Perimeter?

[Scott Tweddale]

Regarding calculating perimeters of "clumps", I would suggest running
r.grow, which essentially adds or "grows" one pixel around all continuous
areas in your input map.  You can then subtract your original clump map
from your r.grow output map using r.mapcalc and then run r.report on
your results.  However, this will only give you total perimeter for 
all category values instead of perimeters for each individual clump.
Also, r.report would only give you areal coverage of linear features based
upon your current resolution, and would not actually provide linear perimeter
measurements.  

Depending on the complexity of your map, d.measure might also be an option.

While on the subject of r.clump, we use a script written by Michael Shapiro
of USACERL which takes an input raster map and produces an output raster
map which contains only contiguous areas above a minimum area supplied by
the user.  This program is often used after running r.clump to eliminate
clumps below a minimum size.  If there is an interest, I can provide the
script.

Scott Tweddale
U. of Illinois GIS Laboratory/USACERL