The TPI is the difference between the elevation of a given cell and the average elevation of the
surrounding cells in a user defined moving window. For landform classification, the TPI is
first standardized and then thresholded; to isolate certain classes, a slope raster (which is
internally worked out) is also needed. For details about the implemented classification, see:
http://www.jennessent.com/downloads/tpi_documentation_online.pdf.
Two methods are available: -the first (devised by Weiss) produces a 6-class landform
classification comprising: -- valley -- lower slope -- flat slope -- middle slope
-- upper slope -- ridge -the second (devised by Jennes) produces a 10-class classification
comprising: -- canyons, deeply incised streams -- midslope drainages, shallow valleys --
upland drainages, headwaters -- u-shaped valleys -- plains -- open slopes -- upper
slopes, mesas -- local ridges, hills in valleys -- midslope ridges, small hills --
mountain tops, high ridges
The second classification is based on two TPI that make use of two
neighborhoods (moving windows) of different size: a s(mall) n(eighborhood) and a l(arge)
n(eighborhood), defined by the parameters sn and ln.
Besides rasters representing the different landform classes, the function optionally returns the
TPI raster, either un- or standarized. The output rasters are plotted on the R graphic console
and returned by the function (as objects of 'RasterLayer' class) within a list.