Hello and welcome to the 2018-2019 Southern Connecticut Weather Winter Forecast!
Each fall, we produce a winter forecast, based off of a number of factors that influence the larger-scale pattern and sensible winter weather in our area. We’ll identify some of the key variables that go into the outcome of a winter, devise a weighting scheme to build a final forecast, analyze each of the variables and what it suggests for the upcoming winter, and finally, tie it all together into a seasonal forecast. We focus primarily on meteorological winter, that is, December through February, but we’ll touch on March where appropriate. Fair warning; this is a long discussion. If you’re just looking for the forecast, scroll down to the end of the piece, but I highly encourage you to read through the discussion and analysis, as it will give you a much better understanding of why the forecast is what it is.
Before we get into the forecast, a word of caution. To quote our last winter forecast:
“Long-range outlooks are an extremely imperfect science, and, as with any forecast, are not the rule of thumb. To mitigate this somewhat, we try to work in probabilities and ranges rather than absolutes, to give a more reasonable boundary of what may happen, but things can and do change in meteorology on a dime, and, as a result, any long-range outlook, including this one, should be taken with a grain of salt, and seen as more informational rather than predictive. Nonetheless, we do strive to be as accurate as possible, and the science behind these methods is real, if a bit shaky at times. We'll keep an eye on how we're doing throughout the winter, and will likely have periodic updates to confirm or refine our thinking from this initial outlook.”
Essentially, the variables that make up a winter forecast are numerous, and if even one or two of them don’t turn out as expected, it can result in a significant shift in the overall paradigm of ground truth. So take it with a grain of salt.
With that out of the way, let’s take a look at the variables that make up the winter forecast. There are five main elements that can contribute to our seasonal forecast. We present them below, along with their approximate weights towards the final forecast.
- Seasonal models (10%): These models are longer-range versions of the global models that we use for short-range forecasting. Here, we look at their composite averages. They have some skill, but have several distinct biases, and as such we weight them relatively little.
- Analogs (25%): We look at past years that have similarities to this year, whether it be conditions observed during the fall, ENSO state, or teleconnectors. Essentially we are looking at how past hemispheric patterns may closely resemble our winter.
- ENSO (25%): El Niño Southern Oscillation, otherwise known as ENSO, represents water temperatures in the central and eastern equatorial Pacific. El Niño represents warmer than normal temperatures, while La Nina represents cooler than normal water temperatures. These phases have distinct impacts upon the larger-scale pattern, and as a result on our sensible weather.
- Teleconnectors (25%): There are many indices measuring current conditions in different parts of the world, from upper level features to regional ocean temperatures. We will discuss them, analyze them, and explain what their current and forecast states mean for our weather.
- Current conditions (10%): Current surface conditions such as Siberian snow cover and surface temperatures, along with the current pattern, can influence our expected winter patterns.
This year, almost the entire forecasting team has written part of this forecast and so each section is signed individually by its author. That said, we work as a team, and have all shared our ideas with eachother and agreed on the outcomes. Our end goal is to produce nine forecasts; temperature and precipitation departures from normal for the winter as a whole as well as monthly for December, January and February as well as a snowfall departure for the winter as a whole. Let’s get to it!
Let’s start by taking a look at the seasonal models. There are four of them; the Euro seasonal, the JAMSTEC, the CanSIPS, and the CFS. Unfortunately, due to copyright issues, the Euro seasonal can’t be reproduced, and I have not seen the actual images myself. So, we will focus on the latter two.
A note on how the seasonal models work. Essentially, they attempt to model the upcoming months in much the same way that the midrange global models attempt to model the upcoming forecast period, that is, start with a set of initial observations and apply some physics package to those observations to make future forecasts for the next period on some interval, then, that forecast is fed back into the model as the initial observations for the next period to be created off of, and so on.
As you can imagine, small inaccuracies in initial data or calculations are exponentially magnified as the recursive cycle continues, and as a result, global models, especially the seasonal ones, are notoriously inaccurate, hence our relatively low weight of them compared to other factors. They are still worth considering and looking at, as what they show (and don’t show) can be valuable when taken in the right context.
First, let’s take a look at the CFS. Here are temperature and precipitation departures from normal on the CFS for December, January, and February, courtesy of Weatherbell Analytics.
Looking at some bigger picture patterns, the CFS and CANSIPS generally follow the climo standard for an El Niño regime where we see a cooler than normal southern tier and a warmer than normal northern tier in the west. Where they differ is how they treat the northeast; the CFS keeps generally colder than normal temps in our area through more frequent home-grown Canadian air masses and Quebec high pressures (like what we’re seeing right now) that allow us to stay cooler than normal relative to the rest of the country, while the CanSIPS allows for warmer pacific air to invade at times when the pattern relaxes. The national precipitation forecasts for those two also generally follow climo, although the CFS featuring a dry December and January is unusual given the propensity for nino’s to be wet here.
