Skip to main content

What's in store for the United States this (2014-15) winter?

After a memorably cold winter in the central and eastern United States last year, and some very cold weather this month, folks are likely wondering if this cold weather is a harbinger of things to come.  The simple answer is “not necessarily,” as the persistence of weather and climate from one winter to the next or even one month to the next is usually fairly low (Livezey and Barnston 1988; Barnston and Livezey 1989; Van den Dool 1994).  While “persistence”—the prediction that recent conditions will continue—is a simple forecast to make, it rarely proves to be as accurate as forecasts made using dynamical models or more advanced statistical methods (1).

So does that mean this won’t be a cold winter in the central and eastern part of the nation? Again the answer is “not necessarily.” According to the NOAA Climate Prediction Center (CPC) mid-November outlook, odds favor below-normal temperatures in certain parts of the country, and many of those areas do turn out to be in the south-central and southeastern United States, as we will discuss shortly.  

Map of US temp outlooks for winter 2014-15

Furthermore, and perhaps more importantly, even in regions where above-normal temperatures are favored, a colder-than-normal winter is still a possibility.  Remember, CPC’s outlooks describe probabilities, which means—as we’ve explained in earlier blog posts—that even when one outcome is more likely than another, there is still always a chance that a less favored outcome will occur.

The El Niño-Southern Oscillation (ENSO) provides strong clues as to what we can expect during winter across much of the United States.  Of course, this only applies when El Niño or La Niña are present, and as we approach winter, we find ourselves still waiting and wondering if El Niño is going to begin or not.  However, despite the reluctance of El Niño to show itself so far this year, CPC forecasters have considered potential impacts from El Niño and have slightly tilted the outlook (particularly the precipitation outlook) in that direction.  

And if El Niño remains a no-show this year, what will this mean for the forecast?  Actually, as you might expect, not much, because the forecasters understand the fact that El Niño has a 58% of developing, which also means that there’s a 42% chance that it won’t.  To see how information about El Niño gets incorporated into the forecast, let’s take a look at the precipitation outlook. (El Niño often has a more robust influence on precipitation than on temperature.) 

The winter precipitation outlook favors wetter-than-normal conditions across the southern tier of the nation extending northward along the East Coast, as well as in southern Alaska, and drier-than-normal conditions in central Alaska, parts of the Pacific Northwest and around the Great Lakes and Ohio Valley.  This pattern is quite consistent with the average precipitation patterns seen during previous El Nino winters.  

Map of precipitation outlook for the U.S> for winter 2014-15

However, you’ll note that the largest probabilities on this outlook are all less than 50%.  This means that while above-normal precipitation across the South is the most likely out of the 3 possibilities (below normal, near normal, or above normal), it’s more likely that we’ll see precipitation that is “not above-normal.” That is, the combined chance that the outcome will fall in one of the other two categories (near normal or below normal) is higher. 

It’s like spinning a climate roulette wheel.  While the “above” area is the biggest piece of the pie, the near-normal and even below-normal areas are not insignificant and could occur.  These are very modest probabilities for an El Niño winter and reflect the reality that El Niño is not a sure bet for this winter.  And even if it does develop, it’s likely to be a weak event, resulting in weak impacts. 

For example, in contrast to this year’s ENSO situation, precipitation probabilities in Texas and Florida during the 2009-10 winter outlook exceeded 50% for above-normal rainfall, and they exceeded 70% during the peak of the 1997/98 event.  In both cases, the most likely or favored result occurred, as wetter-than-average winters prevailed. This year our confidence level is not so high, but we still think the probability for above normal is higher than it would be purely due to chance, which would be 33.33%.

The temperature outlook favors a warmer-than-normal winter over Alaska, the Western United States, and northern New England, while below-normal temperatures are favored across much of the south-central (2) and southeastern parts of the nation.  Probabilities of above-normal temperature exceed 50% along the West Coast, so this region has a significantly reduced chance (just 15%, according to the pie chart) of seeing a colder-than-normal winter. 

