unige-geohealth / accessmod

accessmod 5 : anisotropic accessibility analysis.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Knight's move required

fxi opened this issue · comments

As requested by @SteeveEbener, we should let the user decide if the Knight's move should be enabled for isotropic and anisotropic analysis.

I've ported this option for the transportation mode switcher back in 2014:

https://github.com/fxi/AccessMod_r.walk/blob/a29fa9efb82f616f33cfc3039885bc7499681123/accessmod/timeCostManager.h#L171

It was decided not to display it in the UI, but I don't remember why. Probably for ease of use and performance.

TODO :

  • Low level tests with/without knight's move
  • Add a checkbox in settings:

  • Use Knight's move
    Help text : 'Knight's move' produces a more realistic catchments, but requires more computation time.

Details :
https://grass.osgeo.org/grass82/manuals/r.cost.html
image

This should indeed be available as an option to the user. We had indeed decided at one stage not to have it implemented as default because of the added computation time. But it is now needed, especially for small catchment with walking scenarios that can often generate circular catchments.

Available in fredmoser/accessmod:5.7.21-alpha, and as update in AccessMod Desktop + Docker VM.

After installing 5.7.21 alpha we are getting the following error message when trying to run the scaling up analysis using or not the knight's move:
photo_2022-08-19_17-58-00

Ok.
Can you share the config file, if available ?
data >> filter >> config >> select, then archive >> export
Demo project or another project ?

--

On my side, demo project, it worked.
image

I've pushed a patch : 5.7.21-alpha-1.0
Is this happening with it that version ?

The project presenting the problems is in this dropbox folder: https://www.dropbox.com/sh/f99170i729vek5c/AADQsTBQQw_tvypGWw4LIMrEa?dl=0

Can't manage to get the config file.

Will try with the patch

Just tried with the patch and the error message is still appearing.

Input data and parameters used:

  • population (both) ; moc adjpop
  • merged landcover: ldc single class
  • scenario: scenario single class
  • suitability: pop within 5 km radius - highest more suitable
  • capacity: min: 0 max 1,000,000 capacity 1,000,000
  • start empty layer
  • isotropic
  • use knight move
  • maximum travel time: 60 min
  • max pop coverage: 80%

Thanks for this information. I will try that. I was not able to reproduce with that project, using various parameters.
image

Ok, it should be fixed in 5.7.21-alpha-1.1

It was a bug outside the scaling up analysis, during the creation of the config file. ( used to relaunch an analysis without using the UI or sharing a configuration when a bug occurs )

Great! I will test this as soon as possible and appearing in the list of new versions

Just tested with 5.7.21-alpha-1.1 and it is running!

I will run some more tests and close the ticket in case everything is running correctly

Thanks Fred

Tested for both the geographic coverage and scaling up analysis under 5.7.21-alpha-1.1 and both are running as expected. Thanks Fred.

Closing this ticket