Among the sample points, the lowest elevation of −154 m was
located at Aiding Lake in the Turpan Basin, and the highest
elevation of 8,186 m was located in the Himalayan Mountains.
The frequency distribution of the
SRTM
slope sample points
is shown in Figure 3b. The peak near to 0° indicates a large
proportion of low-slope sample points. The number of sample
points decreased with increasing slope and specifically de-
creased rapidly within the range of 0 to 5° and more slowly
above 5°. The number of points with a slope greater than 45°
approached 0. Figure 5 shows the spatial distribution of these
three slope levels. The proportion of sample points with slope
less than 5° was 52.4 percent, and these points were mainly
located in northeast China, northwest China and part of the
Qinghai-Tibet Plateau. Sample points with slope greater than
45° only accounted for 0.3 percent and were primarily distrib-
uted in the Tianshan Mountains, the western Kunlun Moun-
tains, and the boundary between the second terrain terrace
and the southeastern Qinghai-Tibet Plateau. Sample points
with slope between 5° and 45° were widely distributed in the
hilly and mountainous regions nationwide. The mean slope
of the entire sample was 9.06°.
Considering that the 3-second
SRTM
data were aggregated
from the 1-second data, the derived slope data were attenuated
as the spatial resolution reduced, which meant an overall trend
to gentler topography and smaller average slope, it was therefore
necessary to account for the influence of the attenuation (Yang
et al.
, 2008). In this study, the average slope of
Hc-DEM
and
SRTM
data were 9.85° and 9.02°, respectively, which indicated that the
slope of
SRTM
data reduced on average by 0.83° compared with
the
Hc-DEM
. This reduction produced little effect on the follow-
ing analysis of the
SRTM
elevation error over the entire sample
but should be corrected in practical applications to local areas.
The distribution of the
SRTM
aspect sample points is shown
in Figure 6a. The aspect showed a significant pattern of
concentration in the four main directions (north, east, south,
and west) and four secondary directions (northeast, southeast,
southwest, and northwest). Furthermore, the proportion of
points in the secondary directions was greater than that in the
main directions. Among the main directions, the proportion
of points in the north and south direction was greater than
that in the east and west direction. This pattern was signifi-
cantly different from the reference
Hc-DEM
(Figure 6b), which
demonstrated that the
SRTM
data had a significant problem in
the representation of aspect.
Previous studies have noted that the tracking direction
of the remote sensing platforms has a great impact on
SRTM
(a)
(b)
Figure 3. Frequency distribution of (a) Elevation, and (b) Slope of SRTM data.
Figure 4. The three terrain zones in China.
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
February 2016
139
