floating point | null data in GRASS!
nredding at BBN.COM
Fri Mar 12 17:41:32 EST 1993
The suggestion that you can simulate floating point by a biased
integer assumes that all you care to do is attach a number to a
point. If you are using a GIS to do analyses that involve anything
other than simple addidtion (subtraction) it is not sufficient.
For example, 2.59/.68 = 3.8088, but with integer arithmetic it
becomes 259/68 = 3, while what you'd want is 381. Never mind if
you need logarithms and exponentials. Then there is the problem
of null values when the range goes from very large positive to
very large negative values. It ends up that you write your own
mapcalc routine for each application. A GRASS tool that works in
all situations is clearly preferable, even though you can kludge
something on an ad hoc basis.
More information about the grass-user