Results from simulations of a mesoscale convective system (MCS) that occurr
ed in a weakly forced large-scale environment show significant sensitivity
to the time step size chosen for model integration. As the time step is dec
reased, the movement and longevity of the simulated MCS depart further from
the observations. It is found that the time step-dependent formulation of
the numerical diffusivity is responsible for a major part of this sensitivi
ty in the simulated MCS. The changes contributed by truncation errors may o
r may not be negligible, depending on the physics schemes used in the simul
ations. This suggests that the choice of a specific linearly stable model t
ime step, in combination with the definition of the numerical diffusivity,
may at times be as important as the choice of the basic model configuration
.