The radar looked good but where the hell was the snow? For places along I-84 Tuesday night’s storm was pretty much a dud. There was an extremely sharp cut-off between virtually no snow and a pretty decent amount of snow. In West Hartford I recorded 2.3″ while the CoCoRaHS observer 3 miles east of Southington managed 5.8″ and in Berlin managed 5.3″. Farther south, 8.6″ of snow fell in Shelton and to the east 10.2″ of snow fell in Killingly.
Prior to the event our computer models were showing anywhere between 0.25″ and 0.5″ of liquid in Hartford (the European model was toward the low side of guidance). The setup favored large, fluffy dendrites as the predominate crystal type and >15:1 snow:liquid ratios. The actual amount of precipitation in my backyard? 0.13″ – ouch.
One thing I noticed is that the best radar echoes at 5,000 ft AGL preceded the heaviest snow by quite some time. While some of this was likely due to dry air in the boundary layer sublimating some of the snow (note northerly drain of dry air) I think the larger was because of the snow flakes themselves.
These large and very fluffy snow flakes were gently drifting toward the ground. They fell like parachutes. The fluffier the snow flake – the lower the terminal velocity. Winds through the boundary layer were out of the north around 20 knots – not strong enough to tear the flakes apart (making them less “fluffy” and increasing their terminal velocity) but strong enough to carry them a significant distance from where they formed.
I emailed my friend Bob Hart at Florida State who’s a meteorology professor about the storm in the Hartford area and he agreed with my hypothesis.
i think you are right on the fall velocity.
average fall velocity for a snowflake is apparently 1-1.5 m/s. if we pretend these had half that given their low density — and that the northerly cold flow began at 800mb (?) — don’t recall the sounding. 2000 meters depth would mean the flakes would take about 1 hour to reach the surface. and at an average northerly wind speed of 10 m/s, that means the flakes would reach the surface about 36km further south, or about 20 miles to the south – around Meriden’s latitude.
It would appear that one reason the models were too “bullish” with the northerly extent of the heavier snow on Tuesday night was not because they showed the best lift (and frontogenesis) too far north but because the snowflakes drifted pretty far south from where they were forming. I assume (though I’m not positive) numerical weather prediction assumes a constant fall speed for all snow flakes or maybe doesn’t even account for transport of snow flakes in the horizontal at all?
Where these snow flakes actually fell in large number (of course, not at my house) the snow:liquid ratios were about as great as you’ll see anywhere. In some cases nearly 40:1!
Some of the more noteworthy snow to liquid ratios?
- Oakdale 26:1
- Milford 35:1
- Killingly 44:1!!!!
- Shelton 43:1!!!!
- Bridgeport Coop 32:1
Indeed this wasn’t just an inch or two of snow – some of these ratios were measured with 6″-10″ of snow. That’s Utah kind of fluff.
This may have been part of the forecast bust in Boston too. My buddy Scott Nogueira who’s a meteorologist at WSI said he thought the same issue impacted Boston with models being far too generous and there being an unusual disconnect between snow reaching the ground and radar echoes from above (excluding a case of sublimation/dry boundary layer).
Bottom line is that while the models were too bullish with QPF (part of that is due to some synoptic scale issues with convection off the Carolinas) but part of that bust on the I-84 corridor was because of the wind and the fluff. Apparently we managed to screw ourselves with snow in Hartford because snow growth was TOO good. Go figure.