Snow Cover - Satellite-based Applications on Climate Change - page 210

factor controlling water budget and affecting pastoral farming in many mountain
regions. The high sensitivity of snow to changes in temperature and precipitation
makes it a primary indicator of climate change.
For over three decades, satellites have been actively used for large-scale moni-
toring of snow cover from regional to global scales. Owing to frequent scene
revisits, wide-area coverage and high spatial resolution satellite observations can
effectively supplement ground-based measurements and provide near-real-time
spatially detailed information on the snow cover distribution. Different techniques
both automated and interactive have been developed and applied to infer informa-
tion on snow from satellite imagery. The developed snow mapping techniques
utilize satellite measurements in the visible, near-IR, infrared (IR) and in the
microwave spectral bands.

14.2 Interactive Snow Mapping Technique and Product

Since 1966 NOAA has been generating snow cover charts for the Northern Hemi-
sphere. Maps are drawn interactively by analysts who rely primarily on the optical
imagery from geostationary and polar-orbiting satellites (Ramsay 1998 ). Until 1997
snow maps were produced on a coarse grid with about 180 km cell size on a weekly
basis. In 1998 the temporal and spatial resolution of the maps was increased to daily
and 24 km, respectively, and in 2004 the spatial resolution of the product was
further improved to 4 km (Helfrich et al. 2007 ) (see an example of the daily chart in
Fig. 14.1 ). NOAA's interactive snow cover charts are the longest satellite-derived
record of snow extent. This dataset has been used as the basis for many analyses of
snow cover variability and change on a hemispheric and continental basis (e.g.,
Robinson et al. 1993 ; Dery and Brown 2007 ). At NOAA, the National Weather
Service (NWS) interactive snow maps are used to initialize operational numerical
weather prediction (NWP) models (Ek et al. 2003 ).
From the climatological point of view, a drawback of the NOAA interactive
snow charts consists in the change of the temporal and, especially, of the spatial
resolution of the product which introduced inhomogeneity in the snow extent time
series. Although NOAA analysts are instructed to use a “50% coverage rule” when
deciding whether the grid cell is snow covered or snow free, they typically follow
an “aggressive” approach preferring commission errors in snow mapping to omis-
sion errors and therefore tend to overestimate the extent of a patchy and intermittent
snow cover. D. A. Robinson (2006, personal communication) has found that at
coarser spatial resolution, analysts map considerably more snow in the mountainous
areas in fall and spring. Due to the same reason, the snow extent in coarser
resolution maps may also be overestimated along the snow boundary.
Assessment of the Northern Hemisphere snow extent derived from the NOAA
interactive snow and ice charts has revealed an above average coverage in the
middle of 1970s and a substantial decline in the yearly average snow-covered area
during 1980s and beginning of 1990s. Later in the second half of 1990s, the snow