Also note that both maps include areas where neither above- nor below-normal conditions is favored.  Those areas are shown in white, which represents “equal chances,” and it means that the odds for above, near, or below-normal are all the same (33.33%).  This doesn’t mean that temperature or precipitation is expected to be normal this winter in those regions (the probability for that is also 33.33%), but rather that there’s no tilt in the odds toward any of the three categories.  Thinking back to the roulette wheel, the areas of each region would be the same, so the likelihood of any of the three categories occurring is also the same. 

Making seasonal forecasts is a very challenging endeavor.  Seasonal climate models are not as skillful as weather models, and phenomenon like El Niño or La Niña only provide some hints as to what might occur during an upcoming season.  CPC issues probabilistic seasonal forecasts so users can take risk and opportunities into account when making climate-sensitive decisions.

However, keep in mind that these outlooks will primarily benefit those who play the long game.  The maps show only the most likely outcome where there is greater confidence, but not the only possible outcome.   For example, while the outlook favors above-normal temperatures in northern New England, it wouldn’t be shocking for temperatures this winter to be near-normal or even colder-than-normal. I just wouldn’t bet on it.

Footnotes

(1) However, persistence nonetheless does tend to show positive skill (i.e., it is better than just randomly guessing, or just forecasting near-normal every time), and month-to-month persistence is more likely within seasons with an El Nino or La Nina event (in locations that are influenced by ENSO). 

(2) Note that this means states like Texas, Louisiana, and a few other southeastern states stand a better than average chance of experiencing a repeat of last year’s below-normal temperatures.

Lead reviewer: Anthony Barnston

References

Barnston, A. G., and R. E. Livezey, 1989: An operational multifield analog/antianalog prediction system for United States seasonal temperatures: Part II: Spring, summer, fall, and intermediate 3-month period experiments. J. Climate, 2, 513-533.

Livezey, R. E., and A. G. Barnston, 1988: An operational multifield analog/antianalog prediction system for United States seasonal temperatures: Part I: System design and winter experiments. J. Geophys. Res., 93, 10953-10974. DOI: 10.1029/JD093iD09p10953.

Van den Dool, H. M., 2007: Empirical Methods in Short-Term Climate Prediction. Oxford University Press, 215 pp.

Comments

Permalink
Wow, you're really gonna go that route 33% chance of average, above average, or below average for Minnesota. You guys get paid to do this, we don't we could tell you the same thing that we have that kind of chance for average temps/snow for the winter. Way to make a decision on what this winter is going to do.

Submitted by Keith Pederson on Fri, 11/21/2014 - 12:40

Permalink

I can understand why EC (Equal Chances) is a frustrating forecast.  However, there are times when there is very little model agreement or no clear preceding climate patterns in certain regions of the country.  In those regions, we feel that we should not mislead folks into thinking there is a useful forecast to be made.  If the answer is "we don't know," we should be honest and say we don't know which category is favored.   It is also frustrating for CPC forecasters... they very much would LIKE to see model agrement and a clear climate pattern (like a stronger El Nino) that provides better ability to forecast over many regions of the country.  But there are times when it is not in the cards.  

In the meantime (when we're not making forecasts), many of us develop models and work on applied research projects that we hope will give us a better climate outlooks in the future.  

Permalink
When in doubt, use SWAG. It's better to be completely wrong than to be indecisive. Throw a rooster egg against the barn door and measure the pattern of the yoke spatter. Count the number of fleas on your dog. EC (Equal Chances) is not just a frustrating forecast; it's an improbable statistical anomaly. Stop hiding behind your models and GUESS!

