I’ve been doing some research on the various teleconnections and how they are shaping up for this winter. Based on my research, it appears a very mild winter is likely to occur during 2012-13.
The following indices were researched: the El Nino Southern Oscillation (ENSO), Pacific North-American Index (PNA), East-Pacific Oscillation (EPO), Arctic Oscillation (AO), North Atlantic Oscillation (NAO), Pacific Decadal Oscillation (PDO), Atlantic Multi-decadal Oscillation (AMO), analogs, climatic trends with respect to temperature and precipitation, recent pattern progression/repetition, autumn precipitation/temperature departures, sea surface temperature anomalies in both the Atlantic and Pacific, cryospheric evolution (trends, anomalies), and solar trends.
The following factors are what I put the greatest amount of focus on in making this winter forecast:
- Northern Hemispheric trends
- The Pacific pattern (ENSO, EPO, PDO/PNA, NOI/SOI*, etc..)
- The Northern Atlantic pattern (NAO)
NORTHERN HEMISPHERIC TRENDS
As you’re probably aware, we are currently in the midst of one of the warmest years on record in the Northern Hemisphere. The much warmer than average temperatures over the past decade has significantly affected SSTs and global land temperatures. Below is a graphic showing the mean global temperature departure for the 1999-2008 period against the 1940-1980 average.
Here is a graph of the earth’s temperature trends over the past 60 years, illustrating the impact of ENSO on the global temperature anomaly.
As you can see by these images, the earth is significantly warmer today than in previous decades. Regardless of what you feel is the cause of the trend, the point remains that the earth’s temperature (particularly in the Northern Hemisphere) has been rising rapidly since the 1970s. When making a long range forecast, it’s important to be cognizant of this long-term trend.
The main reason for me I bring this up is because North America and the US are affected by this the most in the winter months as the earth’s tilt is the farthest away from the sun, making temperatures more extreme. Over the past 10-20 years, there has been a marked trend towards warmer than average winters, especially across the Upper Midwest. Here is a map of the past 10 winters (DJFM) across the US versus the 1950-1995 long-term average. Note that the scale on this map is +/- 4 degrees.
Now that we have looked at temperature trends of the past 10 years/winters, let’s take a closer look at the past year across the US. As many of you are probably aware, it has been anomalously warm in the United States for the past 14 or 15 months. The string of warmth is, in fact, unprecedented in the historic temperature record.
Since the first of this year, 2012 is currently ranked number #1 warmest out of every year since 1895 for temperature to date. Here is the NCDC rank of top 10 warmest Jan-July periods in the past 118 years. It is interesting to note that two of the top five warmest Jan-July periods are important analogs to the current year, occurring in developing Nino years with similar global teleconnections. Namely, those years are 1986 and 2006.
In 2012, we have already set the following monthly records (based on 118 years of data):
- January 2012 was the 4th warmest on record.
- March 2012 was the 1st warmest on record.
- April 2012 was the 3rd warmest on record.
- May 2012 was the 2nd warmest on record.
- July 2012 was the 1st warmest on record.
As far as seasons go, we are on quite a streak as well (this is also based on 118 years of data):
- Summer 2011 (JJA) was the 1st warmest Summer on record.
- Fall 2011 (SON) was the 16th warmest Fall on record.
- Winter 11-12 (DJF) was the 4th warmest Winter on record.
- Spring 2012 (MAM) was the 1st warmest Spring on record.
- Summer 2012 (JJA) will also be among the warmest on record.
What all this says is that the warm trend is not only far from breaking, but is now stronger than it ever was.
In conclusion, what the warming trend demonstrates is that it is harder now to get an entire winter average out to be cooler than it was a few decades ago. However, we can still get very cold winters throughout much of the country in this pattern, as occurred during 2009-10 & 2010-11, if certain global factors set up right. Overall, these trends are a relatively small piece to the overall puzzle, but are worth mentioning.
THE PACIFIC PATTERN
ENSO (El Niño-Southern Oscillation)
It is still looking like this El Nino will peak in the low-end Moderate range. If my thinking is correct, Nino 3.4 SSTs will go over +1.1 sometime before the end of this year. The model consensus is slightly less than this, peaking around +1.0C during the OND period.
