And I've been doing some musing.
Here's an extreme detail to show you what I mean:
The orange regions are cliff, the light blue sold and dashed lines are creeks, the grey lines are 5m contours.
There are three renderings of the same track:
- Dark blue is a modern recent GPS track.
- Pink is an (improperly georeferenced) track from the ancient map - you will have to make the obvious translation yourself.
- Dashed black line is the state's topographic map's version.
The GPS and the historical trails are highly conformant to the contours - this is the trail on the ground, following ridges, moving from coll to coll along saddles, avoiding creek crossings. These tracks and the contours are empirically derived, from walking on the ground and from satellite observation respectively, whereas the state topo trail bears no relation to the topography.
What strikes me is how conformant the actual tracks are to the contours.
Which gets me thinking: crowd-sourcing maps is obviously a good thing, because the herd paths are a kind of least-energy path across the landscape, but also that it ought to be possible to generate analytically, for any two points, a least energy path across a given terrain. One would avoid cliffs, creek crossings, and unnecessary crossing of contours. One would presumably add some land-covering smarts (to avoid swamps and scrubby terrain.)
Here's Google Earth's most recent imagery
And here's some from 2006:
