U.S. Winter 2016-17 Seasonal Outlook


Outlook Overview

  • Using three summer and fall datapoints (ENSO, Atlantic tropical cyclone activity, and summer longwave pattern), a first look at the severity, aggregate storm track, and snowfall anomalies for the winter of 2016-2017 is constructed.
  • A weak La Nina, struggling to develop in the central Pacific, will have trouble influencing the North American winter this year, making confidence in seasonal predictions lower than normal. Still, it does appear that the tendency for rapid pattern changes in the upcoming winter could bring normal to above normal snow and ice to the Ohio Valley and Upper Midwest, even as temperature anomalies will likely run somewhat above average generally.

Outlook and Discussion

The last several winters have established a trend of wild regional swings from one extreme of temperature and precipitation to the other. While the winters of 2013-14 and 2014-15 saw numerous all-time temperature and snowfall records fall, particularly in the Ohio Valley and New England, in 2015-16 a major El Nino event brought extreme warmth to the Midwest and East, which was punctuated by several very powerful winter storms. So, the natural question is whether this pattern of extremes will repeat yet again in 2016-17. In general, I believe seasonal forecasting is done best when it’s simplest, so this look ahead at winter 2015-16 will only consider a couple of variable: El Nino state, hurricane activity, and summer longwave pattern. While signals are fairly weak at this lead time, some potential trends are already beginning to reveal themselves in these predictors.

The first step is to identify cases in the historical record in which the primary driver of global weather patterns—the El Nino/Southern Oscillation, or ENSO— is a reasonable match for this year’s Pacific sea surface temperature anomalies. In 2016, this is a reasonably easy exercise in pattern matching, because this year’s La Nina event is in the slow process of emerging. As shown below left, the central Pacific is seeing -1C or so SST anomalies. Additionally, the uncertainty regarding the future evolution of this –ENSO event is declining; most models now suggest little additional intensification through the winter, with a slight possible weakening in early 2017. Thus, the set of potential analog winters begins with years in which the late summer Multivariate ENSO Index was slightly negative and trending cooler.


The next step is to eliminate years in which Northern Hemisphere tropical cyclone activity is not aligned with the near normal hurricane activity observed thus far and expected for the remainder of 2016. As tropical cyclones balance energy flow from the Equator to the Poles, less busy than usual Atlantic hurricane seasons usually prefigure active winters. This was the case in 2013, 2014, and 2015 when well below normal activity preceded severe U.S. winters. Finally, longwave jet patterns for the remaining potential analog years are individually compared to that observed in the summer of 2016, with poor matches thrown out. This process results in five historical years corresponding to 2016 conditions: 1954, 1959, 1981, 1988, and 2000. The anomalous 500mb heights are composited for early (Nov/Dec) and late winter (Jan/Feb) in these years, with a heavier weighting placed on the best fit of 2000. The results, a rough way of estimating whether winter will be colder or warmer than usual, are shown below. The same technique is also used to estimate seasonal precipitation anomalies. It should be noted that while the chance of a moderate or strong La Nina is diminishing, it is not zero. Should La Nina intensify further, these analog years will be less predictive, and a set of years more like 1975-76 and 1998-99 will need to be substituted in this event.


In any case, the analog method suggests at this time, with relatively low confidence, that a weak to moderate La Nina will struggle to apply too strong of an influence to mid-latitude weather in the Northern Hemisphere this winter, which will result in a rather variable pattern over the Plains and Midwest. Many of the analog years had only weak deviations from normal; however, several general trends emerged. The first is that many of these years underwent a rather dramatic anomaly reversal in the central U.S. between early (Nov/Dec) and later winter (Jan/Feb). In early winter, a trough kept the central and eastern U.S. fairly cool relative to average (or at least normal), while not providing a particularly favourable pattern for snow or ice. In several analog years, this trend continued into January as well.

However, this tendency often flipped towards the climatological heart of winter, with Gulf Coast ridging bringing some warmer conditions relative to average north into the Ohio Valley. At the same time, this lifting north of the subtropical jet stream moisture tap—along with the proclivity for these years to see frequent pattern changes—increased the chances of snow and ice over the central U.S., particularly over the southern Plains and southern Ohio Valley states. Of course, with over six months of forecast lead time for February and March, this is only a weak tendency than is subject to being subsumed by a different climate influence in the meantime.

Nevertheless, I think there are a few things that can be said about winter of 2016-17 even at this early stage. The first is that there is nothing in the analog set to indicate that sustained extreme warmth or cold will characterize this year in the Midwest, but that frequent swings and an unstable overall pattern is likely. Second, it seems this instability could increase the frequency of snow and ice events in the south central U.S., particularly later in the winter when more frozen precip is climatologically expected anyway. Overall, my first read on the upcoming winter is for somewhat above normal temperatures in the Ohio River Valley, and near to somewhat above normal temperatures in the Upper Midwest. As for precipitation, a variable longwave pattern means unsettled weather where the changes are happening, which seems to be the east central U.S. For November and December, a lack of truly cold air and unfavourable disturbance track will likely keep snow and ice below normal in the Ohio Valley and Upper Midwest, except perhaps the northwest reaches of the zone. However into January and beyond, the southeastern two-thirds of this region can potentially look for average to above average snow and ice accumulation with more frequent than normal “Lake cutter” lows, with the remainder of the northern Plains probably in the near normal precip bin. In short, while the winter of 2016-17 is unlikely to reach the extreme variability of severe cold, warmth, snow, and drought that last winter’s wild El Nino ride offered, the tendency for a high-variability pattern in the central U.S. means that this region can look for a potentially mild but still possibly stormy and snowy winter.