The following tutorial demonstrates how to create a raster digital elevation model (DEM) by interpolating elevation values from millions of points and 3-D structure lines commonly referred to as breaklines. Creating a DEM in Geomatica with this type of data can easily be done using the following algorithms in EASI: VDEMINGEST, VDEMSETUP and VDEMINT.
A dataset of the Hamilton, Ontario area supplied by JD Barns was used in this tutorial. The dataset consists of 550,000 points with a nominal spacing of 15m. The 98,000 breaklines consisted of 2,100,000 vertices.
Small subset of reference 3D points and breaklines (Below)
VDEMINGEST ingests multiple vector files containing points, breaklines, contours, valleys, ridges, and cliffs, and merges them into vector segments in a PCIDSK (.pix) format file. The vector segments are formatted specifically for VDEMINT which uses them to create a raster digital elevation model (DEM).
VDEMSETUP creates an index file containing a list of raster DEM files that will be generated by VDEMINT.
VDEMINT generates a raster digital elevation model (DEM) by interpolating elevation values of points, contours and 3-D structure lines in vector segments. In addition, 2-D break lines, such as valleys, ridges, and cliffs with no elevation values are also accepted as constraints for interpolation.
1. From the Geomatica toolbar, select the EASI application. The EASI environment window will open.
2. Type s vdemingest to view the required parameters in the algorithm.
NOTE: Parameters can be copied into a Notepad file, modified based on your data and saved as a *.eas file. This file can then be copied into the C:\PCIGeomatics\Geomatica_2014\pro directory and called in the EASI environment.
3. Modify the vdemingest parameters to suite your data. In the example above, \points\*.* and \breaklines\*.* was used. This tells the software to use all of the files in the \points and \breaklines directories.
4. When you are done modifying the Notepad file, save it into the \pro directory as vdemingest_test.eas.
5. In EASI type r vdemingest_test . Hit <enter>
6. Once the script has successfully completed you can move on to VDEMSETUP. Make sure the output vector.pix file from VDEMINGEST has been created and is found in the location you specified. This file will be used as an input in VDEMSETUP.
7. In EASI type s vdemsetup.
8. Copy the parameters into a Notepad file. Modify them to suite your data and save the file as vdemsetup_test.eas in the \pro directory.
9. In EASI type r vdemsetup_test. Hit <enter>
10. The process will run fairly quickly and produces an index.txt file. This file will be used as input into VDEMINT.
11. In EASI type s vdemint.
12. Copy the parameters into a Notepad file. Modify them to suite your data and save the file as vdemint_test.eas in the \pro directory.
13. In EASI type r vdemint_test. Hit <enter>
14. Once the process has successfully completed, you will notice index.pix is created. This file contains the generated DEM file created from all of the point and breakline information.
15. Open index.pix in Focus.
16. In the Files tab right click on the channel and select View > As Pseudocolor
Tests can to be done on each individual dataset to determine if generating a vector DEM from points and 3D vector segments is more accurate than extracting a DEM from stereo imagery in OrthoEngine.
For testing purposes, the resulting interpolated DTM (vector DTM) and a DTM generated from stereo imagery in OrthoEngine were put into a single file with three 32R bands. The two DTMs were subtracted resulting in a third difference band; the amount of displacement indicates the amount of error in the DTM.
The images below show a view of both DTMs using the Live DEM editor in Focus. As you can see the vector DTM has a smoother overall appearance compared to the OrthoEngine DTM.
Detail of interpolated Vector DTM
Detail of corresponding OrthoEngine DTM
The difference image (vector DTM – OrthoEngine DTM) in this area, shown to the left, has some of the worst differences in the entire DTM area.
In the blue area labelled A the vector DTM is up to 16m below the OrthoEngine DTM. In the red area labeled B the vector DTM is up to 15m higher.
For reference, the uniform green area on the left has a difference varying between 1cm and 50cm, and an average a difference of about 10cm.
Looking at a number of areas in this dataset, it appears that about a quarter of the time the OrthoEngine DTM was more accurate than the vector DTM. The OrthoEngine DTM tended to be more accurate in flatter areas than the vector derived DTM. About a quarter of the time the vector DTM appeared to be more accurate. However, about half the time results were inconclusive due to dark shadows in the imagery or dense tree cover making it impossible to see the true ground in the stereo imagery used to derive the OrthoEngine DTM.
- Differences/errors are almost always in deeply shadowed or heavily treed areas. Both types of DTM’s make similar errors. However these types of errors are typically unnoticeable in the final ortho mosaics.
- 3D stereo vector DTMs can model very complex terrain around man made features better. For example bridges, or steep edges of roadways and retaining walls.
- 3D stereo vector DTMs look cleaner and smoother and are closer to what many users and cartographers traditionally expect, even if they may be less accurate than OrthoEngine’s dense DTMs in many areas.
A key aspect to the OrthoEngine method of generating DTMs is labour efficiency. Using imagery to generate a DTM in OrthoEngine stereo DEM extraction is very efficient compared to manually collecting 3D stereo breaklines and points which can be a very labour intensive process. The purpose of this tutorial is to show that if 3D stereo breaklines and points are available, a DTM can easily be created in EASI. The generated DTM will result in a smooth and accurate output.
Live DEM editing in Geomatica can be used to edit common terrain issues (Layer -> DEM editing in Focus or OrthoEngine -> DEM from Stereo processing step-> Manually edit generated DEM).