Reference levels for the connectivity indicator
Opened this issue · 6 comments
This thread holds the discussion about the reference and threshold levels for the connectivity indicator.
The reference value is defined as the mean distance to mire polygons (connectivity) when the analyses is ignoring human infrastructure.
The threshold value for ecological condition is tentatively set to 0.6 (i.e the default value following a linear rescaling).
Probably the ecological condition can still be considered quite good even with a mean of 40% longer distance to surrounding mire polygons. I suggest instead using a negative exponential rescaling, like this:
(source)
The default threshold value shift naturally down when doing this, but one can also define it spesifically, and I think somewhere around 70-90% increasded distance is ballpark where it should be.
I will do this when the downloadning from GEE is finished tomorrow. But I think I will have a larger portion of 0 and 1 values than in your example. Is that a problem?
Thats fine I think. The area aggregated indicator values (e.g. regional) will in any case not be that extreme.
Now I have all the calculations of the connectivity finished in ArcGIS Pro (for several million polygons). So you advice to rescale in this way: Rescaled_Connectivity = exp(-0.6 * [Connectivity]) ? And then find upper and lower confidence interval (2.5 % lower and 2.5 % upper)?
If by [Connectivity] you mean the qoutient, then actually I was thinking
Rescaled_Connectivity = [Connectivity]^0.5
You can get the CI when aggregating to regions, but the reascaling at the polygon level will not produce any variation.
By the way I obtain an average value of 0.715874 and a 95% confidence interval of 0.005124058 for the qoutient