At weather expo, top snow expert shares views

At the American Meteorological Society annual meeting in Seattle, the crowd of top weather and climate scientists has been buzzing about the potential East Coast storm event. Today I caught up with one of the foremost experts on winter weather forecasting, Dr. Louis Uccellini, to get his take on the situation. In addition to getting his thoughts on the forecast, I also asked him why the computer models have struggled to agree on projections for several winter storms so far this season, especially compared to last winter.

Uccellini, is the longtime director of NOAA's National Centers for Environmental Prediction (NCEP), located in Camp Springs, Maryland. Numerous forecasting centers are housed within NCEP, including the Hydrometeorological Prediction Center (HPC), which plays a key role in predicting winter weather events. Along with former Weather Channel winter weather expert Paul Kocin, Uccellini wrote what is essentially the bible of East Coast winter storms. Uccellini is also President-elect of the American Meteorological Society.

On how confident he is that this storm will result in a moderate to heavy snowfall in Washington, D.C. and nearby suburbs:

Uccellini hesitated to say he is very confident in the computer model projections at this point, even though they seem to have largely come to a consensus. Instead, he said the "trendline towards confidence is clearly pointing up, in terms of a significant snow event."

He explained that the models are catching onto a key reason why there may be just enough cold air available during this storm than was initially thought. "We're clearly trending toward a colder solution since yesterday, and I think it makes physical sense," he said.

A storm that moved through New England today, dropping 2-4 inches of snow in the process, has helped reinforce what is known as a "confluent zone" to the northeast of Washington. "That upper level confluent zone is important for keeping cold air in for us," he said, adding, "It's not a textbook confluent zone, but it's an added factor that's helping to keep the colder air in."

As we've noted in our forecasts, the earlier computer model runs had shown warm air moving northwards at the middle and lower levels of the atmosphere, keeping precipitation as rain or mixed rain, sleet, and snow. But the latest projections don't show such a warm air intrusion.

"I'm more confident that the storm will produce heavy precipitation over the mid-Atlantic region, and that it would produce snow," Uccellini said.

He said the models have been showing the possibility of a period of very heavy precipitation as the storm intensifies near the Washington metro area. "I'm not absolutely convinced that for this event... we're going to get that, but the model is certainly pointing to a slantwise convection aspect and we have to pay attention to that."

On the model flip-flops leading up to this, and other, recent winter storms:

Uccellini said that when he examines computer model projections, and their fluctuations from run to run, he tries to determine "if they're occurring within an envelope of solutions... [which are] usually established several days in advance." He noted that if the models depart from those bounds (as at least one model run did on Sunday evening), "then you really know you have a serious [forecasting] problem." In this case, by and large the models were flip-flopping within a particular envelope.

On why predictions for this year's storms have had less skill, especially further in advance, compared to last year's major storms:

Uccellini said he is not surprised that computer models are having a tougher time handling this year's storms, and that the perception that there are more "predictability issues" this year is probably valid. "In terms of storm track and precipitation amounts, there was definitely more consistency" last winter, he said.

But why such a difference from one year to the next, if the computer models have essentially remained the same, and forecast skill is thought to be improving with time?

Uccellini pointed to differences between the origin of winter storms, and the large-scale weather patterns in place, this year compared to the winter of 2009-10 as accounting for much of the disparity.

Last winter was dominated by a strong El Nino, which led to a very intense southern jet stream that transported waves of low pressure off the Pacific, marched them across the country, and then directed them up the eastern seaboard. This year, there is a strong La Nina in place, and the northern branch of the jet stream is dominating the weather in the eastern half of the country.

"My sense is that last year's very strong El Nino, and very strong southern jet stream... that the way the waves [of low pressure] developed in the southern stream and remained essentially coherent along the southern states, lent itself to a more consistent forecast from run to run."

Simply put, Uccellini said, the models don't have as much skill at dealing with storms that originate from within the northern branch of the jet stream versus southern stream systems. "I've not been surprised about the challenges in predictions of these storms versus last year," he said.

The reasons for this are somewhat complicated, but they can partly be traced back to the lack of observational data in the polar regions. Without accurate and timely observational data, computer models won't necessarily simulate realistic projections of future weather conditions (the old "garbage in, garbage out problem").

"I've said all along... that the cross polar flow regime was less predictable than the flow coming in off the Pacific," he continued. "We actually observe the Pacific quite well."

Uccellini said that this year, storm development is much more dependent on the phasing between weather systems embedded in the northern jet, with energy rippling along the weaker southern jet stream. The computer model vacillations often relate to aspects of this phasing process, which, as we saw during the Dec. 25-26 event, can be critical to determining whether Washington receives any snow at all.

"If you have to rely on northern stream systems for east coast storms you're going to have a lot more uncertainty," he said. "I have felt for the last ten years very strongly that we haven't paid enough attention to the northern stream, and this year is proving that point."