Snow Cover - Satellite-based Applications on Climate Change - page 214

depth retrievals at 50-70% for mostly forested locations. Much of uncertainty
associated with SWE and snow depth retrievals from satellite observations in the
microwave results from a strong dependence of the surface-emitted microwave
radiation on other physical properties of the snow pack: its depth, the snow grain
size, density, and stratification (Rosenfeld and Grody 2000 ). Uncertainty in the
forest cover properties and its effect on the upwelling radiation complicate micro-
wave snow depth and SWE retrievals even further. Currently the most accurate
estimates are obtained from region-specific empirical algorithms, where limiting
the geographic domain helps to reduce the variation of sensitive factors (e.g.,
Derksen et al. 2003 ).
As compared to passive microwave instruments, synthetic aperture radars (SAR)
provide much higher, up to several meters spatial resolution imagery. However,
current active microwave sensors operate at frequencies too low to derive informa-
tion on the snow depth or SWE. They can be effectively used only to distinguish
between dry and melting snow and thus can be applied to monitor seasonal snow
freeze/thaw processes (e.g., Koskinen et al. 1997 ).
14.4 Automated Snow Remote Sensing in Optical
Spectral Bands
As compared to satellite passive microwave measurements, observations in the
optical spectral range allow for more accurate mapping of snow cover at higher
spatial resolution. The reflectance of snow is high in the visible spectral band but
drops to very low values in the shortwave and in the middle infrared. This specific
spectral feature distinguishes snow from most other natural land surface cover types
and clouds and therefore is actively used in automated algorithms to identify snow
in satellite imagery.
Since 2000 NASA has produced snow cover maps from observations of the
Moderate Resolution Imaging Spectroradiometer (MODIS) onboard Terra and
Aqua satellites. A suite of MODIS snow products includes global maps of snow
cover distribution generated at daily, 16-day, and monthly time steps at a spatial
resolution ranging from 500 m to 20 km (Hall et al. 2002 ). Several algorithms have
been developed and applied to identify and map snow cover from Advanced Very
High Resolution Radiometer (AVHRR) sensor onboard NOAA polar-orbiting
satellites (e.g., Simpson et al. 1998 ; Baum and Trepte 1999 ). In 2006, an automated
algorithm to identify snow cover in AVHRR imagery was implemented at NOAA
NESDIS. This technique is used to produce daily, global snow cover maps of snow
cover at 4 km spatial resolution ( http://www.star.nesdis.noaa.gov/smcd/emb/snow/
HTML/snow.htm ). Maps of snow cover distribution are also produced from a
number of other instruments onboard polar-orbiting satellites including, in particu-
lar, VEGETATION from the Syst`me Pour l'Observation de la Terre, literally
translated as the “system for earth observation” (SPOT) and the Landsat Thematic
Mapper (e.g., Xiao et al. 2004 ; Dozier and Painter 2004 ).