As late as Friday morning, weather radars across the country showed nary a storm system — even as forecasters throughout the Midwest and eastern half of the U.S. were warning of bitter cold ahead and projected snowfall totals in the feet, not inches.
“If you just pull up the radar, you’re missing all that upper air data,” said Kyle Elliott, director of Millersville University’s Weather Information Center.
That’s where satellite imagery comes in, he said. Satellites capture the upper air that the radar doesn’t and indicate “a storm system even if rain or snow are not showing up on the radar yet.”
Technology, research and experience all factor into the confidence seen in the weather predictions that this week — long before the first snowflake fell — triggered governors to declare states of emergency, flights to be canceled and government officials to start working days ahead to mobilize a response what’s expected to be a massive snowstorm.
Yet even the technology requires sophisticated interpretation and adaptation as more data comes in.
The satellite imagery, for example, combines with general climatology, weather modeling equations and pattern recognition to allow meteorologists to predict what the next days and weeks might bring — even if the radar map appears bare.
Read More
Numerical weather prediction models involve inputting the current conditions — from radar, satellites, weather balloons and more — into a computer that uses different equations to make predictions.
The models, Mr. Elliott said, are tools that offer guidance — not gospel.
“A small error at the beginning can become a big error five or six days down the road,” he said.
In the case of the snowstorm expected to drop upward of a foot of snow across Western Pennsylvania this weekend, models were initially showing the heaviest snowfall across Virginia, with Pennsylvania just catching the northern edge of the system.
“When I looked at these models — I’m really skeptical,” Mr. Elliott said. “That’s not normal, not typical, and the odds favor this system tracking north.”
The pattern recognition portion of forecasting means looking at the weather data as it stands now and looking for the closest match — called an analog — in historical weather data.
In this case, the conditions earlier this week most closely matched those in the days leading up to the blizzard that hit the East Coast over President’s Day weekend in 2003. That storm ended up hitting hardest farther north than the current storm’s early models predicted.
Sure enough, Mr. Elliott said, predictions for this weekend’s storm system shifted several hundred miles over the course of the week, mirroring the 2003 storm.
On top of that, he said, climatology suggests it’s much more within the realm of possibility that Pennsylvania gets 20-plus inches of snow than Richmond or Virginia Beach.
At this point, it appears Western Pennsylvania can expect 10 to 14 inches of snow, according to the latest forecasts from the National Weather Service office in Moon, with the potential for higher amounts in some pockets.
A cold weather advisory is also in effect through Saturday morning, with wind chills in some parts of the region as low as minus-20 degrees.






Comments
Welcome to Post-Gazette subscriber-only commenting community. Please review our community guidelines.