Such an expression of relative risk is quite
familiar in human experience. By rank ordering the
alternatives (on the basis of suitability) and
choosing the best ones, we use a procedure that
forms of uncertainty exist. Thus one might anticipate
the problem of having to express both forms of risk.
For example, given the presence of measurement
error in the development of a multi-criteria
suitability mapping, and a propagation of those
errors to the final suitability map, one realises that
this mapping is only one of a large number of
possible outcomes that might be produced by
randomly introducing the uncertainties in
measurement that exist. Thus by Monte Carlo
simulation (a capability that unfortunately exists in
only a small number of GIS software programs) one
could thus tabulate the proportion of simulations in
which each location falls within a specific threshold
of relative risk, or a specific areal requirement. This
then restores the frequentist notion of probability
and the usual expression of decision risk.