I am trying to understand why the viewshed analysis is taking so long (nearly 124 hours) to run. I have a DEM raster file for all of Colorado which has almost 3400 features and a polyline national forest roads file that has nearly 9000 features with various segment lengths such as 0.002 to 41.5 miles. After the last 124 hour running it completed but no raster data can be seen. The values vary from 0 to 255 and I am unable to view the attribute table. What am I doing wrong?

  • 1
    Did you try cropping a small area and do the same process? – aldo_tapia Aug 28 '17 at 18:41
  • 1
    What GIS software are you using? – PolyGeo Aug 28 '17 at 19:06
  • aldo_tapia - no I havent. I tried selecting one road segment and then reran the viewshed which seemed again very slow – David Aug 28 '17 at 19:20
  • PolyGeo - ArcGIS – David Aug 28 '17 at 19:22
  • 1
    what do you mean by "a raster with 3400 features" ? What do you mean by "no data can be seen" if you have a minimum and a maximum value ? Please focus your question on errors and not on the fact that it is slow (you have a huge number of observer points, so it is normal to be slow) – radouxju Aug 30 '17 at 11:38

Viewshed is slow because it is complex. It will first convert your polylines to raster, so a very long line will converted into a large number of pixels (so one 41.5 miles of road ~= 750 pixels on a 90 DEM). Then each "observer" pixel is taken into account to check if it is visible from any point of your DEM. This is simply huge.

If you want to gain som speed, you could use a outer_radius value that will reduce the searching distance. If you have several computers, you could also do the work in parallel and combine all results in a sum of pixel values. But there is nothing magic to run a complex task.

Concerning your raster, it seems (to be verified) that the sum was limited to 8-bit, and you are asking for far more than 8 bit. I think that viewshed2 has a limit of 32 bits, so this should be enough for you.

| improve this answer | |

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.