2009-2010 Winter Season Forcast

Technical Discussion

Issued November 15, 2009

The following sections are used in the development of my winter forecast:

Each section will have a detailed discussion and explanation as the reasoning for the upcoming winter.

Comparing Past Seasons

Taking a number of considerations into effect, it is possible to compare past winter seasons with the upcoming one.  The term “Analog Years” is used to compare past atmospheric conditions with the current state and make a possible interpolation of what may be expected for the winter season.

The analog years(winter seasons) I have come up with for the current season are as follows and also show the total seasonal snowfall during that season for both Washington, DC and Philadelphia, PA.. The Stations used are Regan National and Philadelphia International Airport.

The reasoning as to why these were chosen and their ranking is explained in the Summary and Analysis section.

Climate and Teleconnection Patterns

These patterns are already well understood in the Meteorolgical community. Rather than describe each in detail, only figures, short-term analysis and projections for the upcoming winter will be discussed.


The waters off the equatorial pacific have been warmer than average for the past several months, thus signaling that we have entered an El Nino.   There are many indicies out there for determining the phase of either one. The ones I use are the Souther Oscillation Index, Nino 3.4 region values and the Oceanic-Nino index. For the SOI, positive values indicate Lan Nina conditions, whereas negative valus indicate El Nino conditions. The threshold for a El Nino is an anomaly of 0.5 degrees Celcuis above average when using the Nino 3.4 and ONI values. For more information on the Nino regions, thresholds and an overview of these values over the past few months, see NCDC's summary.

The anomalies in the Nino 3.4 region have been in the weak phase(0.5 to 1.0 degree celcuis above average) since June.  However, since mid-October, the warming in the 3.4 region has warmed significantly and is now into the moderate threshold.   For the week ending October 11th, the Nino 3.4 value was 0.83.  For the week ending November 8th, the value was 1.70.  Same can be said for the SOI.  That value has gone from a monthly average of 3.9 in September to -14.7 in October.  These values were taken from the Autralian Bureau of Meteorology.  Its amazing how quickly these have changed and this poses problems for the hope of a cold/snowy winter.   When taking a closer look at the ENSO region in the equatorial Pacific, the images show very nicely the rapid increase in anomalies over the past few weeks. Another factor to keep in mind is the depth of the warm water. This will show whether or not the El Nino will sustain itself or weaken rathery quickly. Also, this is more of a west-based El Nino, with the anomalies mainly in the Nino 3, 3.4 and 4 regions. The fourth image is cross-section in time, dating back to May 2006, showing the moderate El Nino of that year and the La Nina, the past two years.

Images courtesty of the TAO/Trition project

Weak El Ninos feature much colder weather and a lot more snowfall than a moderate one, for the most part. Moderate episodes can tend to go either way(pretty much like a coin flip), but do favor warmer and less snowfall.  An example is using the analog seasons I have chosen.  The 2002/2003 season saw a lot of snow from DC to Philly, while the 1994/1995 season saw very little and most the snowfall that season fell in February.  A Strong El Nino will pretty much guarantee the Eastern US a warm and snowless winter.  I don’t expect this to happen.  We’ll need to find something that favors a side, and that will be discussed in the next section, with the PDO pattern.  Here are the model forecasts of ENSO from Sept and Oct from the IRI website.

As you can see, most models are showing El Nino peaking at the moderate phase before leveling off and weaking at the beginning of 2010.

The seasons chosen for comparision were all moderate El Nino seasons. I verified that by looking at the Oceanic Nino Index and Nino 3.4 values. Please see the websites below for the values.
ONI Values

There has been a –PDO since October 2007 and that lasted all the way until May 2009. It has now been neutral for the past few months. There were signs in September that it was trending towards the positive phase. However, that ceased in October as the value dropped from 0.52 to 0.27. See the monthly values. SST’s also confirm it. Please see the SST section for images. It is going to be key this winter as to whether the PDO stays neutral or goes positive. In some of my current research, I am finding that the PDO is showing strong signals as a good predictor for seasonal snowfall and the type of snowstorms. For Philadelphia to DC, postive phases support more seasonal snow and increased chances for big(historic) snowstorms, while negative phases support less snowfall and smaller(nuisance) type snowstorms.

The results have been moderately significant, statistically. This is going back to 1950. The last two winters saw a moderate to strongly –PDO and both winters had below normal snowfall. For some more evidence, lets look at Philadelphia. The PDO was moderately or strongly positive in the 1995/1996, 2002/2003 and 2004/2005 seasons, on average from December to March. Here were the results:

Just wanted to show this as a reference. I'm not ready to claim this as a valid predictor and will need to watch over the next few years for any stronger statistical significance. These results only take the PDO independently and don't consider other patterns of influence.

There is no clear signal for neutral events. Pretty much a coin flip. For the analog seasons, both 2002/2003 and 1986/1987 were highly +PDO, 1994/1995 and 2006/2007 were neutral and 1965/1966 had a -PDO.