Permalink
I don't think you and others who are being critical of the Equal Chance outlook category really understand what it means. First of all, it does not mean everything will EXACTLY match the averages - there is some room on either side of the (lets say 20 year) average for temperature and precipitation that would fall into the normal range. And don't take this the wrong way - I am not trying to be personal, this is for anyone to consider what might be used to create these maps and why the EC category makes perfect sense (the unshaded area includes areas where conditions are expected to be close to normal). I chose to respond to you because you said they should guess and I strongly disagree with that. If you like, you can come back and argue your point. I think more importantly, the models do not always favor above or below average by a large enough probability to use one of the above or below normal categories - IF THE MODELS PREDICT A35% CHANCE OF BELOW NORMAL TEMPERATURES, THAT'S A LOT CLOSER TO THE EC CATEGORY where above and below normal conditions are equally likely THAN IT IS TO THE 40% PROBABILITY NEEDED TO USE THE LIGHT BLUE SHADING. Instructions for reading the probability maps for monthly and seasonal outlooks can be found here:http://www.cpc.ncep.noaa.gov/products/predictions/long_range/seasonal_info.php . As was already explained, the models do not always provide a result that favors one category over the other. If the most likely outcome is near normal conditions or the models do not provide output favoring a better chance of above or below average. All three categories could have a very near equal chance, but it should be possible for the probability of above average and below average to be 40% and near average 20% if these models produce the kind of output I suspect and this would still fall in the EC category. You say to GUESS!, but I do not think that is a good move at all - if the data does not provide enough useful information to make a choice that is more likely to be correct, guessing would be lying to everyone who assumes the maps are based on data and models, not guesses and perhaps deviations from normal conditions are less likely to be large when the models are in too much disagreement. In any case, it would help nobody for them to just guess. It could be that these maps are based entirely on computer models and maybe mathematical formulas and the people making the forecasts do not have any input into what the maps look like, they just enter data recordings or those are collected and processed automatically by computers. It could all just be the output of combined models either with no alterations by forecasters or only changing things by removing data/influences from significantly outlying models - if most models produce results within a certain range but maybe a couple give results far outside that range, it may improve the forecast if those outliers are removed. If there is no significant indication favoring above or below normal conditions, it would be utter nonsense to just guess. You say it is an improbable statistical anomaly. This is only true if you are talking an exactly equal chance (the EC and normal category covers a range of percentages centered on 33%, not just that exact value) and/or you consider normal conditions to be exactly what the 30 year or whatever average is - lets say it is 41deg, instead of a range such as 40-42 or 39-43. They are not saying the probability is exactly equal, but that the chance of above and below average conditions are both less than 40% or maybe the different models are in too much disagreement about what is happening to make a prediction. I'm wondering how much you know about statistics and these maps when you say EC is an improbable statistical anomaly. I do not have the information to run any kind of statistical test to find out if your assertion is false, but it just doesn't sound right to me and I see it differently and feel I understand very well what these maps mean. I could be wrong about ways EC can be assigned but at least I put some thought into it and also realize sometimes there is a better or equal chance of near normal conditions as opposed to conditions being toward one extreme or the other. These maps make perfect sense to me and I feel I understand why the EC category is used and how models can be in such disagreement as to make the output not very useful or even useless, I have looked at the output of many computer models for the same tropical systems and the model forecasts were all over the place with no tracks very close together and wide variation in forecast intensity.I do not know what goes into making these outlook maps and do not know exactly when EC is used (for instance, maybe it would be rounded to the nearest category% - if models show 38% above average maybe it would fall in 40% above instead of EC, where it would be in the 33% probability shade). I do not think I have explained my thoughts very well and I have assumed certain things could go toward assigning an EC that may not contribute to its use at all. I think there is often a better chance of near average conditions which would warrant the use of the EC category over the others and that simple explanation takes care of at least part of the map - maybe there is a 40% of near normal and 20% each above/below in some areas.

Permalink
Reg; ' EC (Equal Chances)' A perfectly reasonable and responsible position to take. The magnitude possible permutations of weather, make daily forecasts tough enough, but a long range forecast without exceptional indicators is nearly impossible. The NOAA and it's staff do a great service for this country and I'am appreciative of the work you all do.

Permalink
The issue is that the article is titled "What is in store for the United States this winter", and it doesn't answer the question (even in a glib, playful way). Of course, it's a better title than "NOAA has no idea what is in store for the U.S. this winter". EC is, after all, not even a forecast. It's an admission. It shows that our models and knowledge are still very rudimentary relative to the system that they seek to characterize. Fortunately, you and the FA agree on what's in store for northern New England. Relatively warm, but pretty darn cold.

