I have som problems with an analysis and would be very grateful for any answers or suggestion.

The background is: - I want to examine the covariance between ecosystem services in an area. I don´t really know if covariance is the correct word, however.

I have identified and combined different variables indicating high biodiversity using the weighted overlay and graded the areas from 1-10, whereas 10 is indicating high biodiversity.

Now I want to investigate how the biodiversity covariates/coexists/relates (english isn´t my native language) with other variables.

The other variables could be biotopes, soil type or other ecosystem services, like recreation.

My data consists of: - polygons with the graded biodiversity and also the same areas in a raster. - polygons indicating biotopes, soil type, precipitation etc. - polygons indicating whether an area is used for a certain activity like mushrooming and so on.

The polygons are divided into green spaces at a county level.

So, what I basically want to do is to se if there are any hot spots where biodiversity are more likely to coexist with an other variable. I would like to visualize it on a map but also get a diagram over the relationships.

How do I do?

enter image description here

Oh, I noticed I wasnt allowed to post pictures. I hope you understand anyway!

Thanks in advance!

enter image description here


Check out the Spatial Statistics toolbox. There are a few tools in there that might be good candidates for what you want to accomplish (Geographically Weighted Regression maybe?).

I'm not sure the best way to handle the biotope, soil typed, etc. variables though, as they aren't really numeric values. You might be able to do something like union the variable data with the biodiversity data. Perform the analysis on that and use your 1-10 biodiversity values as a weight. You'd need to do that separately for each variable layer.

  • Thank you very much for taking time! That´s exactly what I´ve tried (the GWR) but have just received gibberish numbers in the result. I don´t know if there are really any explanatory variables that could be used in the model?? How do you mean, more exactly, in the second part?
    – user21070
    Aug 15 '13 at 6:44
  • I mean, one can hardly claim that an area is showing high biotiversty because people is mushrooming in the area. On the other hand might there be a correlation between biodiversity and mushrooming. The mushrooming polygons are just showing 1 for the presence and 0 for non presence.
    – user21070
    Aug 15 '13 at 7:39
  • Well, first you determine the apparent relationships, and then hypthosesize the "why". The idea was to sort of group the biodiversity numbers with a variable layer. Then perform anaysis on that to try to determine strong (or weak) relationships. A lot of the variables you have aren't numeric, so doing a union gives you a way to give those variables some numbers in a common format, which then lets you do some statistical analyses. Thinking about it again, I guess you wouldn't really be using them as weights unless the variable was already numeric.
    – msayler
    Aug 15 '13 at 21:46
  • Your mushrooming example illustrates something to be aware of. Mushrooming and biodiversity might have a strong relationship, but it could be that mushrooming is high because biodiversity is high, not the other way around. Other variables could potentially work like that too. (maybe that was your point, sorry if I'm repeating the obvious)
    – msayler
    Aug 15 '13 at 21:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.