These are probably the most difficult at forecasting on a seasonal basis. These patterns change much more frequently than the ENSO and the PDO. They are more accurate and predictable on a daily, weekly and bi-monthly basis. Many studies have been done to try and link a connection with how one of these three patters averages over a particular month of the year and if that has any indication to the dominant phase for the following winter. Some of the months that I have seen meteorologists in the field use for connections are May, July, September and October. With that said, the phases of the ENSO and PDO can provide a forcing mechanism to determine what phase the NAO, AO, or PNA will dominate over the long-term.

As more research and findings come about over the coming years, we’ll see if any new results come about and how they will help in future seasonal forecating. Basically, If you want to see a lot of snowfall and much below average temperatures, then the following phases of these patterns will need to dominate this upcoming winter: +PNA, -NAO and -AO. I'm very bullish on the AO in particular. I have done some research and found results that the AO(when in the negative phase) is the most statiscially significant of all climate patterns when it comes to monthly/seasonal snowfall versus all the other patterns. I am also researching as to whether it is the most significant on increases in major/historical snowstorms.

In using the analog years above, here are the dominant phases of these patterns for those winter seasons from December to March of those seasons.


Values Courtesy of the Climate Predition Center

As you can see, there is some reliable information. In four of the five season, the NAO and AO was positve in December. The only exception was Dec. 2002. The reason for that in my opionion was that the PDO was higly positve that month and the preceding months before that time. Also, the NAO was positive for all months of March.

Eurasia and North American Snow Cover

September saw much below normal snow cover across Northern Canada and Eurasia.  However, in October, it recovered very quickly and rapidly caught back up to normal levels. Parts of Eurasia have reached above normal levels.  Here are the snow cover maps from Sept.30 and October 31st versus climatology.

September 30th Snow Cover September 30th Climatology October 31st Snow cover October 31st Climatology

Images Courtsey of the Rutgers Snow Lab

With the rapid increase in snow cover in October, will this have any effect on the upcoming winter and cancel out the much below levels in September?  The answer to that may be no because the snow cover in southern and eastern Canada is lacking this year and the first ten days of November did not provide any incouraging signs for an early winter. There is still very little snow cover in Southern and South-Eastern Canada. The snow cover in Eurasia continues to run above average

Sea Surface Temperatures

Sea-Surface Temperatures during the last three months are vastly different than this time last year.  The main reason is the La Nina pattern that dominated the last two winters.  In September, SST anomalies were very favoring for a cold and snowy winter, with indications of a +PDO developing and hints of a –NAO/-AO dominant winter in the North Atlantic. However, in October, there were some drastic changes, especially in the second half of the month.  The ENSO region has warmed significantly, especially in the Nino 3 and 4 regions.  October did not show continued evidence that a positive PDO was forming.  Since  the PDO was highly negative the past few years, at least a neutral phase would give a slightly higher probability of an increased significant snowfall threat.

2008 2009 Week ending Sept. 26, 2009 Week ending Oct. 26, 2009

Images Courtsey of the Unisys Weather

Finally, I wanted to also point out an area of anomalies off the Newfoundland coast.  Have been looking at this area the past few winters to see if SST’s here have any affect on the weather in the Eastern US.  This is pure speculation and analyzing on my part and have not found many studies to show any validity to it.  Last year, this area saw extremely above normal anomlies during October and November.  Positive anomalies in this area, in my opinion, have a strong signal for a +NAO and inland(western) storm track along the East Coast.  However, I have noticed that this area can change quite rapidly over the coarse of a few weeks to a month period. With that said, this has only an affect on a monthly to monthly basis and not seasonal.

This year, there are negative anomalies and the water cooled more and more the past few weeks.  I have shown last year and this year to show the sharp contrast in differences.  In doing some research, I have seen that in past winters, when there have been major east coast storms producing big snows from DC to Boston, negative anomalies dominated that area.  Even last year shows this as the anomalies went negative in late winter and there was a big east coast snowstorm at the beginning of March.  Also, these anomalies were positive both in fall 2007 and 2008.  This ended up leading to much above normal temperatures along the East coast for December, despite many forecasts predicting below normal temperatures.

Again, this is just some analyzing on my part and will continue to monitor this over the coming years to see if it has any significance.

September/October Synotic Analysis

It has been shown that watching and analyzing the synoptic pattern in September and October can give a good indication as to what the setup will be like for the upcoming winter season. This includes ridge/trough positions and the mean storm track, and temperature and precipitation outlooks. Storm track is important because it can determine what type of winter events you will see, i.e. all snow, Snow to rain, ice/mixed.

Here are the temperature anomalies and 500mb monthly heights for September and October.

September 500mb Anomalies September Temperature Anomalies October 500mb Anomalies October Temperature Anomalies

In September, there is a omega-block look to the anomalies, with strong heighs dominated the North Central and Western US, and all of Central Canada, while below normal heights were found in the Gulf of Alaska and Greenland and a weak area over the south-central US. This type of pattern implies a highly +NAO(negative anomlies over Greenland) and +PNA(positive heights over west Coast). The monthly value of 1.51 for the NAO and 1.29 for the PNA confirm it.