Permalink
You live in Minnesota, I live in Wisconsin, winter is cold and snowy-EVERY YEAR! The "percentage" of cold and snow is 100%. Why would you berate NOAA for it's data and variances, or breaking climate down into percentages? You're right you don't get paid to do this, and it shows.

Permalink
Hi Mike - How has CPC factored in "Lake Effect" precipitation for their percentage outcomes in the great lakes region? Or is it not possible to model such events. Regards, Emmett

Submitted by Emmett on Fri, 11/21/2014 - 15:43

Permalink

Lake effect snow generally effects fairly small regions and due to it's unpredictable nature, doesn't play a role in our seasonal outlooks. Obviously the very impressive snowfall totals we've just seen along the shores of Lake Erie can effect seasonal snowfall totals, but these extreme types of events are not predictable on seasonal time scales.  The average lake effect snow totals are a part of the long term climatology, but there's no way to determine whether a particular season will see less than or greater than lake effect snows.   

Permalink
NOAA's winter forecast is just about the inverse of what most other major weather forecasters (Accuweather, Weather Channel, etc) are calling for. Seems to me NOAA was way off last winter (without stopping to try and look up the maps). Why is NOAA so different in methodology?

Submitted by Bethany Rais on Sat, 11/22/2014 - 14:35

Permalink

This is a great question.  I don't know the methods of other companies (sometimes this is not provided to the public), but I do know of one difference that often shows up in presentation/format.  You'll note that CPC outlooks are presented in a probabilisitic format.  That is, CPC will tell you the percent chance of three different outcomes (the maps show the color of the outcome that earns the highest percent but there are chances assigned to the other 2 categories as well).  There are 3 possible outcomes:  above, near, or below average (temperature or precipitation).  For example, instead of just telling you that there WILL be below average temperatures in a certain part of the country, CPC tells you that there may be a 40% chance of below average temperatures.   The reason we do this is because there is large range of possible outcomes and many different prediction models or methods will give you different answers.  We feel that we should give you information that is up-front with this uncertainty so that you can make the decision that is best for you.   For some people, they will make a decision if there is a 30% chance of below-average temperatures.  Others might not make a decision until there is a 70% chance of below average temperatures.   

Check out the current North American Multi Model Ensemble model temperature predictions for DJF 2014-15:  http://www.cpc.ncep.noaa.gov/products/NMME/current/ustmp2m_Seas1.html  .  There are a lof of different combinations!  Some models predict below-average temperatures over much of the country, but many other models predict above-average temperatures.   And there is generally no "one best model."  Some models perform better in certain years than other models, so we don't necessarily know which one is going to be best for a certain year.  So, we often take all this information (including more than what you see on this webpage) and we build in this uncertainy into our predictions by presenting them as "a percent chance." 

Therefore, this year, as you look at our forecast maps, keep in mind that there is a chance for below-average temperatures over the entire country.  However, it is just higher in certain areas (the Southeastern U.S.) than say over the West Coast, where the chance of below average temperatures is less than the chance of above average temperatures.  

Also, in addition to being clear about the uncertainty in our format, CPC also provide past assessments of our forecasts showing objective scores that tell you how well the forecasts are doing.  We recommend you look through this webpage to see these:

http://www.cpc.ncep.noaa.gov/products/verification/summary/

http://www.vwt.ncep.noaa.gov

Submitted by michelle.lheureux on Sat, 11/22/2014 - 16:05

Permalink

No one says seasonal climate prediction is easy, but our aim is to give the best outlook possible with the best science available.  We certainly would not discourage you from consulting other sources, but would recommend that you seek verification on their forecasts.  We hope to provide a blog post in the fututure that provides gudiance on how to grade our climate outlooks (which can also be found in the verification links in the reply above).  

Permalink
I read an article last weekend I believe it was on CNN where it talked about how there was a conference in Southern California with different Meteorologists from at least around the state and the big comment coming out of that conference was El Nino confidence was only at 10% and they believe it will be a fourth straight dry winter on the west coast from their models. I guess the main question is why the big difference between different groups when I thought they all saw the same information.

