There are some preliminary signals that next winter could be on the cold and snowy side, again.  I should stress preliminary as much can change between now and then.  But let me walk you through my thought process.

A couple weeks ago, Jason Samenow sent me an article on headlined: “Possible signs of snowy winter

Jason asked: “Is it too early to be talking about next winter”?

My answer is that for our audience, it is never too early, especially with our typical hot and humid Washington D.C. summer in full progress. blog discusses the NOAA CFS model forecast for next winter’s Pacific sea surface temperatures and how the forecast seems to favor a colder, snowier winter setup for the Eastern U.S. akin to the notorious 2003-03 and 2009-10 winters. Let’s take a closer look at the claim.

The shoe might fit. Here is a recent CFS model outlook for December to February 2014-2015 (you can download an even more current version here, which looks much the same). I marked the areas for which you should pay particularly close attention:

Notice that water temperatures in the Gulf of Alaska are warmer than normal, while there is a colder-than-normal patch of waters north and west of Hawaii.   Also, notice that the warmest anomalies are just east of the international dateline.  These ocean water patterns share a lot in common with recent snowy winters.

For this post, I examined ocean water pattern for the four El Niño events of the 2000s as they divided cleanly between less snowy (and warmer) vs. colder and snowier outcomes using data from Washington Reagan National Airport, shown here:



These maps below show what the sea surface temperature anomalies looked like for both the warmer/drier and colder/snowier camps:

Notice that our big, snowier winters also had stronger El Niño water temperature warming toward the dateline (vs. the other warmer/drier camp) and there was more of this sea surface temperature anomaly dichotomy between the warm Gulf of Alaska waters and cold anomalies north and west of Hawaii. Just from this simple analysis, I would tend to agree with that the CFS model is showing a Pacific water temperature pattern consistent with our colder/snowier El Niño winters here in the 2000s.

Same but different. There is actually more research on this issue. El Niño events that are stronger toward the international dateline are known as Modoki, which is Japanese for “same, but different.” You can read more about these Modoki events here.  This whole Modoki business helps to explain an old mystery: El Niño winters in the 1980s and 1990s were almost uniformly warm and snowless for the Washington, DC area, but in the 2000s the rate has been 50:50 as shown above. Why did it change?

Blame the PDO. The Pacific Decadal Oscillation is a long-term cycle that takes 50-60 years to fully complete. We are currently in the negative long-term phase, which is closer to the 1950s-1970s than the 1980s-1990s. From the data, it seems that Modoki events are more common during the negative phase of the PDO, which may explain why El Niño winters were snowier during the 1950s-1970s than they were during the 1980s-1990s for us on the East Coast. This also means that the odds are better than normal that we could do it yet again this coming winter (yes, I just said that!).

Physical explanation. Having warmer waters near the dateline increases the chances that tropical convection (thunderstorm) energy can feed into the jet stream in the right spot to fuel stronger North Pacific ridging and leading to more high-latitude blocking patterns (which sends more cold air down to the mid-latitudes). This is an element of the Madden-Julian Oscillation (MJO). Combining increasing blocking with enhanced subtropical moisture flow that all El Niño events provide means a cold + stormy combo that gives us our crazier winters.

Time for the caveats. I would definitely file this situation as “interesting” at this early stage and not anything approaching a forecaster’s level of confidence.

Here are some reasons why:

  • This is only a model projection and a lot could change on the model inputs between now and the autumn to adjust this view; I’ve noticed that in recent years, the CFS model has overestimated El Niño intensity, which could work against this forecast by winter; other model data availability for next winter is very little this early in the game
  • For the statistical fanatics, I fully agree that nearly all sample sizes for El Niño events are too small, particularly now that you break them up between Modoki and non-Modoki cases; we use physical modeling to help fill the gaps and as described above, we believe we understand it, but there are always risks (thanks to chaos theory) for other variables to become more dominant/influential in the winter outcome
  • The CFS model itself is not calling for a cold/snowier winter for the Eastern U.S.  It forecasts warmer-than-normal temperatures and near normal precipitation; sea surface temperatures are easier to forecast long-term because they change more slowly than air temperatures due to specific heat differences, so at some point the CFS split (between the Pacific ocean temperature pattern supporting a snowy winter versus its forecast for a warm Eastern U.S.) should be resolved- probably this autumn
  • There is still no guarantee that an El Niño will occur through this winter; the odds seem to be growing, but some data we track (like subsurface ocean warmth) has weakened

In other words, we’re saying there is a chance that this winter’s El Niño could be a Modoki version, which is colder and snowier than normal. Based on the evidence, my qualitative assessment would be a 55% chance, based on the long-term –PDO cycle boost. Stay tuned!