search

bowlerbear / distributionChange

a project to explore the multi-dimensional nature of species' distribution changes
MIT License
0 stars 0 forks source link

defining core vs marginal regions #2

Open bowlerbear opened 2 years ago

bowlerbear commented 2 years ago

so at the mo, this step takes the predicted map of occupancy probabilities and splitting into two 3 discrete regions - core, marginal and absent - based off the data in 1990:1995. I explain how this categorization is done below, but it looks like this: image

But i need to think carefully how to do these categories. At the moment, i use a moving window to calculate the mean occupancy prob surrounding a given grid. I assign it as "core" if the cell has a mean window occu prob (OP) of 0.5 or higher (i.e., at least 50% of neaighbours likely also occupied). I assign it as "marginal" if the mean window OP is between 0.2 and 0.5 and "absent" if is marginal window OP and also the focal grid is unlikely occupied (occup prob <0.2) (this is because few actually have exacetly zero).

This is obviously all a bit arbitrary - and not sure how to justify it better. Vary the thresholds should vary by species too?? Or I just just cut_intervals or cut_number on the OPs???

bowlerbear commented 2 years ago

Based on the above the results of species change in AOCC within each region, look like this

image

so increasing species change most in formerly absent or marginal areas and decreasing species change most in formerly core areas but i think this has to be true cos of maths....

coreytcallaghan commented 2 years ago

Hey - sorry for never getting back to you.

Yeah, I see your point about being 'arbitrary'. But I think this is fine.

I think the key point is that the figure with the relative changes on the x-axis should show the same general pattern regardless of the cutoffs, right? So, in theory, they are all relative to one another. So I guess I would just do 2-3 more 'cutoffs' to show that the pattern stays the same...

I think anyway, as you mention because of the maths.

P.S. I'm happy to block off all morning tomorrow if you wanted to work on this a bit and push this a bit further along.