Hi Ed,

Thanks for reminding about the SREF plumes below... Now that we know what verified for this week's 
NYC-LI snow event, attached are the SREF plumes for the QPF at LGA for each of the runs on the 19th 
(33-42h lead time), as well as the mean snowfall plumes... Not one member from the 19/03Z run had 
significant precip, and there was a significant d(SREF)/dt towards more snow in the mean, albeit 
lots of spread. It is disturbing that there was a 0% chance of significant snow at 42-h lead time 
in the SREF, and it appears that all models (at least US models -- GEFS and NAM) had a low 
probability of snow for (00-12Z on 19th), which transitioned high probability of heavy snow 
by 18-00Z on the 20th as the simulated storm became more intense and closer to the coast... 
This common issue among models suggests that resolution/physics were not the big story, but 
rather some very important piece of initial condition info upstream (ensemble sensi suggests 
western Canada may be one place to look...). This case will likely not make the NWP "drop-out" 
list as defined by large-scale metrics, but it is just as interesting... The only good news 
is that because the NYC area had a major snowstorm earlier in the month, most people took the 
change in forecast in stride (more acclimated to the snow). If this week's storm was the 
first big snow of the season, the rapid change in the short-term forecast may have caused 
more of a stir...

-Brian