Moving window analysis is a prominent means of analyzing the spatial variability of landscape patterns at multiple scales.
A new computational framework is presented that overcomes technical and computational barriers to the use and implementation of moving windows based landscape pattern analysis of raster maps.
For a small window of 41 × 41 pixels, computation time was reduced by a factor 600 compared to the most commonly used software. These gains will be greater for larger windows.
The framework facilitates distance-weighted moving window analysis, which currently is not commonly used. Distance-weighting prevents discontinuities due to the discrete delineation of windows and is the natural choice for geographical analysis.