In this exercise, several techniques in generating and manipulating different types of surfaces are explored. DEM, Slope, and Curvature concepts are discussed using the Salzburg catchment data. Primary data used was a 10-meter DEM which all the other products are derived from.
General procedures and short discussions are outlined in the succeeding sections.
For this exercise, I chose the Tauruch River watershed located within the vicinity of Radstadt, Untertauern, up to the Obertauern area. The catchment data from UNIGIS Salzburg was filtered to show only the smaller sub-catchments within the Tauruch-Pongau river. A total of 12 smaller catchments are located within the area.
The Digital Elevation Model (DEM), the most important layer in this terrain analysis task was pulled out from the Organization's content. This is a 10-meter resolution DEM covering the Salzburg province. I clipped the DEM to the extent of the catchment area so that, all the other surfaces to be derived will have the same extent. Also, this will make processing faster as compared to when the whole DEM is used. Here, we can see the elevation surface covering the watershed. Lower areas are shown in light blue while those with high elevation are shown in white.
The slope surface was derived from this DEM layer as well. Using the Slope tool in ArcGIS Pro, the generated surface shows that the maximum slope value within the catchment is 81.8 degrees.
I noticed that the generated slope raster has some kind of gridded effect even if the Resampling type is set to bilinear or cubic interpolation. This might be due to the spatial resolution of the DEM which may be solved by resampling the DEM to a lower resolution.
Since we are interested to know the average slope for each sub-catchments, I used the Zonal Statistics tool to generate another surface showing the average slope for each zone.
The resulting layer shows that the average slope ranges from 20.5 to 28.5 degrees. To symbolize the zonal statistic surface, I used the geometric interval classification method with five classes. Geometrical interval classification works well on data that are not distributed normally such as in the case of the mean zonal slope surface.
Here is the histogram of the mean zonal slope raster for reference. Looking at the distribution curve, most of the values are concentrated on the right side of the curve meaning within the catchment area, most pixels have a value within the 24.9 to 26.7 range. We can verify from this histogram that the distribution is indeed not normal, thus, validating our choice of using geometric interval classification.
The resulting raster shows the mean slope for each of the four elevation zones. The values range from 15.45 to 59.21 degrees for the whole catchment. Comparing this to the first zonal statistic that we did for the sub-catchment, the values are relatively higher, mainly because of the extents or area of the zones. Since dividing the catchment area by elevation interval produces bigger areas, the range of slope values considered also becomes higher.
To explore the impact of DEM resolution on the generated slope surface, I resampled the original 10-m DEM to obtain a 100-m elevation model. I also generated a Slope surface using this DEM to compare it with the Slope surface from the 10-m DEM.
Since the resampled DEM is coarser and contains a bigger grid size, the generated slope surface also changed significantly. Because the computation of slope in GIS heavily relies on neighborhood concepts, it follows that the result will be also dependent on the scale. In this case, having a bigger grid size produces a smoother surface (less variation of slope values) as seen on the range of values. In general, reducing the resolution of the DEM also results in a reduced slope value (from max. value 59 deg using the 10-m DEM to 26 deg using the 100,- DEM).
While this is the case, the choice of DEM resolution will still depend on the scope of the application that we want to accomplish.
Can you find a way to establish a metric for curvature WITHOUT using the curvature function?
I think one alternative to establishing a metric for curvature besides using the curvature function is by generating an Elevation Profile.
Although the generated values apply only for a single line, multiple lines can be generated within a fixed interval, resulting in a number of elevation profiles. Arranging these profiles correspondingly can give us a graphical representation of the curvature of the terrain. Below is a sample of multiple elevation profiles with a relief map backdrop.
