You raise a good point. We know that for some volcanoes, VAFTAD
overpredicts the spatial extent of the plume. This likely occurs
because of steam in the plume or other uncertainties in the source. For
a real volcano, based on satellite imagery, a "reduced ash" version is
run if it would be more appropriate. The "hypothetical eruptions" web
page (
http://www.arl.noaa.gov/ready-bin/vaftadtst.pl) has VAFTAD output
from both the full-ash case and a "reduced ash" case. The differences
in these two products indicate some uncertainty in the forecast based on
the uncertainty in the ash/steam mix in the source. Typically for a
large eruption (greater than about 30000 ft) there are only small
differences between the reduced and standard run, for smaller eruptions
there is a greater difference.
Barbara Stunder