rossbernet/ArcGIS cost distance

## ArcGIS cost distance
task: create cost distance raster for each of a series of shapefile for libya using a friction layer derived from open street map road network


steps:
1. obtain street network from OSM. I got mine from here: http://download.gisgraphy.com/openstreetmap/pbf/LY.tar.bz2
then used osmosis to convert pbf to xml.

In arc...
2. reclassify road network such that roads have a value of 1 and non-roads have a value of 5. I set pixel size equal to 200m x 200m

3. Use Cost distance tool.
Input: shapefile and road friction layer.
Parameters: I set a max distance of 200,000 because processing for the whole country was too time consuming, and may not have even been useful.

4. Then I reclassified each of the resulting cost distance rasters to values of 1 - 20, inversely (such that pixel values would be higher closer to the points/lines) And would add together in a normalized fashion (maybe?)


That is 95% on the raster creation. Then Rob put them on an online server and added them together and set slidable parameters. ( I don't know what exactly he did)

Optionally:
1.  I treated population centers differently to provide nuiance and take into consideration the different sized population.
As far as I can tell, there is no way to mimic the cost distance spread effect to reflect shapefile table properties.
If that isn't clear. I mean. I wanted smaller cities to look like they had a smaller impact.
To do this, I divided the population shapefile into small, medium, and large cities. (arbitrary cut offs...)
And then ran each with a different cost distance maximum. (100k, 150k, 200k)
	task: create cost distance raster for each of a series of shapefile for libya using a friction layer derived from open street map road network


	steps:
	1. obtain street network from OSM. I got mine from here: http://download.gisgraphy.com/openstreetmap/pbf/LY.tar.bz2
	then used osmosis to convert pbf to xml.

	In arc...
	2. reclassify road network such that roads have a value of 1 and non-roads have a value of 5. I set pixel size equal to 200m x 200m

	3. Use Cost distance tool.
	Input: shapefile and road friction layer.
	Parameters: I set a max distance of 200,000 because processing for the whole country was too time consuming, and may not have even been useful.

	4. Then I reclassified each of the resulting cost distance rasters to values of 1 - 20, inversely (such that pixel values would be higher closer to the points/lines) And would add together in a normalized fashion (maybe?)


	That is 95% on the raster creation. Then Rob put them on an online server and added them together and set slidable parameters. ( I don't know what exactly he did)

	Optionally:
	1. I treated population centers differently to provide nuiance and take into consideration the different sized population.
	As far as I can tell, there is no way to mimic the cost distance spread effect to reflect shapefile table properties.
	If that isn't clear. I mean. I wanted smaller cities to look like they had a smaller impact.
	To do this, I divided the population shapefile into small, medium, and large cities. (arbitrary cut offs...)
	And then ran each with a different cost distance maximum. (100k, 150k, 200k)