One of the things I miss least about doing TV weather is the pressure to come up with a winter outlook. When I was still on TV this past autumn, I didn’t have much input into a weeknight presentation, and I was glad of that.
If you’ll take the time to read what I wrote for The Buffalo News on Nov. 1, you’ll get the gist of what makes such outlooks a much riskier venture than forecasting out to seven days. (Hey, I just re-read that article, and it’s not bad.)
Call it meteorological cowardice on my part, but seasonal outlooks for the winter do not have very good verification scores. Take, for example, this temperature probability outlook from the National Weather Service Climate Prediction Center, issued on Oct. 18.
Here is how temperatures have been working out from Dec. 1 through Feb. 26, courtesy of the University of Illinois Midwest Regional Climate Center. Green represents temperatures running below the seasonal average, and brown represents above average.
This should not be interpreted as a putdown of the CPC. They have a mandate to issue a winter outlook in mid-October, even when confidence is low, because it’s so early in the season. A private forecasting company that has to remain unnamed did a little better, but not much. I haven’t done an exhaustive search of all winter outlooks, but in what limited sampling I took, I haven’t found any that would win a gold star.
My curmudgeonly attitude toward winter outlooks has worsened in the last 10 to 15 years. The more we have learned about variables that can affect weather patterns over months, the more we have learned about the uncertainties in such outlooks.
And before any warming climate denialists jump on these uncertainties, the time scale I’m referring to isn’t really climate prediction, despite CPC’s naming. It’s weather. It’s weather patterns over a three-month period, not climate over decades. The different time scales are critical, since climate prediction involves cancelling out the week-to-week or even month-to-month “noise” of short-term weather patterns. The modeling is markedly different between weather and climate, and it's not comparable.
In the early 1980s, when the El Niño Southern Oscillation (ENSO) began to get more attention in the scientific literature, there was some hope ENSO might be the most important variable to making more successful seasonal outlooks, for both temperatures and precipitation. Hopes rose during the winter of 1982-83, when a strong El Niño was predicted and verified. It turns out a strong El Niño, if well predicted, is the most useful tool we have. It greatly increases outlook confidence for temperatures and somewhat increases confidence for precipitation. The typical weather impacts of a well defined El Niño are described in this short NOAA article.
The problem is, few El Niños truly are very strong. When a weak El Niño or a weak La Niña or a neutral ENSO are in evidence, the effects become much more of a crapshoot (pardon the technical jargon). Only three “very strong” El Niños have occurred in recent decades: 1982-83, 1997-98, and the most recent and extraordinary episode, 2015-16.
Since very strong El Niños can be predicted some months in advance, confidence for those infrequent winters’ temperatures running warmer than average in the northern United States and in our region increases considerably. With more Pacific and less polar air crossing the Great Lakes, we generally experience less lake-effect snow.
In the last super El Niño, even as winter ran out the clock, March 2016 was 5.9 degrees above average and snowfall for Buffalo totaled 0.9 inches, with 48 inches for the season at the end of the month. The winter outlook the previous autumn was more reliable than usual.
With weak ENSOs, we are more at the mercy of many other oscillations. Just running off a few of them, there is the Arctic oscillation (AO), the North Atlantic Oscillation (NAO), the Eastern Pacific Oscillation (EPO), and the less understood Madden Julian Oscillation (MJO).
While ENSO phases are often seen coming months in advance, time scale for predicting the phases of the MJO, NAO and AO are usually just a few weeks.
Moreover, the very important MJO has eight different phases that are recognized, and each of its eight phases have different impacts in North America during the four seasons. The MJO phases have sometimes wiped out the typical impact of, say, a strong La Niña. But there are no magic models to put all these variables — pieces of the puzzle — together in a nice, neat fit very often. And a mean warming climate is always lurking in the background, with its growing effects on global weather.
We do have models generated by several nations, including the United States, that are doing a better and better job of giving us the chance to look out several weeks in advance for colder or warmer patterns. Thanks to tools like those, I was able to write this in a Dec. 19 article: “My best estimate is an eventually weakened polar vortex will bring some true, below-average cold to our region and much of the east, but the biggest transition will hold off until mid-January.” That model-driven prediction actually worked out very well. Ten years earlier, I wouldn’t have touched going out that far in advance.
On the other hand, lest someone of my ilk grow overconfident on looking out a month in advance, Dr. Michael Ventrice of IBM’s the Weather Company noted on Feb. 28 that models such as the European had been predicting at the end of January warm temperatures for the western United States and cold in the east in February.
To quote Michael, “Whoops.” This is how temperatures actually worked out, compared to average.