The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
ILWIS 3.4 and profiles plotted. Figure 5 shows an example of
the terrain profile along one of the transects.
Transect A-B: DEM Elevation X Cumulative Distance
Figure 5. Terrain profile along Transect A-B
3.3.5 Assessment of the cartographic quality of the SRTM
DEM: One major requirement for deploying contours
interpolated from SRTM elevation data in 1/25,000 mapping is
that the contours must be of good cartographic quality.
However, SRTM DEMs have been shown to suffer from a
number of gross, systematic and random errors propagated from
the Synthethic Aperture Radar (SAR) imaging system. As
demonstrated by Koch, A. and Lohmann, P., (2000), SAR
imaging system is affected by errors due to baseline tilt angle,
baseline length, platform position, phase and slant range. These
errors are known to affect the accuracy and the quality of
SRTM DEM and its derivatives such as slope, aspect and
contours. In contours interpolated from the SRTM DEMs for
example, such errors may manifest as artefacts such as short
pieces of unclosed contour lines, self-intersecting contour lines
and contour lines intersecting other contour lines with different
contour values. The issue of correcting errors in digital
elevation models has received the attention of many researchers.
For example Ping Wang (1998) employed 2-D Kalman filtering
approach to generate optimal estimates of terrain variables from
a noisy DEM. The approach comprised of using a 2-D Kalman
processor, a function for outlier detection and a two-step
filtering procedure. More recently, Emanuel Mahler (2001)
implemented the wavelet transform model for scale-dependent
filtering of high resolution digital terrain models. In our study,
we investigated the cartographic quality of contours
interpolated directly from SRTM DEM. The objective was to
determine the suitability of using such a product in 1/25,000
topographical mapping and to propose, employ and evaluate a
methodological approach in order to determine its effectiveness
in improving the cartographic quality of contours derived from
SRTM DEMs. The steps involved in this accuracy assessment
method were:
(1) interpolating contours directly from the SRTM 90m
DEM and visually inspecting the quality of the
contours
(2) re-sampling the SRTM 90m DEM to twice the
resolution (45m) and visually analyzing contours
derived from the new DEM;
(3) subjecting the SRTM 90m DEM to further
processing by re-interpolating the surface from its
elevation data and visually analyzing the quality of
contours derived from the new product;
(4) visually comparing the results of the above
procedures.
The goal of the first step above was to visually assess the
cartographic quality of contours interpolated from the CGIAR-
1351
CSI 90-m resolution SRTM DEM without subjecting it to
further processing. To achieve this goal, the contour
interpolation function available in ArcGIS 9.2 GIS was used to
create a vector contour map with a vertical interval of 5m (a
vertical interval value recommended for a 1/25,000 topographic
mapping). A small window extracted from the resulting contour
map is presented in Figure 6(a). From this map extract, it can
be seen that direct interpolation of contours from the 90-m
resolution SRTM DEM produces artefacts as highlighted in the
encircled areas of the map.
The second step in the assessment of the cartographic quality of
the SRTM data was executed to visualize the impact of further
processing of the data set on its cartographic quality. This
procedure involved simply applying a pixel densification
operation by a process of raster re-sampling. In this study, we
executed this process by performing a cubic interpolation of the
90-m resolution SRTM DEM to re-sample the data from 90m to
45m. The result of this operation is presented in Figure 6(b) as a
map extract using the same window coordinates as in the first
step.
The third step in the assessment of the cartographic quality of
the SRTM data involved re-interpolating the SRTM DEM
surface from a point raster map derived from the SRTM DEM
using the Inverse Distance Weighted (IDW) point interpolation
function. The principle of this interpolation technique, also
called the moving surface function, is well documented in the
literature (see, for instance, ILWIS, 2001). This technique
involves determining a new output value for a given cell by
fitting a polynomial surface through all points weighted by their
individual weight factors considering only points that fall
within a certain limiting distance (search radius) towards this
pixel. The weight functions ensure that points close to an
output pixel obtain a larger weight value than points which are
farther away from an output pixel. In this study, we executed
this interpolation operation by adopting a methodological
approach as follows. First a point map was created by running a
small Visual Basic 6.0 program developed in-house. The
program created a point map in which the points corresponded
to the centres of the cells in the SRTM 90m grid-based map.
The point map (with SRTM DEM heights as attributes) was
then imported into the ILWIS 3.4 environment where its
variogram surface was calculated. The calculated variogram
surface enabled an optimal search distance for the moving
surface interpolation to be established. For the point data set
covering our study site, this value was found to be 2.5 times the
grid cell size. The actual interpolation was executed in ArcGIS
9.2 environment using the IDW interpolation function. This
operation was executed twice (for 90-m and 45-m resolutions)
each time setting the search distance of 2.5 times the cell size
and a weight exponent of 2. The resulting raster grid map was
then passed to a contour interpolation function to create a 5-m
vertical interval contour map. The resulting contour maps are
presented in Figures 6 (c) and (d).
To further visualize the spatial relationships between the three
resulting contour maps above, we extracted a small portion of
these maps and overlaid them in the ArcGIS 9.2 GIS
environment as shown in Figure 6(e). To investigate the degree
of fitness of the hydrography with the derived contour maps, we
super-imposed the hydrographic network as shown in Figure
6(f).