Over the years meteorologists have decided the best way to forecast changes in the atmosphere is to treat it as the physics problem it is. That means computer models full of math. Weather modelling is among the most complex tasks run through supercomputers.
Here’s the problem. The atmosphere is infinitely complex. Any change (or observational error) anywhere affects everything else. Our models can only hope to approximate the actual truth.
They get pretty close, but on a night like tonight where the difference of a few degrees (temperature or latitude, take your pick) can make a difference the computers fall short!
Right now the 18Z NAM is calling for half a foot of snow at Hartford before an icy mix glazes everything and puts a cap on accumulations. The 18Z GFS says under a half inch of snow followed by sleet then rain.
The forecast is easier near the shore where the turnover to rain will happen sooner (if there’s any snow at all).
It’s not that these models are so different. It’s just a tiny difference decides snow versus rain. Think on/off switch versus volume knob.
Both solutions are tough to buy into! My gut feeling is to take a compromise. That’s what I’ll be doing on the news tonight.
More than likely the final call will be 2-5″ inland before the changeover with mostly slush then rain on the shore.
Is this guaranteed? Hello?
What does look reasonably certain is the onset of precipitation after noon. That means Wednesday morning is good, but Wednesday night isn’t.
I’ll be up most of the night checking, rechecking then rechecking again. I never sleep easy before it snows.