Sunday, November 13, 2005

Least Cost Path Analysis Through Military Terrain

Tuesday evening (yes, the day before the Tech Fair), I am meeting with a geoscience class to guide them through a workshop that I devised 2 years ago. The workshop is entitled 'Shortest Path Analysis Through Military Terrain'. However, I believe that ESRI changed their function from shortest path to least cost path, so I might have to update the name accordingly. I have not glanced or even thought about this exercise in at least a year, and so this blog entry is my attempt to jog my memory.

So, here is the scenario envisioned during the workshop. Texas is under seige, of course from an unidentified/unmentioned enemy. It is our job to take out/destroy certain key facilities in areas currently under the control of the enemy. So, we gather base data including DEM elevation data and certain TIGER files (roads, railways, urban areas, hydrology). Then we gather current intelligence including the locations of the key enemy facility in this quad, enemy mine fields, and the locations of enemy troops.

Then, of course, we plan the optimal path from a randomly generated starting point to the enemy facility. The shortest (least cost) path function is used with various rules defining the cost of traversing through any peice of terrain.

I actually have not experimented with the differences between running this exercise on ArcGIS 8.3 and 9.0, so I forsee where my free time will be spent tomorrow... The old versions (designed for ArcGIS 8.3) of this exercise can be found here: (Workshop without data provided | Workshop with data provided).

1 comment:

Anonymous said...

In fact, there was a US Army Corps of Engineers mapping system called TEM (Terrain Evaluation Module) that was written to support answering this kind of question. Back in the 1992-5 timeframe, MITRE and WES (Waterways Experiment Station) were working on the software. I was working on MATT (Mapping Analysis Tool for Transportation) (the original web pages seem to be gone). MATT was going to use TEM as the underlying map server but a schedule mismatch (i.e. we had to be done sooner) meant we wound up writing our own map server. The people at WES had a group that characterized the ability of military equipment to cross particular kinds of terrain (e.g. how steep a slope can vehicle type "A" climb if the slope is dirt and it's been raining for 3 days). They also dealt with questions like "how long will an asphalt surface last with convoys consisting of vehicle types "B" and "C" running over it?". I'm not sure how advanced the modeling got. At any rate, the GIS element to calculate the least cost path is going to be far easier to deal with than getting usable data that is not just an approximation.