The reason I believe a somewhat stronger, although early peaking event is likely, is from research conducted by StormchaserChuck several years back. He found the earth undergoes a major change sometime in the Spring between March and May. At this time, future ENSO and NAO trends are set for the rest of the year. The second big signal, after equatorial Pacific subsurface water temperatures, to a coming Moderate to Strong ENSO event is given by springtime MEI values. Where the Mar-Apr MEI value is fairly neutral (i.e. below 1 and above -1), and this is followed by an increase in Apr-May by greater than 0.4C, the odds tip strongly in favor of a Moderate to Strong El Nino. In 2012, this pattern occurred:
The Mar-Apr MEI number was 0.059
The Apr-May MEI number was 0.706
This means that the change between these months was +0.65.
Prior to this year, there were only 8 years (since 1950) where this pattern occurred; that is, where a weak Mar-Apr MEI (nothing over or under 1.0) rose greater than +0.4 between Mar-Apr and Apr-May with the Apr-May MEI also being weak (nothing over or under 1.0). These 8 years were:
Every one of these years (100%) saw a CPC defined El Nino by the end of the year.
Every one of these years (100%) saw a MEI of greater than +0.5 by Jul-Aug.
Every one of these years (100%) saw the MEI reach +1.0 before the end of the year.
Every one of these years, with the exception of 2006, saw a Nino 3.4 max that was at or greater than +1.5 sometime before it died down.
Here are the maxes for my individual analog years (based on 1971-2000 climatology).
The 57-58 Nino maxed out at +1.6 making it a STRONG Nino.
The 65-66 Nino maxed out at +1.6 making it a STRONG Nino.
The 72-73 Nino maxed out at +2.1 making it a STRONG Nino and the 3rd biggest on record.
The 82-83 Nino maxed out at +2.3 making it a STRONG Nino and the 2nd biggest on record.
The 86-87 Nino maxed out at +1.6 making it a STRONG Nino.
The 97-98 Nino maxed out at +2.5 making it a STRONG Nino and the largest on record.
The 02-03 Nino maxed out at +1.5 making it a MODERATE Nino.
The 06-07 Nino maxed out at +1.1 making it a low-end MODERATE Nino.
The above correlation is based solely on data through the Spring. To get a better idea as to where we are heading, it’s important to look at what recent data is showing us as far as the progression of this El Nino goes.
The September OHC (Ocean Heat Content anomalies) are now warmer than they have ever been at any time this year, as the subsurface warm pool has gained a lot of strength over the past few weeks and began to surface. Below is a loop of the subsurface temperature anomalies of the past 2 months, which shows how OHC is progressing.
As of August 29, the weekly Nino 3.4 anomaly was measured at +0.9C, and is still rising. We have now risen +2.1C since early February, and +0.4C since the beginning of July. Nonetheless, the tri-monthly ONI for JJA was only 0.1C. This suggests that the September ONI will remain relatively low. Since 1950, there have been 7 times where the Sept ONI was < +0.8 that still went on to produce an ONI max of +1.0 or greater in the winter, as I am projecting:
1957 – Sept ONI (+0.8)
1963 – Sept ONI (+0.8)
1968 – Sept ONI (+0.2)
1986 – Sept ONI (+0.7)
1991 – Sept ONI (+0.8)
1994 – Sept ONI (+0.7)
2006 – Sept ONI (+0.6)
1957 – ONI max +1.6 (January)
1963 – ONI max +1.0 (Nov-Dec)
1968 – ONI max +1.0 (Jan-Feb)
1986 – ONI max +1.3 (January)
1991 – ONI max +1.8 (January)
1994 – ONI max +1.3 (December)
2006 – ONI max +1.1 (Nov-Dec)
Going by Nino 3.4 data (i.e. the ONI), it is safe to say that a Moderate or even borderline Strong El Nino peak is still not out of the question (although a Strong max is far from likely) climatologically speaking. I currently favor a low-end Moderate event.
Subsurface ENSO anomalies (or OHC) usually give a good indication as to when the El Nino or La Nina will peak. Climatology says that the peak in surface temperatures usually comes 1-2 months after the OHC has maxed out. With that being said, if you compare the location of where the warm pool below the surface currently is to where it was in past years, it is safe to say that the OHC will max out in October of this year. Given that, and how the ONI has progressed throughout the year, my call would be for Nino 3.4 temperatures to peak at the beginning of December, then decline through the winter. I expect the peak to be between +1.1C and +1.4C, making this a low-end Moderate El Nino event.