The JAMSTEC, on the other hand, paints a very different picture where the boundary between cold and warm is pushed much further north (into Canada), allowing for a nationwide below normal winter. While that solution would be very nice for snow lovers, I don’t think it’s realistic compared to El Niño climo and what we’ve seen so far this fall.
All in all, it’s hard to get a clear signal from the seasonal models. They all offer plausible solutions (discounting the extremity of the JAMSTEC a bit) and I don’t see any real red flags from any of them. Averaging them out argues for a near normal winter here, and so that’s what we’ll go with for this portion of the forecast.
Overall Summary of the Seasonal Models: Near Normal Temperatures and Precipitation. –SA
First, I’ll start by telling you my analog methodology for this year. The current ENSO state is a weak El Niño. So when I gathered upper air analogs for the progression of autumn, I considered weak to moderate El Niño years, “La Nada” years, and only the very weakest La Nina years. All other La Nina’s were automatically eliminated from consideration, as were any strong El Niño’s, since those ENSO states are nothing like what we have this year. Upper air matches were given three “points” for the study. Good ENSO matches were then given two points and a good QBO match was given one point. The years were then cross-weighted to find commonality between years. Finally, points were summed up to calculate strong analogs, moderate analogs, weak analogs, and “throw away” years. The “throw away” years will obviously not be mentioned in this forecast.
After doing all this, the following years came out to be my strongest analogs.
2002-03- this year is being widely talked about in the meteorological community, and is the preferred analog of most meteorologists I have spoken with about the upcoming winter. It was a weak to moderate El Niño, the upper air pattern has been a great match, and the fall to this point has progressed very much like the fall of 2002. 2002 also had a hot summer. There were, of course, differences. The summer of 2002 was dry over much of the east, while this year was wet. I am not sure what impact, if any that would have on the future of the pattern evolution. The other question I would have is whether or not this El Niño event will be as strong as 2002-03. I do not want to confuse people and make them think that 2002-03 was a strong El Niño—it was not. But by this point in 2002, the El Niño was stronger than what we have now. This does not mean that it cannot change, however. This analog will be weighted 3x.
1959-60: Very weak El Niño, decent match in the upper air pattern, similar pattern evolution to this point. This could be a very good analog, especially if the El Niño remains weak. Another year that fits the same logic would be 2003-04; another good evolution in the pattern and upper air, and a very weak El Niño. Those two years could be very good analogs if the El Niño doesn’t strengthen from here on out. The last “strong” analog that showed up and kept showing up was 1986-87. The strength of that El Niño was somewhere in between the last two mentioned years and 2002-03, and could be where we’re headed this year. It also had a very similar pattern evolution and was a good QBO match. We’ll also weight this analog 3x.
We have two “second tier” analogs this year. The first “second tier” analog is 2014-15. That ENSO was very similar to that of 1986-87 and could be an excellent match for this year. So from that standpoint, we have a very good analog. However, the pattern that summer was very different from this one, so I struck it down to a medium level analog. The other second tier analog is 1980-81. Of all the upper air years, this is probably the best upper air match. However, it’s a bad ENSO match, since that year was a weak La Nina, and we are not headed that way this year. But since I decided to weight upper air more than ENSO and it was not a moderate or strong Nino, I labeled that year as a medium level analog. If the ENSO strengthens, this year would then become weaker. We’ll weight these at 2x each.
Finally, we have two weak analogs. They are 1976-77 and 2009-10. 1976-77 is a good ENSO match, as its ENSO numbers are very close to what I expect this year- a weak El Niño (unless it strengthens more). However, it’s a bad upper air match, so I only gave it small consideration. Finally, 2009-10 is a good QBO match, but it is a poor upper air match. It was an El Niño, but it was a moderate to strong El Niño, and I don’t expect this El Niño to get nearly that strong. We’ll weight these at 1x each.
