0

I am working on a project where I need to create viewshed areas around towers with a radius of 5000 meters. The underlying DEM I created using free Lidar data. Uptill now I did all this on a laptop with a core i7 proc and 8GBs of RAM. Pretty powerful but does not seem to be enough to complete the task within a reasonable timeframe.

Therefore I am considering the setup of a couple of virtual machines on AWS and I need input on what you guys out there would recommend in terms of resources: CPU, amount of RAM etc.

The process is the following:

  1. Merge DEM tiles into one huge TIFF that fully contains the DEM within 5000 m of the tower.
  2. Use this DEM (in TIFF) format in Grass 7.4 (r.viewshed) to generate the viewshed area.
  3. Vectorize the result viewshed raster and delete areas that have a DN value of 0 (not visible)

Even a smaller subset (a sector of 10°-20° from north) took half of the night to produce only the viewshed so I am kind of desperate to speed things up no matter what.

  • What resolution is your DEM? How many Towers? – BERA Jan 10 at 12:34
  • Say a dozen towers are located in my immediate area of interest. Pixel width and height are both 0.0000013 arc degrees Ideally if I could speed things up for one tower that would already be progress. – EZMapdesign Jan 10 at 13:01
  • 1
    0.15 meters to be more exact. – EZMapdesign Jan 10 at 13:10
  • I would use the DEM to model RF propagation and how buildings and vegetation obstruct signal. – EZMapdesign Jan 10 at 13:12
  • 2
    r.viewshed with --verbose flag gives you an estimate of required memory. Also I would recommend for each r.viewshed computation limit first the computational region based on the position of the tower and max distance. – Anna Jan 17 at 3:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.