I have a forest cover image having 5 land cover classes (very dense forest, mod. dense forest, open forest, scrub and non-forest) and a road shapefile with five types of roads. I'm thinking of creating a new image containing only two classes (forest and non-forest) and calculating fragmentation index (km/sq.km)(which is basically road density) per 10 km cell. But how do I calculate the index separately for each class?
Dear Adrian,
I do not know which GIS software you are using, but they all have an intersection function that allows you to create a new layer that includes only the roads present on one class of land cover. You can then extract the intersected roads' lengths from the attributes table. With more information on your software, you could probably get a more precise answer.
That's for the technical part. However, conceptually, when you have different types of habitats, the fragmentation of an habitat might not only be due to roads but also to borders with another habitat. In you case, with two habitats, the length of borders is the same for both habitats, but with more than two, some habitats could have longer borders than others.
Hope that helps!
We have addressed some of these questions in the past, and basically have to come up with some refined criteria for habitat, such as when is a low density forest with savanna understory a non-forest, and when does it function more like a forest such as a woodland with still some grassy or forb understory. Also how small should the polygons go to be considered either forest or non-forest, how big should an inclusion be such as a small patch of trees in a savanna, is still functioning as a savanna, and a small opening in a forest is still forest. I am unsure of a tull technical agreement, but such as a low density forest of 10-20 BA(basal area)/acre of trees with savanna understory is still considered grassland for habitat purposes. Woodlands commonly 40 BA/acre of trees even with grassland understory from perhaps burning is often considered forest. These kinds of decisions are typically best resolved by discussion or agreement with wildlife biologists, ecologists and foresters or silviculturalists, and may vary with the circumstances and importance of specific habitats and fragmentation to species involved. Also you might want to look at type of road, whether it is open for full time use, or a temporary or intermittent road, open seasonally or just for specific activities at some low frequency, but heavily grassed and relatively narrow in width, and closed to traffic uses and disturbances most of the time. And it never hurts to analyze the problem with one set of criteria, and then reanalyze with one or more other sets to make a slightly different point relative to the situation. In GIS, you can set and describe polygons of tree density or habitats and roads to perfect its utility to your subject, which probably has to do with habitat fragmentation. The difficulty as your question suggests in how to best look at road fragmentation to a forest dominated landscape with mixture of non-forest. The approaches you use will determine how best to accomplish. If small inclusions of forest or non-forest of a select size are to be combined with the dominant land cover type, then that would be your first step, then decide if you want to evaluate any road or just certain types of roads, and whether skid trails or landings of timber that are still barren should or should not be included as a disturbance of habitat.
Dear Charlotte,
Thank you for your input. I am using ArcGIS 10.0. I have extracted the road length as per your suggestions.
By borders, do you mean that with more habitat types, there would be greater edge effect among the different habitat types?
Dear William,
Thank you for your suggestions. Some of the things you mentioned such as road type and usage have already been considered. I'm a little unsure as too whether to consider clumping all forest types into one class or to consider them as separate classes since each forest type would respond differently to road effects. And wouldn't other factors such as patch size, shape, their configuration and also their area-edge ratio affect their response to roads?
Exactly, with more habitat types, there would be greater edge effect among the different habitats (rather than just with roads), and these edge effects might differ between habitats.
- Badges
- Science topic
- Similar topics
- Geoscience
- Geoinformatics
- Geographic Information System