Now some notes on the QBO. The QBO is the Quasi Biennial Oscillation. This is a relatively new player in the field, so there isn’t a whole lot of understanding on how it works or what its large-scale effects are on the atmosphere. Most tend to believe that a negative QBO correlates to more high-latitude blocking up north, which would then produce more cold and snow here. However, there have been winters with a strongly positive QBO that were cold and snowy here. Conversely, there have also been winters that had a strongly negative QBO that were warm and fairly snowless here. So it certainly is not a one-to-one correlation. Because our QBO index is weak this year (as in not strongly one way or the other), and also because the understanding of the QBO’s exact impacts on our weather is still in its infancy, I did not weight in heavily in my forecast. If a year was a good QBO match but did not qualify in any other area, it was not included. Additionally, we’re in somewhat uncharted territory this year regarding QBO. It’s weak to moderate negative at the moment, but rising rapidly. There really aren’t any great QBO matches in the sample. The best matches are 2009-10 and 1986-87, which interestingly enough, were also El Niño years.
When visualizing these analogs for temperature and precipitation departures on a monthly basis, the composites look like this.
All in all, the analog package for the season suggests below normal temperatures and near normal precipitation. You can see the seasonal composites (December/January/February) below.
The El Niño – Southern Oscillation is an inter-annual variability across the Pacific Ocean that impacts global weather. We normally call the two extreme phases of the oceanic portion of the oscillation “El Niño” and “La Nina”, which are two terms I’m sure everyone has at least heard once before. ENSO is an atmospheric-oceanic coupled system, which means its variability depends on both the workings of the atmosphere and the ocean. The typical range of this oscillation is anywhere between 2-7 years, and the strength can vary.
There are three phases to this oscillation: normal conditions, La Nina, and El Niño. During normal years, the Pacific Ocean is characterized by very warm waters over the western Pacific (Australia, Indonesia, etc). On the eastern side of the Pacific, cool water is upwelled off the South American coast and Central America. This setup would tend to support warm, rising air over the western Pacific and cool, sinking air over the eastern pacific. Rising air is associated with low pressure and convergence, and cool sinking air would be associated with high pressure. High pressure tends to flow to low pressure, and the resulting flow would be an easterly flow from the Eastern pacific to the Western pacific. Meteorologists call this wind the trade winds, which is a component of the Walker Cell.
Finally, the model consensus takes into account 18 dynamical models and 8 statistical models and plots the 26 models against each other, and the data strongly represents an El Niño event average at a 0.6 degrees above the average sea surface temperature. With the current oceanic observations and statistical model output, it is clear that we will have a weak El Niño year. Therefore, we can expect a sporadically active Pacific Jet, which will occasionally drive above average temperatures and shortwave patterns.
Overall Summary of ENSO: Slightly above average temperatures and above average precipitation. –LD
During the wintertime, teleconnections are key to understanding the synoptic, or broader scale pattern. Teleconnections are critical to my two-week outlooks during the winter. Overall, these indices measure the current atmospheric conditions in certain regions of the globe (for our purposes, the northern hemisphere). Often, they measure the pressure in a region, which is shown on a 500mb map as ridges and troughs and blocks.
The major teleconnections that we look at can be divided by the ocean basin that they dominate. For us, that means that we usually divide the teleconnection pattern into Atlantic and Pacific (PAC) patterns. We focus on the four “biggies”, two in each basin. In the Atlantic, we look at the AO and NAO. In the Pacific, we look at the PNA and EPO.
Arctic Oscillation (AO)
First up is the Arctic Oscillation, or AO. This index measures pressure anomalies in the Arctic, and correlates with temperatures for much of North America during the winter months. A positive phase of the AO (+AO) represents below average heights (think troughing or low pressure). A negative phase of the AO (-AO) represents higher than average heights (think ridging or high pressure). In a negative phase of the AO, the Polar Vortex, which exists year round, is more disturbed and weaker, resulting in conditions that make it more likely for Arctic air to surge southward over the eastern US. Another result of this pattern is a more southerly storm track. This is the opposite when the AO is in a positive phase, as the Polar Vortex is stronger and undisturbed, and less likely to be displaced, along with its much colder air. For us, the AO is an important indicator of potential temperatures and the availability of cold air during precipitation events.
Here's a look at what a positive AO looks like graphically. This map is from 1993 and is not representative of our current conditions. With below average heights in the Arctic, the Polar Vortex is consolidated and undisturbed, leaving higher heights over the eastern US. Looking at the Hartford temperature records, 8 of the 14 days shown below were above normal, with two days with departures of +22 degrees!
The second important teleconnection in the Atlantic is the NAO, or North Atlantic Oscillation. It is a close relative of the AO, but is different in a few ways. For this teleconnection, forecasters examine two areas in the Atlantic. One is an area of low pressure near Iceland, and the other is an area of high pressure near the Azores. For the winter however, we look closely at the heights around Iceland. The strength and placement of these features has a significant impact on the placement of the jet stream, a critical feature in the delivery of cold and precipitation through developing areas of low pressure.