Monthly temperatures confirm the heights with above average anomalies in the North central and Western states, below average in the south-central staes and near normal in the East and Southeast states.

Quite a different pattern set up in October. There were well above normal heights in Alaska and the Western Arctic regions. Negative heights were found over all of the Central US and Mountain West and also off the Newfoundland Coast in the North-Central Atlantic. A small area of positive heights were found over the Bahamas and Florida. Temperatures were below average for all of the Central and Western US. Like September, temperatures were near normal for many areas along the East Coast and Southeast States. The only above average temperatures were found over Florida.

Precipitation both months favored above normal anomalies in the Southern States and Tennessee Valley. The signal was stronger in October and showing a pattern that we would see with a developing and strengthening El Nino.

Storm tracks featured mostly a dominant ENE trajectory through the United States from the southern plains through the Tennessee and Ohio valleys into the Northeastern US. There was only one coast event and that occurred in Mid-October.

All images in this section are courtesy are the Earth System Research Laboratory

Summary and Analysis

The rapid increase in the sea-surface temperatures at the surface and sub-surface in the western El Nino regions has significantly changed the complexion of the outlook for the upcoming winter. There is high confidence that we will be in a moderate El Nino for at least the first half of the winter. I do not see it reaching the strong phase. The big question still remains when we head into 2010. Does it weaken, and if so how fast? Will we see it weaken as rapid as thte 2006/2007 Winter Season? All analog seasons were moderate En Nino seasons. The differences in each being the timing of phase changes and locations of the warm pool in the ENSO region.

These are my rankings of the analog seaons with comments.

  1. 2006/2007- Fits the current trend of the ONI very well; West-Based El Nino. Rapid weaking of El Nino to weak phase. Neutral PDO
  2. 1965/1966- Potential good AO match. Also fits ONI very well. Moderate El Nino for most of the winter.
  3. 1994/1995- El Nino not west-based and slow weaking during winter
  4. 1986/1987- Moderately positive PDO; could end up being a great match for snow totals if PDO strengthes
  5. 2002/2003- Strongly positive PDO; Too cold and ignoring snow totals

The theme in all of these seasons is a warm start to the winter and then turning sharply colder sometime in January and continuing into all of February. The only exception was the winter of 2002/2003. There was a lot of snowfall and it was a very cold winter. I want to point out that the main difference in that winter was that the PDO was highly positive. This cannot be taken lightly and would caution anyone that is using it as a strong analog season. I pretty much consider it a throw out and it is ranked last of all the seasons I have used. Same can be said for the 1986/1987 analog, but to a lesser degree. The PDO is the "X" factor of the winter. If it strenthens into a moderate or high +PDO, then all bets are off and there will be a greatly increased threat of bigger snowstorms during the cold part of the winter season . The PDO is a pattern I am still learning and have very little skill in predicting it beyond a few weeks time.

In four of the five analog seasons, there was no snowfall at all in December or March for both Philadelphia and Washington, DC and December temperatures were above normal. All the snowfall occured in January and February. Again, the exception was 2002/20003. The following charts show this nicely.

Philadelphia, PA(KPHL)
Season Temperature Departure / Montly Snowfall
December January February March
1965/1966 + 3.1, 0.0" + 0.7, 16.0" - 0.6, 11.4" + 1.5, 0.0"
1986/1987 + 2.4, 0.4" + 0.7, 15.1" - 0.6, 10.1" + 3.9, 0.0"
1994/1995 + 6.1, 0.0" + 7.8, 0.0" - 1.5, 9.8" + 4.9, 0.0"
2002/2003 - 2.0, 8.4" - 3.7, 6.3" - 4.9, 29.6" + 1.7, 0.2"
2006/2007 + 5.3, 0.0" + 5.9, 2.2" - 6.8, 6.5" + 0.5, 4.7"
Washington, DC(KDCA)
Season Temperature Departure / Montly Snowfall
December January February March
1965/1966 + 3.4, 0.2" - 4.5, 21.3" - 1.6, 6.9" + 2.7, 0.0"
1986/1987 + 0.9, 0.0" - 0.5, 20.8" - 0.5, 10.3" + 1.9, 0.0"
1994/1995 + 4.8, 0.0" + 5.0, 3.9" - 3.1, 5.8" + 2.0, 0.4"
2002/2003 - 2.3, 7.1" - 3.8, 4.5" - 4.4, 28.7" + 0.6, 0.1"
2006/2007 + 4.7, 0.0" + 5.8, 1.3" - 7.2, 5.9" + 1.2, 1.9"

Maps showing temperature and precipitation anomalies for all the analog seasons accross the US.

There have been some other analog years used by other Meteorologists in the field. These include the 1957/58, 1963/1964 and 1977/1978. Of the three, I don't like the 1963/1964 season as it was mostly a weak El Nino and the 77/78 winter was one of the coldest on record and there are so many factors to look at, so that is not a good one to go by. However, the 57/58 season is intruiging. That was the only time since 1950 where we had a El Nino event following a multi-year La Nina. However, it was a strong El Nino event and is why I have left it out. Perhaps those could be good matches.