I'm trying to evaluate fragmentation of an ecosystem. While It is easy to calculate in ArcGIS 10.1 values such as Area, perimeter or minimum distance between fragments, I'm having problems calculating other parameters that are used in software such as Fragstats. I tried to use this software but I have problems with the memory, as my raster have a very high resolution.

I wanted to calculate first the "edge effect" also called Core Area in Fragstats. I mean, if I subtract for example the first 100 m from the border of those fragments, how much area do remain? I tried to make this with buffers but it is very difficult to calculate polygon by polygon.

Is there any way to calculate this more automatically? Maybe another options to calculate similar parameters to those of Fragstats in ArcGIS?

  • Have you established a minimal mapping unit for your raster and applied a sieve filter? This is an important step when using classified raster data for evaluating habitat characteristics as it mitigates the "salt-and-pepper" effect. This also makes things more computationally efficient. – Jeffrey Evans Jun 2 at 15:13
  • @JeffreyEvans This is a useful comment but, if the map is correct, applying the sieve tool will lead to an underestimation of the fragmentation. I would not apply it in all cases, only after considering if it would disrupt the connectivity patterns of the species of interest. – radouxju Jun 2 at 15:48
  • @radouxju, this is true but, many people do not consider the spatial structure of a resulting vegetation classification and it can have a huge influence in evaluating fragmentation. Now, to the question, FRAGSTAS is functionally defunct. Two suggestions: the landscapemetrics R package or, have you tried Guidos (forest.jrc.ec.europa.eu/en/activities/lpa/gtb)? The Guidos model functionally does exactly what you describe using mathematical moropmetrics and yields a classification of cores, corridors, and peripheries. Very nice model with corresponding body of literature by Peter Vogt. – Jeffrey Evans Jun 2 at 16:03

If you work with vectors, apply a negative buffer of 100 m n all your polygons. Then you can measure the area of the remaining polygons intersected with your grid of interest.

If you work in raster, you can use the shrink tool, then create a binary raster (0,1) where the shrinked value equals the initial raster (with the Con tool), then use a focal stat to have the mean of your binary raster (=the proportion of pixels that are away from a boundary)

| improve this answer | |

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.