During a negative phase (-NAO), higher heights are seen over the Icelandic region, translating (usually) into lower heights over the eastern US. This generally means colder than normal conditions in the eastern US and a jet stream that is in a favorable condition for storminess. The opposite is true with a positive phase (+NAO) of the NAO. It is important to remember that a –NAO is not critical for a colder than normal or snowier than normal pattern. Other teleconnections/pattern orientations can overwhelm a +NAO to produce more wintry sensible weather. However, the NAO is critical to the development of our biggest winter weather events. This is because a –NAO, especially if it is west based, often implies more blocking, which “blocks” storms from being too fast a mover and enhancing the potential for the phasing of different aspects of the jet stream, most notably the northern stream which supplies the strong cold air mass and the southern stream which supplies copious precipitation.
Not all NAO phases are created equally. The strength and position of the Atlantic pattern has a significant impact on potential. While the AO is an important indicator of potential temperatures and the availability of cold air during precipitation events, the NAO is an important indicator of storm track and the ability to produce a blockbuster event.
Here's a historical look at a negative NAO from 1966. The period below produced a majority of days with below normal temperatures, and two 6” events in Hartford, pushing the monthly average above normal.
Pacific/North American Pattern (PNA)
Here at SCW, while we believe in the power of having the Atlantic on your side during winter, PAC is King. The Pacific pattern is critical to understanding both temperature and storm potential. The PNA is one of the most important teleconnections, with significant influence over North America. A positive phase of the PNA is seen when there are higher heights over the eastern Pacific and western US. This means there is ridging over the western part of the continent that translates into a trough delivering colder air over the eastern US. The opposite is true in a negative phase. The strength and orientation of the PNA helps determine temperature and storm track potential.
Perhaps saving the best for last, the last major teleconnection we look at is the East Pacific Oscillation, or EPO. The last few years, it may not be hyperbole to say that the strength of the EPO has saved a few of our winters (for winter lovers) in the face of a hostile Atlantic pattern. The EPO is fundamental in building and delivering cold air to the continental US or torching the continent. For this pattern/teleconnection, we examine heights over the northeast Pacific and Alaska. Higher heights over the region translate into troughing in the east, both work in tandem to deliver an ample supply of cold and sometimes frigid airmasses.
The strongest and best oriented –EPO phases can facilitate what we call “cross-polar flow”. If you like big cold, that’s the flag for something big. This is the extreme cold that comes straight from the pole without much moderation. If you hate cold and snow, a big positive phase of the EPO virtually does the opposite. It cuts off the cold supply not just by changing the flow of cold air, it torches the source region for cold to begin with, by flooding the continent with relatively warm Pacific air. If you have an awful PAC pattern, you might as well close the shades on any type of normal winter.
Below is a look at the positive and negative phases of the EPO and their effects on our sensible weather.
This year, the ENSO state helps inform a lot of our thinking for the season ahead. Starting with the Pacific, we’ve generally seen a pattern that has been favorable toward delivery of colder than normal and stormy conditions. This is expected to relax a bit in a “reshuffle” in mid-December, with both the PNA and EPO becoming more hostile toward cold air delivery, but we expect this to be transient, and for the PAC pattern to become favorable again for cold air delivery and active storm threats (not necessarily all snow) in later December and generally through the rest of meteorological winter.
As for the Atlantic, we are more confident that the lack of “help” we’ve received over the last few seasons overall, particularly when it comes to the NAO, will change. The AO looks to be in a mostly negative phase during the winter months. One of the factors in believing this is the fact that we’re heading into the colder season and starting to see a strong and relatively persistent –AO, which favors continued pushes of cold air as we see the atmospheric response to ENSO conditions continue.
Here’s a look at the state of Siberian and North American snow cover so far this season. We can see that at the end of October (the period measured by the SAI index), the extent of Siberian coverage was below normal, but by the first few days of November coverage rebounded to towards the middle of the pack, and has since remained there. I find it unlikely that the index is so connected to the calendar that a shift of a day or two is enough to make a notable difference in the performance of the index. Across North America, snow cover is solidly above normal.
Snow cover in Siberia has a much more questionable correlation by way of the SAI. The SAI is an index developed by Dr. Judah Cohen that claims that the extent of snow cover in October in Siberia is correlated to the development of a Sudden Stratospheric Warming event and hence the development of a -NAO and -AO. Until 2014, the index had performed very well, and there were high hopes for it’s future. However, the past several years have not been kind to the index. As such, we reduced the weight of this section in this year’s forecast to reflect its diminished importance, but I still think it is a factor worth considering.
