Ensemble Cloud Model Applications to Forecasting Thunderstorms
or
The Ensemble Heresies

Using Cloud Scale Models

When a forecast is made using a single cloud model run, to which particular storm does it apply?
Observed storm behavior is variable, so how is a single realization verified?
With more sophisticated models that contain numerous storms and time-varying environments, how literally should results be interpreted?
If not interpreted literally, how are error bounds placed on results?
Deterministic forecasts do not explicitly or objectively provide guidance about possible forecast errors.
If a model run provides only one possible state, how likely is the atmosphere to attain this state?
What is the range of other, equally plausible states it might also attain?