Submitted by Ryan Souza on Mon, 11/24/2014 - 15:11

Permalink

This is a interesting point and I believe touches on a core point.  Individuals can have the same information at their fingertips and come up with different interpretations of the data.  To ameliorate this at CPC we often employ "consensus based" approaches to developing outlooks meaning that we do not rely on a single forecaster but a group based decision process.  I'm not familiar with the specific conference you are discussing, but it is possible they were discussing how even If California gets a normal or above normal rainfall winter this year, that would still not replinish the large deficit and drought could continue in many parts of the state.  

Submitted by michelle.lheureux on Sun, 11/30/2014 - 10:16

Permalink
I have to agree with Keith. Your long range predictions are frustrating. For months you've been forecasting that this winter will be especially warm (and I believe last year's long range forecast was also completely wrong). Now that the weather is the 'polar' opposite, you're changing your tune. I read that your requested budget for 2015 is 5. 5 billion dollars. How is that you need 5. 5 billion dollars to do what the Farmer's Almanac has gotten right for the past two years using nothing more than moon cycles and divination wands? Even the Weather Channel beat you guys to the correct forecast for these blustery, below average, cold days.

Submitted by Jeff on Tue, 11/25/2014 - 11:39

Permalink

The winter outlook is the period from December-February, so despite the cold weather as of late, we still haven't yet entered into the period we are forecasting.  As for CPC's verification statistics, we recommend you check out these webpages so you can see how good (or bad) the outlooks are:

http://www.cpc.ncep.noaa.gov/products/verification/summary/

http://www.vwt.ncep.noaa.gov

While we don't use moon cycles and wands because they are not discussed in the peer reviewed climate literature, we would not want to dissuade those who would prefer using these methods for prediction.  

Submitted by michelle.lheureux on Sun, 11/30/2014 - 10:24

Permalink
The Farmer's Almanac is very often wrong. At best, it is a stopped clock (a stopped clock is completely accurate only twice per day!) that you are cherry picking from. As for the NOAA's budget... if you actually look at what the NOAA does as an organization... it isn't just making wild predictions about how cold the winter will be.

Permalink

Aside from NOAA's budget, it is true that NOAA's climate forecasts factor in all of the known influences on seasonal climate fluctuations, all supported by the historical observations. They are not made using dice or roulette wheels, or by licking one's finger and holding it up to the wind. Science is used wherever possible, and when there are no clues, the "equal chances" forecast is given, with no tilt of the odds in any direction. The smallness of the usual tilts of the odds reveals in an honest way the fairly large uncertainty of the forecasts, even when an ENSO event is in progress.

Permalink
This isn't a forecast but more of an overall guidance. NOAA historically does not make a weather prediction but more of a percentage of what possibly could happen. They do not state things that they think will happen. It is literally the safest way to make a forecast by not forecasting a thing but just saying there maybe a chance for this in the huge section of the nation. I would like NOAA to go back and look at their so called percentage forecast last year and then look at how well that worked out for you. If I remember correctly you completely blew the entire percentage forecast bad. I don't think you could have been more wrong than what you ended up being. So now looking at this year with a good analog package with a weak Nino 3-4 and years such as 76-77 and 93-94 this years so called forecast above will be wrong. How was your November forecast??? I am pretty sure you thought it was going to be warm and it was record breaking.

Submitted by Stacey on Wed, 11/26/2014 - 06:43

Permalink

Today is the last day of November, so in a couple days we will have the data to see how good CPC's November 2014 outlook was.  Later on, check out the verification pages that I've linked in the comments to evaluate the quality of all of CPC's past outlooks.  You are correct that in providing a percentage chance we are not ruling out any outcome, but at the same time we are providing the user information on how much we prefer in a certain outcome (e.g. a 55% chance of above average temperatures).  We believe that given the large uncertainty in seasonal prediction, this is a more honest approach than telling the user "it will be warm."  However, if the user would prefer to have their climate outlooks in degrees Fahrenheit, one can use the mostly likely or median prediction (the 50% tile) to obtain this information, but keep in mind that there is an equal chance the observations will occur below or above this value. We provide these maps here (for preciptiation and other seasons see the side bar):

http://www.cpc.ncep.noaa.gov/products/predictions/long_range/poe_index…

Permalink
Greetings. I have enjoyed reading this page! The older people who live around me (Missouri) told me this past (late) summer that the upcoming autumn and winter seasons would most likely be cold and long because of what they perceived in the seed of the common persimmon. So far they have been right on with their assessment. What do you think about folk ways of predicting the weather?

