It’s been all or nothing for winter snow in Philadelphia for the past 4 years. In fact, we’ve only had one seasonal snowfall anywhere close to average in the past 7 years. What’s going on?
Things have really changed. As I explained last year, some of the “rules” of winter forecasting no longer seem to apply. I speculated that it might have something to do with the unbelievably fast ice melt in the Arctic.
The ice melt has accelerated since about 2009, and we’ve seen dramatic changes in weather patterns since then. No one can prove that one is the direct cause of the other, but it seems logical that there is at least some connection.
THE AO, NAO, & SAI:
My friend at The Weather Channel, Stu Ostro, has documented changes in the Arctic and surrounding areas, largely related to HIGH pressure. Big highs to the north force storm tracks farther south. This seems to be connected to the huge snow seasons here in 2009-10 and 2010-11. High pressure in the Arctic leads to a “negative AO”, or Arctic Oscillation. The –AO is clearly connected to increased snowstorm chances in our area. It is also related to the –NAO, or North Atlantic Oscillation, which has an even stronger connection with our snowstorms. If we can predict whether the AO and/or NAO will be positive or negative, we can do a good job in predicting the upcoming winter overall. Winters with an overall +AO have very low chances of being big snow winters, and the higher the AO, the lower the chance.
Until recently, the AO and NAO were thought to be basically unpredictable beyond 2 weeks. But some new research (some published; some not) suggests that has changed, too.
The first breakthrough came from Dr. Judah Cohen from a company called AER. His research found a connection between October snow across the globe in Eurasia and winter snow in the Northeast U.S. I heard about his research long before it was published, so I’ve been following him for nearly 10 years. It has been a better predictor than anything else I’ve seen. Look at how well he did worldwide for last winter!
The one area he did badly was the U.S. Of course, most of us don’t care about any other part of the earth, so the forecast was a “bust” here.
But even Dr. Cohen’s method has changed. He used to look at total snow cover at the end of October. Now it’s the rate of change of the snow during October. That makes a difference this year, so, even though the Eurasia snow was the 4th most ever recorded, the rate of change wasn’t very large. His new forecast tool is called the SAI (Snow Advance Index). That strongly favors a +AO overall for the winter. But he admits that it conflicts with the total Eurasia snow cover, and that’s the first time he’s seen that strong a difference. He still says other factors lead him to predict a +AO. (My forecast was made 24 hours before Dr. Cohen was quoted on this subject and the coming winter. His quotes have not led to any changes in anything below - OR my final forecast)
Here’s a graph of the correlation between SAI and AO:
That’s pretty impressive. But it’s still not as impressive as what’s next.
The OPI-something new (and foreign):
Not all forecast advances occur in the U.S. Europe has the best overall computer model (The European, or ECMWF). Now an Italian meteorologist named Riccardo Valente and some colleagues have developed an index called the OPI (October Pattern Index). Instead of tracking October snow in Eurasia, this one looks at upper-air patterns.
What has caught the eye of so many forecasters in the Eastern U.S. is the correlation Riccardo has found between the OPI and the overall winter AO. Take a look at this graphic he supplied:
It’s amazing how close those lines are to each other-and even closer in the past 10 years. This work has not been peer reviewed and published yet, so it has some people skeptical. But my interest is always to explore new research and test it myself, published or not. Remember, that’s how I was talking about Eurasian snow in October years before it became common knowledge.
If the OPI is right, the Arctic Oscillation will be clearly positive overall this winter. That doesn’t mean we get zero snow and cold. It means the odds strongly favor LESS snow than our average of 22". All it takes is one big storm during a temporary -- AO pattern to add a lot of inches to our seasonal total.
Meteorology is not an exact, precise science (you probably know that already), and seasonal forecasting is even more difficult. So we use percentages and "play the odds." When both the SAI and OPI suggest odds favor milder, less snowy weather, it makes sense not to ignore it.
El Nino or La Nina? or Neither?
In the past, I never would have waited this long to start talking about El Nino/La Nina. It used to be the first thing I looked at. But not anymore-not unless it’s a pretty strong El Nino or La Nina (the opposite). It’s just become more obvious that our nearby North Atlantic/Arctic have become more important overall.
We can’t ignore the Pacific, of course, but if we have "neutral" conditions, how does that help us? Ocean temperatures in the Tropical Pacific were neutral last winter, and may be again this time. Here’s the latest map of ocean temps compared to normal:
If anything, it’s a weak La Nina (colder than normal). But not far beneath the surface, there’s plenty of unusually warm water:
Computer model forecasts keep conditions at the surface neutral throughout winter, but if some of that warm water gets up to the surface, things change. As we’ve seen in the past, El Nino forecasts are far from perfect.
Other Parts of the Pacific:
Sometimes forecasters get so focused on the Tropical Pacific with El Nino that other areas are ignored. And there are more indices that describe them: PDO and EPO in longer terms, and the PNA in the short-term.
The PNA is pretty simple: a ridge of high pressure in the western U.S. is a +PNA, and it leads to colder and sometimes snowier weather in the East. But the PNA varies a lot during the winter, so it doesn’t do us much good as a seasonal forecast aid.
The PDO (Pacific Decadal Oscillation) covers much of the Pacific between the tropics and the Arctic. A -PDO tends to lead to milder, less snowy winters around here, and a +PDO trends the opposite way. We’ve had a -PDO the past 2 winters, which corresponds with the lack of snow, but it was also negative in 2010-11 when we had 44”. The PDO should be negative overall this winter.
The EPO (Eastern Pacific Oscillation) covers a smaller area in the North Pacific, closer to the U.S. Coast, so maybe that’s why it correlates better with seasonal weather here. A –EPO leads to cold weather tendencies in much of the U.S. (including here) for all 3 winter months. But it’s far from fool-proof: it was clearly negative two winters ago, and it was a very mild winter. That may have been because a very negative AO dominated everything. The EPO should be negative this winter.
Putting it all together:
So, let’s summarize the factors and what they suggest (in general):
SAI - "Mild" winter
OPI - "Mild"
El Nino - Neutral
PDO - "Mild"
Overall, it’s a tendency toward above average temperatures and below average snow. But it’s not that simple.
Here's what I expect:
- Even more extreme temperature fluctuations than normal-very warm, followed by very cold
- Fewer blocking patterns than the past few years. Neither cold nor warm spells will last for weeks
- N&W areas will be closest to average snowfall, while S&E areas will see well below normal snowfall
Philadelphia 14-18” (average is 22”)
Allentown 30-34” (average is 34”)
Atlantic City 10-14” (average is 17”)
Wilmington 12-16” (average is 21”)
December - +1 degrees
January - +2
February - -1
March - +2
What could go wrong? (Or right, depending on how you feel about winter):
- The total October Eurasian snow cover could be more of a factor than the snow increase (would lead to more snow)
- An El Nino develops as that extra warm subsurface water makes it to the surface (more snow)
- The PDO changes to positive (colder)
(If somehow all 3 of those things happen, it could be another one of those big winters. I’ll keep you updated on any changes)