A generalized non-regression technique for evaluating the fractal dimension of raster GIS layers consisting of non-square cells.

Natural landscapes often reveal extremely complex patterns that can only be very rougly characterized by methods of Euclidean geometry. In contrast, fractals can be applied to a variety of landscape ecology problems because they conveniently describe many of the irregular, fragmented patterns found in nature. This paper focuses on a fractal-based measure of landscape complexity for grid-based GIS layers. A non-regression technique for measuring the distribution of diversity within a raster database consisting of square cells is generalized to incorporate any regular shaped grid cell (eg regular polygon, rectangle) that forms a continuous, fully tessellated grid.