Crowdsourcing Snowstorm's Westward Model Shift

I was surprised to see this HRRR forecast model prediction of tonight's Northeast snow showing no snow for Harrisburg and York, PA, and showing the axis of heaviest snow (4-8") over or west of State College, PA. This disagreed with overnight AccuWeather and NWS forecasts that showed it further east. This storm will be a good test of last minute "nowcasting" by the new higher-resolution models that we have access to this winter season. I thought I'd "crowdsource" this forecast on the WeatherMatrix Facebook page so my readers could weigh in.

Henry Margusity reminded me that the HRRR model may not be seeing far enough ahead to predict the entire accumulated snowfall. This was from the 10Z morning run, which only sees out to 01Z tonight (about 8 PM). It makes sense that more snow will fall over the higher elevations of Pennsylvania early (because the cold air is there and to the northwest); you can see these mountains in the topographic map below.

Joe Moore commented: "Check out the RR (or, as it's new name, the RAP), which goes out further than the HRRR and is much closer to becoming operational." [The map through 05Z, or ~1 AM is shown below. It's not as high-res, but it shows a similar solution, even with the later snow].

Since I don't follow these things closely, I asked Joe what he thought of the RR model. He replied "I think the rapid refresh is fairly reliable, but still far from perfect. Also, the rapid refresh is initialized with the GFS boundary layer conditions, as opposed to the RUC which uses the NAM. (This is due to the larger domain of the rapid refresh.) And, as the name implies, the HRRR is initialized off the rapid refresh."

Jenn Grover pointed out the comparison of two different versions of the WRF model's simulated radar, shown above for 10 PM, one of which had the axis further west -- but one that didn't. Yet another thing to consider.

So what does it all mean? For this situation, the high-res models tell us that the snow storm is likely trending west, but whether or not it really will, we'll have to wait and see. Although the infoglut provided by social media could overwhelm forecasters, even the best meteorologists (and I do NOT include myself in that category) can use crowdsourcing to find out additional important information, especially in extremely short-range forecasting (nowcasting).

This is an example of how Social Media is revolutionizing weather forecasting, something I'll be writing about in WeatherWise magazine's Jan-Feb. 2012 issue, and it's not at all unseen here at AccuWeather -- when our company was started 50 years ago, our founder Joel Myers noted that the average consensus forecast of his entire meteorology class would always beat the best daily forecasters - which is why we have a twice-daily map discussion here at HQ to get all of the meteorologists on the same page - an internal crowdsourcing if you will.