l C-band,
VV SAR data to map the glacier snowline within 50-75m of ground based measurements. The technique could separate wet, melting snow from glacier ice and bedrock,
but was not applicable when the snow was dry, since dry snow was transparent to Cband SAR. Some research has been done on freeze-thaw applications in permafrost
areas (England, 1990; Zuerndorfer et aI., 1989) but other sensors with higher resolution (e.g. Landsat) maybe more suited to mapping permafrost areas (Leverington and
Duguay, 1997; Duguay and Lewowicz, 1995).
The spatial and temporal patterns oflake ice freeze-up and break-up can be monitored using passive microwave measurements over large lakes because of the large
difference between the microwave emissivity of the lake ice and fresh water. Due to
the coarse resolution of satellite passive microwave radiometers, the technique is
limited to large lakes (> 1 00km 2 in size). The use of passive microwave remote
sensing for lake ice thickness determination is limited to longer wavelengths (e.g., 15GHz frequencies), which can be sensed through the entire ice thickness, as opposed
to shorter wavelengths (e.g., 37GHz) which are emitted from the near surface and are
sensitive to the overlying snow cover (Hall and Martinec, 1985). The potential