In densely forested or vegetated areas, photogrammetry is notoriously problematic for topographical mapping. One of the solutions to this issue is manual interpolation, but the process can be extremely slow and cumbersome. In certain cases, aggregated and optimized digital surface models (DSM) can be used for the extraction of general contour outlines of major land shapes. This procedure is typically more accurate and always much more efficient than manual interpolation. This tutorial covers the approach using ArcMap, but QGIS or other GIS applications are also suitable.
The aggregation method works best on densely forested landscapes with very pronounced land shapes and fairly even canopy profile. For example, on a recent project in coastal Ecuador all other survey options were first evaluated and assessed to be either above budget allowance or were simply physically impossible to conduct. The land was then surveyed through photogrammetry with a budget UAV (DJI Mavic) and the imagery was processed on-site (see Figure 1-1).
Since the site is almost entirely densely forested, the resulting DSM was extremely noisy to the point of being unusable. Cleaning up the noise manually would have taken too long and would most likely not have arrived at a satisfactory result anyway. So we decided to aggregate the DSM.
Aggregating simply means reducing the resolution of the raster, essentially pixelating it in a controlled manner to a degree that will smudge the vegetation detail but preserve the general land shape. When contours are extracted they will provide a general outline of what the terrain is like (see Figure 1-2), instead of outlining all the surface features and vegetation.
The process of aggregation in ArcMap is fairly straightforward, but you will have to experiment with various levels of aggregation to determine which one works best for the given situation.
- Drag and drop your DSM file into ArcMap
- Bring up the Search by going to Window → Search (Ctrl+F).
- In the search field type in Aggregate and press Enter.
- From the list of options, select Aggregate (Spatial Analyst).
- Under Input Raster, specify the original DSM layer.
- Under Cell Factor enter a value for the degree of aggregation that you need. The value varies from DEM to DEM, so you have to experiment. In my case a value of 180 works best. I arrived at that number after four attempts of trial and error. See Figure 1-4.
- Under Aggregation Technique (Optional) choose MINIMUM. See Figure 1-3.
- Press OK.
Before proceeding with any further processing or extraction, it is a good idea to fill in any 'holes' within the aggregated DSM that would have a negative effect on outputs.
- Bring up the Search (Ctrl + F) and type in Fill.
- From the list of options, choose Fill (Spatial Analyst).
- Under the Input Surface Raster, specify the aggregated DSM layer.
- Press OK.
Now that the optimization of the DSM is complete, you can proceed with contour extraction following the same process outlined in this tutorial, using the aggregated DSM generated through the process above as your base.
Bear in mind that you won't be able to produce meaningful raster outputs from the aggregated DEM because they will appear so pixelated as to be unusable. You can however produce good quality vector based maps (see Figure 1-5 below. Note: Google Earth's terrain is a poor reflection of the land shape on this site) from the extracted vector contours, such as elevation, slope, catchment, drainage and other maps.