The Omniscape algorithm is based on an omnidirectional implementation of the connectivity tool, CircuitScape. It uses a moving window approach to calculate connectivity. For more information on the OmniScape approach please refer to McRae et al (2016) For this project, we tested several different moving window sizes from 250 meters to 1250 meters. A pixel size of 90 meters was used for computational efficiency. We used the OmniScape Python script provided by TNC. Omniscape requires two inputs: a resistance surface and a source surface. We used the standard resistance surface described above, along with the habitat layer described above as the source surface.
Current FlowThe raw output of Omniscape is converted into into percentiles of value called “current flow”, with values of 1 to 100. For further discussion, we will refer to the 500m window size runs. The image below shows Omniscape current flow outputs.
Classified Normalized Current FlowAnother important way to interpret the OmniScape outputs is to first create the Regional Flow Potential map (see below) by running Omniscape with a value of 1 for the resistance surface, and using the normal source surface, then dividing regional flow potential by current flow to create the normalized flow potential map. The result is then a continuously valued surface that is then classified into different classes to yield the classified Normalized Current Flow. We used the approach described in McRae et al (2016), with assistance from Aaron Jones, with the standard set of five flow classes – channeled, intensified, diffuse: high, diffuse: low, and impeded. There is no set formula for where to draw the break lines between classes, but rather an expert opinion approach based on a suite of considerations: (a) logical distinctions; (b) what seemed to be sensible relative footprints for each class; (c) what would make for more intuitive mapping; (d) natural breakpoints between any modes in the flow’s distribution; and, (e) what we knew to be on the ground.