I have been forecasting the onset of severe weather events for a long time. Of all the things I have learned, there is one truth I hold to most closely. Forecasting for severe weather is done on Day 1.
The use of models to pinpoint the best severe is almost useless. All I use the models for is to get a feel for the overall synoptic pattern. To interrogate the model output to a deeper degree is to play a game of futility. I am much happier looking at 500mb winds, 850 theta-e and winds, CAPE and CIN to get a quick feel for stuff.It becomes apparent when there will be a significant severe weather pattern. It also becomes apparent when there will not be a large outbreak but a chance for a few good storms. Shear, moisutre, lift, instability really is sufficient for forecasts beyond the day one period.
However, once it gets to Day 1, then I analyze the data. Given enough experience, it is amazing what you can learn from a mesoanalysis. I do not need model guidance to know where the threat is most likely.
Do I use models anyway? Of course. Hey, I am a meteorologist. Of course I look at models. If I am excited I look at everything and all models. Give me more output.
But do I need models? I would love to have nothing but analysis - colored pencils, maps and images. I would bet I could beat a model 85% of the time. And believe it or not, that makes me feel good. Who doesn't want to beat a computer.