The decline throughout the winter should not be rapid, as there is no real cold water below the surface to enforce the drop like there were in other years. Based on the max, and expected decline rate… Dec, Jan, and maybe the beginning of Feb should have Nino 3.4 SSTs at or above +1.0C (Moderate Nino state), with late February into March dropping back into a weak Nino state. Given the 1-2 month lag between the SSTs and its effects on the pattern, I will classify this winter as being a low-end Moderate El Nino.
Since I dont believe that only SST numbers show all the effects of ENSO, I am going to use the MEI for potential winter ENSO analogs. The MEI is a measure of all areas of ENSO, including the SOI, OLR, SSTs. Since April, the MEI has been consistantly running higher than Nino 3.4 SSTs. This is one of those years where the SSTs are weakest part of the Nino present. Based on how the MEI has been progressing compared to SSTs, here are my favorite winter MEI analogs:
To illustrate the likely effects of a MEI winter similar to this one, here is a composite temperature anomaly map for the years given above:
PDO (Pacific Decadal Oscillation)
With that said, I thought it prudent to look elsewhere in the Pacific to get a better idea of how things are likely to progress this winter. In poring over the data on PDO, I noticed a strong correlation between August PDO values and the ensuing mean DJFM PDO modality.
All El Nino years since 1950 and their corresponding August & DJFM PDO modalities are given below. I’ve highlighted all years with a -PDO modality in August, and their ensuing winter PDO modality.
1951-52: August – negative, DJFM – negative
1957-58: August – positive, DJFM – positive
1963-64: August – negative, DJFM – negative
1965-66: August – negative, DJFM – negative
1968-69: August – negative, DJFM – negative
1969-70: August – positive, DJFM – positive
1972-73: August – positive, DJFM – negative
1976-77: August – positive, DJFM – positive
1977-78: August – negative, DJFM – positive
1982-83: August – positive, DJFM – positive
1986-87: August – positive, DJFM – positive
1987-88: August – positive, DJFM – positive
1991-92: August – positive, DJFM – positive
1994-95: August – negative, DJFM – negative
1997-98: August – positive, DJFM – positive
2002-03: August – positive, DJFM – positive
2004-05: August – negative, DJFM – positive
2006-07: August – negative, DJFM – negative
2009-10: August – positive, DJFM – positive
16/19 years, or 84% featured a positive correlation between October and the ensuing winter, DJFM. There were eight years with a negative PDO in August, six of them were followed by -PDO winters. The two exceptions were 1977 and 2004, which both had a positive DJFM PDO that was preceded by negative values in October. I feel this is unlikely in 2012, given the overall -PDO regime we have been in recently and no evidence of warming SSTs off coastal California.
Note, too, that of the four closest MEI analogs (1957-58, 1965-66, 1994-95, and 2006-07), three of them occurred during a -PDO, despite the fact that El Nino events are rarer during -PDO modalities (only eight of the 19 cases above occurred during a -PDO). As you can see on the graphic below, the warm pool south of Alaska and the cooler anomalies along the California coastline are traditionally associated with a -PDO regime. Note also the developing El Nino visible on the SST anomaly map below.
PNA (Pacific/North American Pattern Index)
The correlation between the PDO (which is more of a SST index), and the PNA (which is a pattern index) is greater than 0.80 in the winter months. Given the expectation of a -PDO winter, it might be reasonable to expect a correspondingly negative PNA. Somewhat surprisingly, however, the correlation disappears in -PDO, El Nino winters. Only three of the eight cases had a -PNA. Those were 1951-52, 1965-66, and 1968-69. In all remaining years, the mean DJFM PNA was positive. Since PNA and EPO are inversely correlated, this arrangement would favor a -EPO, which tends to favor enhanced blocking in West Canada and Alaska.
Unfortunately, StormchaserChuck has shown that the impact of -EPO is minimal during El Nino winters. This was evident during the winters of 1991-92 and 1994-95 and, to a lesser extent, 2006-07. During these winters, the EPO was quite negative, but the Pacific flow across the northern US and Canada simply overpowered the -EPO effects. Because of this, it is probable that the EPO will not be a significant factor during this winter.
