Identifying phenomenon trends

[W. Fredrick Limp]

There are a couple of potential ways:
i.zc would probably be effective in locating the edges and you might try
that.  You could use two neighboor filters for max and min and then
use r.mapcalc to calculate the difference.  The largest
difference is the maximum tone change within the neighborhood. Or you
could use r.mapcalc's own "neighborhood" operator map[-1,+1] etc. to
identify the maximum difference in the defined neighborhood