Submitted by david on Thu, 11/27/2014 - 20:25

Permalink

Thank you very much.  I personally find these methods very interesting!  However, what stops me from weighting them is that I don't understand the physical connection between persimmon (or another method) and the upcoming winter.  I'm guessing, but it would seem to imply that there is a preceding climate pattern that causes the seed to take on certain attributes (i.e. a mild autumn).  More directly, we can can just look at the climate pattern and see if that helps us to predict the winter.  What we find is that persistence forecasts (using the past to predict the future) are not as good (less skillful) than today's state-of-the-art statistical and dynamical models.  Also, keep in mind that when CPC states "winter outlook" we are discussing the period from December-February, so we have yet to see how things turn out.  

Submitted by michelle.lheureux on Sun, 11/30/2014 - 10:31

Permalink
Well I for one appreciate the guidance, which is all it is, same thing with economic forecasts. Its a guidance and weather and micro climates are all over the place, rarely is the temp/precip avg, its either above or below the line, you can't forecast short cycles of extreme weather which is really what our weather is. Look at where you live, the history of the weather, what local weather offices say and go from there. Doesn't mean we won't get some strong blocking high that will throw all the forecasts out the window.

Submitted by Eric L on Mon, 12/01/2014 - 09:56

Permalink
Nice summary and informative responses. One question: Why do the CPC outlook forecast maps stop at the Canadian border, yet it appears the various model runs that feed into the forecast are not so restrictive ?

Submitted by Pete H on Mon, 12/01/2014 - 19:01

Permalink

Great question Pete. Obviously, our seasonal forecast of impacts won't just suddenly stop at the Canadian or Mexican border in real life and the models we use are global in nature. However, the mission of the Climate Prediction Center is to provide forecasts for the United States and thus our forecasts follow the borders of the US. 

If you are curious as to seasonal forecasts for Canada, I urge you to check out Environment Canada's (our counterparts to the north) forecasts located here.

Submitted by tom.diliberto on Wed, 12/03/2014 - 08:55

Permalink
hey, CPC folks, i wouldn't worry about the complainers too much . . . they seem like the types who think rocket science is hard. what you are trying to do makes rocket science look like knitting! keep up the good work . . . and thanks.

Submitted by willi0000000 on Wed, 12/03/2014 - 13:21

Permalink
I totally agree with will0000000 above - and would like to commend you also for your work, and for your explanations on the graphs which are well-written and make them understandable to us non-rocket science types :-) ! I also want to commend you for your civility with your answers to those who might not have been civil in their comments - and for using those as an educational opportunity as well. NOAA and Climate Prediction Center are my go-tos - in addition to prayers for a warm winter in the Northeast! Thanks again!

Submitted by Ellen on Thu, 12/04/2014 - 18:36

Permalink
Wow, I am amazed by the hostility. I am a retired USAF pilot and have depended on NOAA and their predictions for years. Predictions are just that, predictions. It's not an absolute definitive answer and if you think it should be, I recommend you educate yourself about the science of meteorology as well as the mathematics of statistics and probability and what an EC answer truly is. If you hate the snow and related conditions of living in the great lakes area, move! I am a southwest native and live in Michigan temporarily (another 4 years) and trust me, no one hates the snow more than I do. I prefer the deployments I had to the middle east over measurable snow but the other three seasons here are beautiful. NOAA does a great job and I will continue to use the information they provide.

Submitted by Nitrobunny on Sun, 12/07/2014 - 12:19

Add new comment

About text formats