Expect the following indices this winter in the Pacific:
- Moderate El Nino, weakening throughout the winter
- Negative PDO, but positive PNA
- Somewhat negative EPO, but having minimal impact
The best analogs for the upcoming winter, based on the state of the Pacific include:
1994-95 (personal favorite)
The composite temperature anomaly map for these six winters is given below:
THE ATLANTIC PATTERN
NAO (North Atlantic Oscillation Index)
Aside from extreme ENSO events such as 1982-83 and 1997-98, the NAO is probably the most important long term factor that determines how the pattern sets up across North America and Europe. StormchaserChuck described a novel approach for predicting the NAO state up to six months in advance with reasonable accuracy.
First, some background, the NAO is a pressure index that lies roughly between Iceland and the Azores. A lower pressure in Iceland with a higher pressure over the Azores would result in a +NAO. A higher pressure in Iceland with a lower pressure over the Azores would result in a -NAO. Sea temperatures both on the surface, and below the surface have a big effect on pressure changes in these regions. For more on the relationship between subsurface temperatures and pressure indices, see this thread:
In a nutshell, Chuck’s method involved dividing the Northwest Atlantic, from the Davis Strait south to near 35N, into two regions. Below is a map of the Gulf stream and North Atlantic steering currents. The red dot on the map is approximately where the Azores are. The blue dot on the map is where Iceland lies. Around these two areas are where the NAO is measured. The black circled area is one of the areas Chuck’s research focused on. Chuck called this region “Area B.”
The 2nd area that Chuck focused on is the dark blue circled region just south of Greenland, which he referred to as “Area A.”
As shown in the images above, the currents that lead to the Azores ultimitely come from the North and West, and are originally located south of New Foundland (the black circle in the 1st map). The currents that lead to Iceland come from all directions, but mostly west of where the island is. This is very important to note since subsurface temperatures play a major impact on pressure indexes, and subsurface temperatures travel with the ocean currents.
Chuck focused on two times of the year where SSTs are most changeable due to the subsurface temperatures emerging — namely, the Summer solstice and the Winter solstice. In between these seasons (Summer and Winter), Chuck posited that the warm or cold SSTs tend to plunge below the surface as the Earth’s pattern changes, comparing it to the seasonal progression of subsurface temps in the ENSO regions. This mechanism explains why we usually know if a Nino/Nina will be moderate or strong by the Summer, and also the reason why ENSO events usually reach their peak around Dec 21st.
With that being said, Chuck theorized that one could use May-August SSTs (i.e. centered around the Summer solstice) in certain areas of the globe to accurately predict where warm or cold subsurface temperatures will travel to in a given time (depending upon how strong the current is). He did just this with the NAO.
To get an idea as to what state the pressure indexes near Iceland and the Azores will be in the winter, Chuck broke down May-Aug SSTs across areas A and B in the maps below to get an anomaly.
To get a very accurate reading, Chuck took every year since 1949, magnified the map, and broke it down block by block, like what is shown below for 2006. He added the boxes up, and divided them by however many blocks there were to get an even number for the entire area. This was based on the anomalies from 35N to 70N as well as 70W to 30W, with everything north of 52N being counted as area A, and everything south of 52N being area B.
Since the total of Area A came out to be higher than Area B, Chuck multiplied the reading from area B by 1.51 to make it even. To get a total anomaly from the two regions, Chuck subtracted the number he received from Area A from the number he got from Area B. The total data of Area A, Area B, and the total of the two regions for every year from 1949 to 2005 can be found at the link below:
Using this method alone, Chuck reports a Standard Deviation of just 0.47
Looking closer into the years that had the most error, Chuck discovered that the ENSO state can throw this formula through some loops. In the 6 strong ENSO winters since 1950, where the MEI averaged out to be above +1.5 (strong Nino) or below -1.5 (strong Nina), the average deviation was off by +0.66. Seeing how the ENSO state plays a strong role on the entire Global pattern, he added a +0.50 anomaly to the years that were and are expected to be a Strong Nino or Nina.
In the 12 Moderate ENSO winters since 1950, where the MEI averaged out to be between +1.0 and +1.5 or between -1.0 and -1.5 for the winter, the average deviation was off by +0.24. Since Moderate ENSO events can impact the NAO, but not to the extent of a strong ENSO event, he added +0.20 to every winter that was a Moderate Nino or Nina.
