International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3
forest factor F and time- and space-varying c in (2) is
to correct the systematic errors in (1) so that only
random errors remain.
Figure 4: Difference maps between the new and old SSM/I SWE algorithms for October 1990 through May 1991.
3.3 Comparisons between the new and old
algorithm for SWE retrieval
Figure 4 shows maps of monthly biases, i.e.,
difference between SWE estimate from the new (2)
and old algorithm (1). Shades of blue indicate that the
new algorithm is estimating less snow than the
original global snow algorithm, while shades of red
indicate that the new algorithm is estimating more.
The pale blue and sage colors in Figure 6 indicate
areas where there is little difference in SWE estimates
by the two algorithms. Because snow crystals are
typically larger in tundra regions than in forested
regions, snow thickness and SWE were overestimated
using the original algorithm. Snow was under-
estimated using (1) in forested areas because the
microwave emission from the trees was overwhelming
scattering from the underlying snowpack. The most
noticeable negative values are found in the northern
portion of the Mackenzie River basin and on the north
slope of Alaska. With the inclusion of a forest factor in
. Istanbul 2004
Inter
—
th
th
PN pec} N
—