search

pierreroudier / clhs

A R implementation of the conditioned Latin Hypercube Sampling method
12 stars 9 forks source link

Scale/range of the cost layer #14

Closed mim25 closed 4 years ago

mim25 commented 4 years ago

Hello, Apologies if this is simplistic.

What is the scale/range of the cost layer? 0-100, 0-10, -100 -100, weighted etc. ? and Can negative scores be used i.e. to define preferred sites.

Thanks in advance

dylanbeaudette commented 4 years ago

I'm not sure, but in the past I have used weights of 0, or near-0, for "easily accessed" parts of the landscape such as roads and improved trails.

pierreroudier commented 4 years ago

I does not matter, because the algorithm is simply trying to minimise that cost.

Following a great idea from @dylanbeaudette back in the days, I have mostly designed it to be used with a cost layer derived from eg GRASS GIS's r.walk.

Negative scores could be used, although like @dylanbeaudette I tend to just flag them as zero.

Please do re-open the issue if you spot a behaviour that is inconsistent with these notes.