| Summary: | The South African Water Act (Act 36 of 1998) (NWA, 1998) requires the calculation of the
amount of water required for ecological sustainability in aquatic systems. Part of this process is
the classification of slopes according to geomorphological class, which in turn requires, inter
alia, the derivation of river longitudinal profiles from which to calculate these slopes. This has
prompted the need to develop a method for obtaining these slopes that is fast and repeatable,
and can be applied at both a national as well as sub-catchment level. Input data sets are
required that are consistently available at a national as well as a sub-catchment level. This
study will assess the results of using a semi-automated GIS procedure to derive longitudinal
river profiles and slopes, based on nationally available data sets, in a test catchment.
In recent years the use of Digital Elevation Models (DEMs) is replacing contour lines on
topographic map sheets as the source of elevation inputs required to construct longitudinal
profiles. The main question put forward is: can river longitudinal profiles and slopes generated
from a DEM and based on 1:500 000 mapped river lines adjusted to within 50m of 1:50 000
mapped river lines, be used as effectively as river longitudinal profiles extracted from 1:50 000
mapped contours and based on 1:50 000 mapped rivers lines?
Primary catchment X, situated in eastern South Africa, is used as the test area for this study.
River channels in this catchment represent a range of slopes, from steep mountains streams to
flat lowland rivers. The assessment is undertaken on 109 rivers identified at 1:500 000 scale in
primary catchment X. These river lines are based on those originally scanned and vectorised
from 1:500 000 topographic map sheets. These lines are available at a national level, have
been connected to form a continuous network and horizontally adjusted to improve locational
accuracy to within 50m of the river lines on 1:50 000 topographic map sheets (DWAF, 2003;
DWAF, 2006). Profile elevation values extracted from three medium to low resolution Digital
Elevation Models are examined in this study.
This study compares slopes based on the elevation values extracted from DEMs according to
adjusted 1:500 000 river lines, to those extracted from contour lines on 1:50 000 topographic
map sheets according to 1:50 000 scanned river lines. These input data sets and any limitations
associated with them are discussed. A semi-automated method used to extract and compile the
elevation and distance values required to construct longitudinal profiles and the statistical tests
and procedures used to compare elevation and slope values, are also described. Comparisons are formed around two reference scenarios. In the first elevations are extracted at
the intersections of river lines with 1:50 000 scanned contour lines. The second reference
scenario uses these same derived longitudinal profiles, but divided into five sets of equal
horizontal intervals: 100m, 200m, 300m, 400m and 500m.
Finally, the conclusions that can be drawn form these results, together with any
recommendations for either improving or even replacing the data sets and methods described in
this study, are presented. It is found that, that when comparing slopes derived from 1:50 000
contour line elevations to those based on DEM elevations, steep slopes tend to be more
underestimated by the DEM than flatter slopes. More than 90% of profiles based on contour
intervals and more than 90% of slopes derived at 500m horizontal distance intervals show no
significant difference between slopes. It is finally suggested that the adjusted 1:500 000 river
lines available from DWA (DWAF, 2003; DWAF 2006) combined with elevations from medium
to low resolution DEMs can be used as a substitute for 1:50 000 river line and contour linebased
profiles. It is also suggested that the automated GIS procedure used to extract and
combine these values can be applied in other areas where the 1:500 000 river lines and
medium to low resolution DEMs are available.
|
|---|
