topic Re: Determining whethera slope goes up or down... in ArcGIS Spatial Analyst Questions

Determining whethera slope goes up or down...

MichalBirkenfeld — Mon, 22 Oct 2012 13:12:15 GMT

Hi,
I'm trying to create a cost surface for hill walking time expenditure, based on slope and distance.
To do so, I need to know whether at a specific cell the slope is positive (= walking uphill) or negative (= walking downhill).
Any ideas?
Thanks
Michal

Re: Determining whethera slope goes up or down...

DanPatterson_Retired — Mon, 22 Oct 2012 15:24:00 GMT

Perhaps Path Distance tools http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#/Understanding_path_distance_analysis/009z00000022000000/

Re: Determining whethera slope goes up or down...

MichalBirkenfeld — Mon, 22 Oct 2012 16:09:58 GMT

Hi Dan, thanks for the quick reply...
I've tried using these, however, the "path distance" requires an existing cost surface.
As for the path distance allocation tool, I couldn't find where the actual factors should be entered (i.e. the conditions that transform a certain slope to time costs).
Any suggestions?
Michal

Re: Determining whethera slope goes up or down...

DanPatterson_Retired — Mon, 22 Oct 2012 17:22:29 GMT

Try searching PathDistance on this forum ie http://forums.arcgis.com/threads/10487-Pathdistance-problem?highlight=PathDistance

Re: Determining whethera slope goes up or down...

MichalBirkenfeld — Wed, 24 Oct 2012 10:03:42 GMT

You were right! Path distance is the way to go, and I've finally figured out how my data needs to be stored to use as input...
However, I've encountered a new issue - I keep getting a "spider-web" type raster, where the cost just increases with distance, with no regards to the other factors.

I'm using a 5m cell size DEM as the surface raster, and a ASCII table with the vertical factor.

I found a very long thread, dealing with the exact same problem:

http://forums.arcgis.com/threads/23888-Problem-with-DEM-and-Path-Distance?highlight=path+distance

I'm not sure whether this makes me feel better or worse, but I was wondering if you had any ideas??

Thanks
Michal

Re: Determining whethera slope goes up or down...

RamB — Wed, 24 Oct 2012 11:36:22 GMT

Hi,

I suggest you work with Aspect map (spatial analyst). For example let us say you are walking north, that is if you are on an average moving north compared to your starting point.

Step1. Develop an aspect map of your study area.
Step2. convert your walking path to a raster (path-raster), but of same size and extent as your study area. Such that cells beneath path are valued 1 and all others valued 0.
Step3. Multiply your path raster with your aspect raster and get the output.
Step4. In the output raster if the values are between 90 degree and 270 degree, then you are walking uphill. if the values are between 270 degree and 90 degree, then you are walking downhill. A negative value in your output suggests that you are walking on a flat surface.

This is because you are walking north and the degrees are measured from north.

If you are walking in any other direction, just play around with the degree ranges to get a solution. Let me know if you dont follow my english.

regards

Re: Determining whethera slope goes up or down...

MichalBirkenfeld — Wed, 24 Oct 2012 13:55:04 GMT

Hi,

I suggest you work with Aspect map (spatial analyst).

Hi Srirama,

Thank you for your reply!
That works well, especially when working with a specific path/direction.

I am trying to model catchment areas, or exploitation areas around archaeological sites, so my main interest is in defining walking time parameters (for example, what area is controlled from the site, within 1 hour walking distance).

I think path distance tools are perfect for that, as they calculate the cost of moving through the terrain, while taking slope and direction into account. Unfortunately, I've encountered a major problem with the tool, which, according to the thread I found in the forum, has been driving people crazy for over a year now... The tool is basically calculating the cost according to distance, with no regards to the vertical factors I've put in. I'm working with a 5m cell size DEM as the surface raster and an ASCII table for the vertical factor.

Here is the thread on the issue http://forums.arcgis.com/threads/23888-Problem-with-DEM-and-Path-Distance?highlight=path+distance

any suggestions?

Thanks!
Michal

Re: Determining whethera slope goes up or down...

RamB — Wed, 24 Oct 2012 20:01:05 GMT

The tool is basically calculating the cost according to distance, with no regards to the vertical factors I've put in. I'm working with a 5m cell size DEM as the surface raster and an ASCII table for the vertical factor.

I understand you are going crazy 🙂 But if i am not asking for too much. Which vertical factors are you considering ? Please elaborate.
Thank you.

Re: Determining whethera slope goes up or down...

MichalBirkenfeld — Thu, 25 Oct 2012 06:05:41 GMT

I'm using an ASCII table that includes the slope (in degrees) and a factor (calculated according to hill walking reconstructions), which adds x amount of time cost for up hill walking or for descending down a very steep slope, and subtracts x amount for walking down a moderate slope.
It basically has two columns. one with the slope degrees starting with -90 and up to 90 and the other with the appropriate cost.

And yes - I am going slowly crazy over this 🙂 I've tried every idea that appeared on the older thread, including transforming the rasters to integers, the source file into a raster and back, etc. everything except magic 🙂

Thanks again
Michal

Re: Determining whethera slope goes up or down...

RamB — Sat, 27 Oct 2012 09:21:44 GMT

Hallo,

Finally I had time to work on this. I do not get euclidean shapes. I get proper anisotropic contours. Are you giving any cost raster as input ? I did not. I guess it get confused with cost raster if you use it. I just used DEM and a point file. It is a 32bit raster with 30x30 resolution. I changed to 5x5 resolution, that also works.

Did I get you right ?

regards,

PS: extent gives a problem I think

Re: Determining whethera slope goes up or down...

MichalBirkenfeld — Sun, 28 Oct 2012 07:00:01 GMT

Hallo,

Did I get you right ?

regards,

PS: extent gives a problem I think

Hi!

Thank you for taking the time to help!

You got it exactly right... I don't get what I've done wrong. Are you using ArcMap 10.0?

The only thing I can think of is the table - how is yours built? is it saved as a .txt file?

everything else is exactly spot-on...

Thanks again
Michal

Re: Determining whethera slope goes up or down...

RamB — Mon, 29 Oct 2012 05:46:26 GMT

Hallo,

Yes, I am using arcgis 10.
1. I have a dem
2. I have sites as points
3. I select one point of interest, so analysis is done for one selected point.
4. the inputs to the model are a point file with one selected location, the same dem as vertical raster also, a text file with toblers method of vertical factors (-90 to 90), factor has 6 digits after decimal, and degrees and factors are separated by a tab spacing.
5. extent can be default or full extent does not matter, full extent took about 100 minutes to solve ( I have a slow computer)
6. Results are as expected 🙂

please check your txt file is tab separated, check units correspond to your shapefile units

regards

Re: Determining whethera slope goes up or down...

MichalBirkenfeld — Wed, 31 Oct 2012 07:59:27 GMT

Well, you got it - the problem must have been the extent.
It took forever to run (apparently my computer is even slower than yours), but with a restricted extent I did get better results.
Now I just have to think how I can double-check the validity of my results... It seems to give a bit too small of a radius.

Thank you very much for your help!
I appreciate it very much!

Michal