Thanks for the response on the z option on r.mfilter. A followup
question: Is the "divisor 0" a known bug in r.mfilter? I compared
the results from a 5 x 5 r.neighbors average with a 5 x 5 matrix
of 1's with a divisor of 0. The results for were the same, yet
there were zero valued cells in the map layer.