I have a land cover map of forest/non-forest pixels. Based on this map, I would like to separate my landscape into various fragments.
As for now, I have used a distance threshold to regroup neighboring pixels into fragments, but this seems to also include potential corridors and stepping stones, resulting in very large fragments.
Does someone know how I could integrate shape or other criteria to effectively identify fragments vs corridors in the landscape?
Hi Marie,
You could use the LecoS plugin in QGis to identify the forest fragments.
This is the reference of this usefull tool:
Article LecoS - A python plugin for automated landscape ecology analysis
Best,
Sacha
Thank you Sacha!
I don't know if I missed something but this seems to do pretty much the same thing as I already did, grouping neighboring pixels into fragments.
My concern is, see the attached image for an example, that sometimes these fragments are actually corridors, and often corridors and actual fragments are grouped into a same fragment because of their proximity.
I'm wondering if there is a way to distinguish the corridors from the fragments based on shape, number of surrounding pixels, etc..?
Did you explore raster to polyline to extract corridors? If successful, neighborhood operations (maybe preceded or followed by polygonization) could yield patches followed by selecting small patches ('fragments').
You may also consider FRAGSTATS
Hi Hein,
Thank you for your reply!
I am not sure I understand though, you're saying that by vectorizing my raster it will be easier to distinguish fragments from corridors using the neighborhood analysis? How so?
Hey Marie-Lyne,
what software are you using for your analysis? You can check clump function from R package raster (https://www.rdocumentation.org/packages/raster/versions/2.6-7/topics/clump).
You can use it to identify connected patches of cells, and you could use some criteria to remove patches that should not be considered as "forest patch" (minimum number of cells in a patch, patch shape, or maybe even use a moving window filter to remove salt & pepper patterns)...
Mirza.
Hi Marie,
So from what I understand, you need to make a classification of forest fragments.
You can use a software called Guidos tollbox that makes Morphological Spatial Pattern Analysis.
http://forest.jrc.ec.europa.eu/download/software/guidos/mspa/
Best,
Do you know the "Patch Analyst" plugin for ArcGis?
It calculates landscape metrics such as minimum and maximum size of the fragments, area, size and edge effect, etc. It works with raster or vectors, returns you as a result a worksheet with several values referring to landscape parameters.
In this link you will have access to the download:
http://www.cnfer.on.ca/SEP/
Article Geospatial analysis of forest fragmentation in Uttara Kannad...
- Badges
- Science method