While such
models have not been
run explicitly for Montana, they have been applied to other locations and can provide insights — though
sometimes with conflicting results — for considering impact on Montana wheat.
I am astounded by this, for two reasons: 1) if the
model can
sometimes «go off the rails» for no apparent reason, then it seems unlikely the
model is a reasonable representation of the Earth's climate, and 2) at an absolute minimum, there should be a careful investigation into the stability of
model behavior, with lots of
runs, so that the probability of a «crazy»
run is at least known.