DESCRIPTION

Edge effects and core area analysis of landcover fragments. This module can compute static edge effects (defined edge depth) and dynamic edge effects (based on surrounding landscape). The impact of the surrounding landscape can be accounted for and the resulting core area is provided.

NOTES

This module is generating core areas based on defined edge depths. The edge depths can be increased by the values of a costmap (e.g. urban areas could have a more severe impact than secondary forest on forest fragments). Moreover a friction map ( propmap within the fragments can lower the impact of surrounding landcover types and hence an increased edge depth (e.g. a river or escarpment which might lower the edge effects). Moreover a dist_weight can be assigned in order to increase the weight of closer pixel values.

Distance weight

The assigned distance weight is computed as:
w(d) = 1 - (d / d_max)^(tan(dist_weight * 0.5 * pi))
where:
d = Distance of the respective cell d_max - the defined maximum distance dist_weight - the parameter how to weight the pixel values in the landscape depending on the distance
the dist_weight has a range between 0 and 1 and results in:
0 < dist_weight < 0.5: the weighting curve decreases at low distances to the fragment and lowers to a weight of 0 at d=d_max
dist_weight = 0.5: linear decrease of weight until weight of 0 at d = d_max
0.5 < dist_weight < 1: the weighting curve decreases slowly at low distances and approaches weight value of 0 at higher distances from the fragment, the weight value 0 is reached at d = d_max
dist_weight = 1: no distance weight applied, common static edge depth used

propmap

The propmap minimizes the effect of the edge depth and the surrounding matrix. This has an ecological application if certain landscape features inside a e.g. forest fragment hamper the human impact (edge effects).
two method exist:
propmethod=linear: propagated value = actual value - (propmap value at this position)
propmethod=exponential: propagated value = actual value / (propmap value at this position)
If 0 is chosen using the linear method, then propagated value=actual value which results in a buffering of the whole region. In order to minimize the impact the value must be larger than 1. For the exponential method a value of below 1 should not be chosen, otherwise it will be propagated infinitely.

EXAMPLE

An example for the North Carolina sample dataset using class 5 (forest): For the computation of variable edge effects a costmap is necessary which need to be defined by the user. Higher costs are resulting in higher edge depths:

# class - type - costs
#   1	- developed - 3
#   2	- agriculture - 2
#   3	- herbaceous - 1
#   4	- shrubland - 1
#   5	- forest - 0
#   6	- water - 0
#   7	- sediment - 0

r.mapcalc "costmap_for_corearea = if(landclass96==1,3,if(landclass96==2,2,if(landclass96==3,1,if(landclass96==4,1,if(landclass96==5,0,if(landclass96==6,0,if(landclass96==7,0)))))))"

now the edge depth and the resulting core area can be computed:

r.pi.corearea input=landclass96 costmap=costmap_for_corearea  output=landcover96_corearea keyval=5 buffer=5 distance=5 angle=90 stats=average propmethod=linear

the results consist of 2 files: landclass96_corearea: the actual resulting core areas landclass96_corearea_map: a map showing the edge depths

AUTHORS

Programming: Elshad Shirinov
Scientific concept: Dr. Martin Wegmann
Department of Remote Sensing
Remote Sensing and Biodiversity Unit
University of Wuerzburg, Germany