I am developing a model where I aim to estimate the bimodal landscape orientation (azimuth angles) of ridges, valleys and fjords. The purpose of this work is to develop an index on how landscape orientation relate to bird migration from a certain direction.
Calculating horisontal landscape orientation is straight forward but gives in my opinion only meaning for regular shaped features with a distinct directional shape (lets say from east to west). The challenge is how to estimate the direction of irregular shaped features (having a very curly perimeter of outliers and inward bends). I would like to use a moving window to calculate mean angles of sub-polygonal elements of each irregular polygons in order too measure how these mean azimuth angels deviate from the master polygon mean azimuth angle (along its long axis). I also want to figure out how to use morphological shape metrics to estimate the angular shape dispersion of each irregular polygon (in order to remove those polygons that disperse equally in all directions).
Any ideas out there?