The foot package provides functions to calculate summary statistics of
geometric measurements of building footprint polygons. Footprint shapes
representing buildings are becoming more widely available by being detected
and extracted from very high resolution satellite imagery. Such datasets are
spatially detailed but often are unlabelled. However, the size, shape, and
distribution of buildings can suggest possible land uses or differences in
structure use, socio-economic status, etc.
The foot package is designed to provide a set of consistent and
flexible tools for processing 2D vector representations of buildings. The
functionality includes basic geometry and morphology measures, distance and
clustering metrics. These calculations are supported with helper functions
for spatial intersections and tiled reading/writing of data.
The measurements in foot include: area, perimeter, nearest-neighbour
distance, angle of rotation for a bounding rectangle, as well as a binary
indicator of structure presence and a count of structures.
These measures can be summarised as a mean, standard deviation, coefficient of variation, the nearest neighbour index of clustering, or a (normalised) entropy measure for the angles. The output can be formatted as a data table or as a gridded dataset.
The foot package provides convenience
functions (calculate_footstats and
calculate_bigfoot) to wrap common analysis steps
together, taking a list of measurements and parameters and returning a
collected output.
In addition to bulk processing helper functions, there are additional utility functions supplied with the package to support identifying nearest neighbours, adjacent raster cells, creating zonal indices for spatial data and providing efficient I/O and parallel processing.
This work was undertaken by members of the WorldPop Research Group at the University of Southampton(https://www.worldpop.org/).