Jump to content

# Cost Analysis Polygones (Creating Frontiers)

4 replies to this topic

### #1 dav Posted 21 April 2010 - 06:24 PM

dav

Contributor

• Validated Member
• 34 posts
• Gender:Male
• Germany

Dear Community,
I want to create frontiers (Polygone-Layers) between points which are representing an equal DEM-derived cost value. Meaning I want to create some "cost-derived thiessen polygones". Unfortunately I'm not an expert at scripting or programming (I'm starting using R). Do you know any program or packages, which have such functions? Does anybody of you have some experience in working with cost-functions and has any suggestions where I can find some good sources for starting? I'm just starting with the cost-issue, but there are so much possibilities.
I've got a lot of other ideas dealing with cost-analysis. For example telling the function that it should create a polygone or polyline around a point, at a special value (based on a DEM) (e.g. the power of a traveller for one day).

I really want to do such cost analysis and if there are no ready programs, I have to code them myself and I will do if necessary. So if you got any suggestions, I will be very very grateful.

Some good theory is described here, but not the methods I want to use: http://dissertations....leusen/ch6.pdf

Thank you, dav

### #2 dsl Posted 22 April 2010 - 12:17 AM

dsl

Master Contributor

• Validated Member
• 225 posts
• Gender:Male
• Location:Denver
• United States

Some commercial packages that have raster cost functions are ArcGIS with the Spatial Analyst extension, and IDRISI. Open source options might be GRASS, SAGA GIS, and something like SEXTANTE for gvSIG. You can check out www.freegis.org for other sources.

Cost, as it implies, uses attributes to make locations more prohibitive to traverse. For example, just using distance means that the further something is away from a feature the more costly those distances are. You can use distance in combination with a friction surface to identify locations of high cost. My experience with cost surfaces is with Multiple Criteria Evaluation.

However, your description sounds a bit different to me. "create a polygone or polyline around a point, at a special value (based on a DEM) (e.g. the power of a traveller for one day)." You might be interested in network analysis, particularly what ESRI calls service areas. These are areas of travel from a point along a road network.

Hope that helps,
David

### #3 dsl Posted 22 April 2010 - 12:18 AM

dsl

Master Contributor

• Validated Member
• 225 posts
• Gender:Male
• Location:Denver
• United States

Accidental duplicate post..

### #4 dav Posted 23 April 2010 - 11:12 AM

dav

Contributor

• Validated Member
• 34 posts
• Gender:Male
• Germany

hey dsl,

In fact what I want to do is the following:

I have 10 ancient settlements in a landscape. I do not know any roads or any other information in vector data (besides of rivers, but this makes it even more complicated). Now I want to create a model that shows the territory of each settlement, assuming that the 10 territories are covering the whole landscape. I know some works, where people used a DEM for creating a cost layer (travelling time in this model ist strongly based on the relief , because nobody knows ancient landuse). So one should be able to create a cost distance layer out of the slope and aspect rasters (e.g. with SAGA anisotropic analysis, if I am right). For a traveller it's easy to go downhill on a light slope, but when it gets steeper it nearly takes as much effort as going uphill (there are some information about those issues). So I don't know if a ready built in cost function in a GIS is considering this or is focusing on other questions and so uses another function.
So my aim must be to write my own cost function, no?

Then finally, to create the territories, i have to use my cost distance raster for each settlement and combining it with the other 9. the boundaries will be created at the place, where the cost values will be the same. so I will have 10 cost-based terriotires. of course this is only a model and starting point. I know someone who used an accurate cost function in combination with a kernel density analysis (on the basis of a fix cost a man travel a day. (unfortunately I don't know that person so well, so that I first need to get involved in this issue)

thanks for your help, sry I am a newcomer in these issues (yet )

### #5 dsl Posted 23 April 2010 - 11:50 AM

dsl

Master Contributor

• Validated Member
• 225 posts
• Gender:Male
• Location:Denver
• United States

You may want to look into Multi-Criteria Evaluation (MCE) and Multi-Objective Decision Analysis. It might be a bit overkill for this project however.

I suggest you use your slope raster surface as a friction surface, and then calculate a cost surface using your territory origins as the origin point. You can include all of them. The GIS will then calculate a cost distance surface based on the origin and slope friction surface. Those areas with a higher cost will correspond to higher distances and higher slopes. You will need to pick a threshold for the territory boundary. I think SAGA will do this, definitely IDRISI and ArcGIS.

Using MCE you can add in other criteria beyond slope. You could include Land Cover. If your territories are near desert, then that would probably be excluded. Or large bodies of water.

Cheers,
David

#### 0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

-->