There was no strong error rate one way or another with ENSO neutral or weak ENSO winters, so Chuck left these unadjusted. Including ENSO into the formula brought the average deviation down from 0.47 to 0.42.
Here are two charts showing the correlation between the predicted and actual NAO states using Chuck’s formula:
With that background out of the way, it’s time to look at where SST anomalies were this summer for the relevant region.
We can see that most of the NW Atlantic is running well above normal. However, the warmest anomalies seem to be located in the vicinity of Area B along and south of PEI. This would favor a +NAO, especially when the 1.51 multiplier correction factor is applied to Area B. In addition, the expected Moderate El Nino would somewhat amplify this effect. I am not going to count pixels as Chuck did, but there does seem to be support for a positive NAO on this basis.
It’s worth noting another strong correlation involves October NAO state. It has been noted that the winter (DJFM) mean is often the opposite of the October monthly NAO. In other words, a strongly negative or positive NAO in October is typically followed by the opposite during the winter. Obviously, it is not yet October but we should have more information once we get to that point. In addition, above normal October snow cover in Siberia is thought to favor a negative NAO during the winter, according to research by Dr. Judah Cohen. Currently, snow cover has been running below normal, but we could see rapid increases over the coming month or so.
Finally, it was also pointed out to me that, during El Nino years, the NAO index is frequently the same sign of the QBO index (h/t NorEaster07).
Currently, the QBO is near -28, which would favor a -NAO. However, this QBO reading is on the tail end of the its biennial falling cycle. The QBO has now peaked and will be beginning its rise towards a maximum value sometime next year. In fact, the year closest to this one is none other than — 1994! Yes, during 1994, the negative-QBO cycle peaked in July and August right around -28, which is close to where we currently reside. However, the QBO index rapidly rose over the fall, reaching above zero by December! The QBO remained positive throughout the winter of 1994-95 and the NAO was, as predicted by this correlation, highly positive. The complete QBO data is available here: http://www.esrl.noaa.gov/psd/data/correlation/qbo.data.
Note the striking similarity in this year’s QBO and that present in 1994. There have been other years with a similar pattern, but none featured an El Nino, like that expected this winter and that occurred during the Winter of 1994-95. In light of the QBO signal and the current observed SSTs in the NW Atlantic, it appears that a positive NAO should be the favored state this winter. That is not to say that periods of -NAO cannot occur, but that they will be the exception rather than the rule.
AO (Arctic Oscillation)
With respect to the AO, El Nino years overwhelmingly tend to feature a -AO. But the AO is closely related to the NAO (correlation ~0.85 during the winter). In the years that featured a predominantly +NAO, a +AO was also favored. This was the case in both 1994-95 and 2006-07, which have consistently appeared as the closest analogs to the current year.
It’s worth noting, also, that Arctic sea ice is at unprecedentedly low levels as shown in the graphic below.
It’s not yet known what effects this may have on winter conditions, but I suspect it favors a +AO. There is some anecdotal evidence that low sea ice leads to more and stronger Arctic cyclones. Deep low pressure over the Arctic favors a -NAO and -AO, as I already explained how those indices operate. In August, we observed a record-breaking 963mb low over the Arctic Ocean and currently a 988mb low is affecting the same region. Whether this trend continues is not yet knowable, but it does lend some support to the idea. I should add that this is one proposed mechanism to explain the warm, wet winters characteristic of the mid-Pliocene in the middle Latitudes and may become a more permanent fixture in coming years.
Based on the information mentioned above, I am using the following to make a forecast for the winter:
- Warm seasonal trends.
- Low-end Moderate El Nino, peaking early.
- NAO, averaged greater than +0.5.
- PNA, averaged greater than +0.5 (with a negative PDO).
- EPO, averaging somewhat negative to neutral, but having minimal impact.
The following analogs appear as the best fits: 1986-87, 1994-95 and 2006-07. Of these years, I prefer 1994-95, followed by 2006-07. For the composite analog map below, I triple weighted 1994-95 and double-weighted 2006-07.
I plan on producing some maps of my own, as well as going into greater detail about how I foresee the winter. Basically, however I suspect it will be mild overall with less than normal snowfall. The best chances of snow & cold will be from late January and on, with the first half of winter likely featuring unusual warmth.