By: Tom Di Liberto
I know it may seem a bit ridiculous to say that winter is over when the eastern half of the United States keeps getting repeatedly smacked by nor’easters. But from a meteorological perspective, winter lasts from December 1 – February 28. And it is over. Which can mean only one thing! It is time for the ENSO Blog’s annual post looking back at how NOAA’s Climate Prediction Center’s Winter Outlook did.
Let’s start with the temperatures
The Winter Outlook, which came out in November 2017 predicted a tilt in the odds towards warmer than average conditions across the southern United States (1) and stretching north along the East Coast. In contrast, the Pacific Northwest and the northern Plains were slightly favored to be colder than average. The Outlook was most confident in above normal winter temperatures over the desert Southwest and Texas, with odds of more than 50%. Meanwhile, in Alaska, the Outlook favored a colder than average winter for southern parts of the state and a warmer than average winter for the northern parts.
In reality, the temperature outlook wasn’t that far off. Winter was much warmer than average across northern, central and western Alaska, the southeast and southwestern parts of the country, while the High Plains, parts of the Pacific Northwest, southeastern Alaska, and southern Texas experienced below normal temperatures. Elsewhere, temperatures were around average.
As you can see, reality rarely is as smooth as the Outlook simply due to the limitations in our forecast abilities. But still, even at a quick glance, this wasn’t a bad forecast. Certainly better in many places than a random guess.
But if there is one lesson I hope you take from my many posts on verification, DON’T TRUST YOUR EYES! Ask for the (scientific verification) numbers!
nutshell (2), the Heidke Skill Score measures how many hits and misses our outlook had and subtracts how often we are likely to have gotten the right answer just by luck. A perfect score would be 100. A score greater than 0 is better than random chance.
Did we beat 0? You bet! This year’s winter outlook scored a Heidke Skill Score of 25 for the continental U.S. if we include all the areas where the forecast said “equal chance,” and a 35 if we only evaluate the places where the outlook got off the fence and favored a specific outcome—above or below-normal (3). And that’s pretty good!
And because we are good, transparent scientists, we’ve put all of CPC’s seasonal forecasts and their verifications online for all to see. Most of our temperature outlooks have performed pretty well (a long-term HSS average of 15 which is greater than 0) going back in time, showing that our forecasters do have skill in this whole seasonal outlook business. We aren’t flawless, but you’re still better off coming to us than asking the groundhog.
What about precipitation?
Precipitation is usually harder to forecast than temperature, as one big storm can have a large impact on the seasonal totals. This leads to a much noisier observed pattern than temperature. This year’s winter outlook forecasted higher chances for a drier-than-normal winter across the southern tier of the United States and higher chances for a wetter-than-normal winter across the northern United States—a generally La Niña-like pattern.
In reality, the western and southeastern regions of the United States observed a drier-than-average winter, while the Missouri and Ohio River Valleys, Great Lakes and western High Plains recorded a wetter than average winter. Not too bad! If we look at the verification number, the outlook scored a HSS of 23 (or a 32 if we only looked at locations with one favored outcome). Remember, a score of 0 would mean we got it right no more often than we’d expect by chance. So, this is a good score for a precipitation outlook when we consider how difficult it can be to predict precipitation.
For a full look at how CPC’s precipitation outlooks have done, please go here. You’ll note that precipitation scores in the last five years are lower and not nearly as consistently positive as temperature scores, reflecting just how hard it is to predict seasonal precipitation totals based on our current tools and techniques. However, on average, both temperature and precipitation have HSS numbers above 0.
So were you guys wrong, or were you right?
Both… and neither! This outlook had more hits than misses (yay!), but there were still misses (boo!). It all sort of depended on where you lived. That is the nature of these winter outlooks.
But in figuring out just how good a winter outlook is overall (or whether to trust it), you cannot just look at the one (or two) times a winter outlook was good (or bad). That’d be like proclaiming your favorite baseball team is the greatest ever because they won on opening day (or vice versa).
It’s important to see how ALL of the outlooks have performed to get the true picture. After all, if in the last 10 years, eight of the forecasts were worse than random chance but two were much better, I’m not sure I would rely on that outlook in the future (4).
What actually happened this winter then? La Nina? Other stuff?
Lots of things. For one (and a reminder that this is the ENSO blog), there has been a La Niña active over the Pacific Ocean. And as we have covered numerous times on the blog, La Niña can influence the winter over the United States by impacting the jet stream and the placement of high and low pressures over the mid-latitudes. But we can’t blame EVERYTHING on La Niña, and no serious scientist does.
Also active during this winter was the Madden Julian Oscillation. Its influence on the United States was already covered extensively by Michelle last month. In short, as the MJO moved around the globe, it led to prolonged periods of colder and warmer than average temperatures across much of the country this winter. Though, while this certainly impacted monthly climate, it likely had less impact on climate patterns averaged over the entire winter.
Anything else, you ask? Well, there was always the randomness of the weather as well as variability in other atmospheric oscillations like the Arctic Oscillation.
As we often say, ENSO is not the only game in town!
Wait, why then do you talk so much about ENSO when making the outlook?
Basically, because it is the one influence that is actually predictable a season ahead of time. We know that other things will impact the winter temperature and precipitation patterns besides ENSO. But we don’t know exactly HOW those other atmospheric features will behave more than a couple weeks in advance.
In contrast, we usually know what phase it will be—El Niño or La Niña—months in advance. The other things mentioned, like the MJO or other patterns of middle to high latitude variability, are much more random, which means they don’t give forecasters nearly as much of a heads up (5).
This leaves forecasters in a tricky spot in searching for climate signals, which is one reason they often turn to “old reliable” ENSO (6), and why reality always looks different and noisier than the seasonal predictions.