The chart below explains the logic behind the index. We have seen some early signs of disruption in the stratosphere, but a full on split of the polar vortex and resulting SSW (Sudden Stratospheric Warming – the most surefire way to promote blocking across the northern tier) has not yet occurred.
Summary of Current Conditions: Little clear guidance one way or another from the SAI, but strong snow cover over North America suggests potential colder outbreaks being more intense later in the season. Sea surface temperatures suggest a storm track that would bring above average precipitation to our area. All in all, slightly below normal temperatures and slightly above average precipitation -SA
I’ve long believed that you have to trust your instincts when it comes to forecasting; if you go against your subconscious in favor of someone else’s opinions or a model consensus that doesn’t quite make sense, more often than not, you’re going to be burned. In this case, my gut is calling for a big winter in terms of cold and snow. Weak El Niño’s are our most prolific ENSO phase for winter weather, and this one seems to be playing out about as textbook as you can get for some sustained periods of winter.
With no real warning signs in the pipeline and with the history we’ve had with setups very similar to this one, I think we have to lean a little bit more bullish, and since the team started looking over data way back in September, the general consensus among all of us has been for a strong winter. El Niño’s often struggle in the early season before turning more wintry late; given that we’ve already seen a notable snow event and sustained cold and it’s only early December, I think we can put to bed some of the concerns that could be raised about the pattern never flipping. All in all, given the forecasting talent we’ve got on the team, to discount the collective gut feeling of five excellent meteorologists and forecasters that we’re going to see a good winter would be lunacy.
Overall Summary of Gut Feeling: Below Normal Temperatures and Above Normal Precipitation. –SA and the SCW Team
Now it’s time to put all of this together into a forecast. To recap, here are our summaries of each section.
Overall Summary of the Seasonal Models: Near Normal Temperatures and Precipitation.
Overall Summary of the Analogs: Below Normal Temperatures and Near Normal Precipitation.
Overall Summary of ENSO: Slightly above average temperatures and above average precipitation.
Overall summary of the teleconnectors: Progressively colder temperatures relative to normal and above normal precipitation.
Summary of Current Conditions: Slightly below normal temperatures and slightly above average precipitation
Overall Summary of Gut Feeling: Below Normal Temperatures and Above Normal Precipitation.
Winter 2018-19 Forecast
Of the six factors that we looked at, we had four calling for below normal temperatures, one for near normal, and one for above normal (although the above normal one, ENSO, highlighted how we could see it turn cooler as well, which we may already be seeing happening). The data seems pretty convincing that we will see temps below normal, and we think we will see the deltas from normal progress further into the negative category as we enter the winter.
December: 0-1 degree below normal
January: 1-2 degrees below normal
February: 2-3 degrees below normal
Overall: 1-2 degrees below normal
Nothing we’ve looked at suggests any sort of dry spell will persist this winter, as every single indicator is either normal or above normal for precipitation this winter. We don’t see any reason to vary from that consensus, and so will go with a slightly above normal three-month precipitation departure of 105-145% of normal for the winter.
We think that December will be the driest month (only on account of being below normal so far with not much in sight for the next week), but we still think we will end up relatively near normal when all is said and done. We once again expect February to be our wettest month as a pattern ripe for coastal development emerges. January is the wild card; while the analogs suggest below normal precipitation for January, we really don’t see much to back that up. If we assume that January ends up normal, and February ends up notably above normal, that puts us at slightly above normal for the season overall, in line with our projections.
December: 80-120% of normal precipitation
January: 90-130% of normal precipitation
February: 120-160% of normal precipitation
Overall: 105-145% of normal precipitation
Finally, to the section that everyone really cares about. Cold temperatures and above normal precipitation usually suggests above normal snowfall unless there is some indicator to the contrary, a La Nina bringing a cold/dry, warm/wet pattern, for example. In this case we don’t see any reason to differ from the suggestion.
This winter, we’re going with 110-150% of normal snowfall for the winter (remember normal is ~27” in Bridgeport, ~45” in Hartford and in the 50-70” range in most of the hill towns). We expect February to be the snowiest month, with December likely to be our least snowy just based on what we’ve seen so far. January leans toward above average snowfall as we begin to see the atmosphere really respond to ENSO, especially in the lead into February.
December: Near normal snowfall
January: Slightly above average snowfall
February: Above average snowfall
Overall: Above average snowfall
As always, thank you for reading and trusting SCW. Please like, share, and interact with us on our social media--on Facebook to share our discussions, and follow us on Twitter @southernctwx to retweet our posts. Hit the buttons below to join. From our entire team, we wish you a Happy Holidays and we look forward to tracking for you this winter!