ID, MT and OR: Oh my

Unlike Dorothy and Toto we were whisked away to Montana ... far from climatology for this time of year.

Efforts centered on diagnosing how this low CAPE low moisture environment would yield severe storms and if those storms did indeed arise if they would be severe. The shear was plentiful and the mountains, and accompanying upslope flow, appeared sufficient and necessary for storms across ID and OR. Montana was its own temptation and indeed I have been thinking all day about model cancer given the robustness of model storms there. Multiple lines of what appeared to be squall lines with UH tracks were present.



It is indeed difficult to swallow the model solutions when dew points exceed 52 in central MT from model precipitation and then as a result suggest signals we will use for forecasting subsequent deep convection. The real atmosphere seemed less hospitable to deep convective coverage (number of storms and area those storms covered).

Yet again afternoon updates seemed suboptimal. Whether this was meteorological in origin or personal was up for debate. It is our hope that as an event nears, uncertainty will decrease. We should be evolving to a state of more confidence. This can only occur when observations and model solutions converge or at least are similar enough to warrant both confidence and reduced uncertainty.
 After all I can be confident I made a poor forecast even when the uncertainty is small! Models were consistent and thus one could argue they were certain. But this didnt make us feel any better. There were enough "signals" in the real atmosphere and perhaps some "noise" in the model output. In the end our team decided to side with the models, with hesitation. This is an experiment and sometimes you need to be wrong or think you will be wrong, in order to understand more about the forecasts. Then we can learn about our forecasting techniques and procedures to improve next time.

[Saved space for tomorrows forecast evaluation]
Update: Storms overnight did indeed form a line and multiple clusters over MT. Model signal of a dew point pool kind of verified. Loosely. The signal the models were sending us is that once storms formed, they would be ale to boost the dew points where they rained. As far as the big area of moisture pooling, it was totally overdone in areal coverage. It was just that the NME doesnt have a large number of observations, especially within the large area of max dew points, to really dampen that model precipitation signal.

The verification for our specific forecasts were not too shabby. We managed to capture most reports in the full period 15% probability and captured the evolution of the cluster and elevated line that managed to make its way to MN overnight.