alpha-beta-soup/train-landscapes

Viewshed analysis of a train route

train-landscapes

Viewshed analysis of a line feature draped over a DEM.

How to

First, obtain a raster digital elevation model. Edit the variable R_DEM to point to it. Viewshed will be determined on this surface.

Secondly, obtain a line vector feature (a shapefile's good). Edit the variable LINE_SHP to point to it. You may like to simplify the line or edit out the tunnels (where nothing is generally visible). The viewshed will be determined using a sample of points along this line as observer locations.

1. In a terminal: $ grass64 or $ grass70 as per what version of GRASS GIS you are using. (GRASS 7.0 will be faster for this task.)
2. Set up and/or connect to the GRASS workspace where your project data is available and will be stored.
3. Type $ sh generate_los.sh or $ sh generate_los_70.sh according to your GRASS GIS version, in the command line. Note that while GRASS 6.4 is the current stable release, the viewshed functionality in GRASS 7 is improved in performance and in cabability. Indeed, the maximum visible distance is adjusted in the GRASS 6.4 script to reflect the fact that the documentation of r.los suggests a maximum of 1000 rows and columns in the input raster used for the visibility analysis (given its resolution). The GRASS 7 script does not require this (r.viewshed).
4. Add an argument to this statement: either -train or -road depending on whether you want to run the intervisibility analysis for the road route, or the rail route.
5. Execute your command, and wait (probably a very long time).
6. Examine your result (enter exit to quit GRASS).
7. Create tiles/tweak your slippy map.
8. ????
9. Profit.

Note that for a very long and/or detailed line feature and/or a high-resolution DEM and/or a large DIST_PTS variable and/or a large MAX_VIS_DIST the procedure can take several hours to complete. You might like to test with non-default values first.

Requirements

generate_los.sh assumes GRASS 6.4. For GRASS 7.0 (currently not the stable release), use generate_los_70.sh or see original repo (by @pierreroudier).

I used QTiles, a QGIS plugin, to create tiles. There are other methods.

Inspired by

A long, beautiful train journey in New Zealand, the economics of choosing a train over driving, and this blog post on doing the same thing with R/GRASS.