Sinds 1974
Vereniging voor Weerkunde en Klimatologie

De laatste weerflitsen

  • 07:24
    (6/8) Zwaar bewolkt
    Windrichting: ZW
    Windkracht: 2 = Zwakke wind
  • 22:50
    (8/8) Geheel bewolkt
Bakker & Co Het magazine voor weerliefhebbers Weerhuisje

Interview with Judah Cohen

Judah Cohen is one of the world's leading scientists in winter forecast. In 1999 he and Entekhabi demonstrated a strong statistical relationship between Eurasian snow cover extent (SCE) anomalies in the fall and the Arctic Oscillation in winter, later on he found a strong correlation between Eurasian snow cover advance and winter AO. Nowadays he is Director of Seasonal Forecasting at AER and still publishing lot's of articles. Time for an interview.

Question 1. Please if you would give us some information about yourself, your background and interests, career and current work.


I am Director of Seasonal Forecasting at Atmospheric and Environmental Research (AER). Previously I spent two years as a National Research Council Fellow at the NASA Goddard Institute for Space Studies and two years as a research scientist at MIT's Parsons Laboratory where I still maintain a Research Affiliate position. I received my Ph.D. in Atmospheric Sciences from Columbia University in 1994 and have since focused on conducting numerical experiments with global climate models and advanced statistical techniques to better understand climate variability and to improve climate prediction.


Question 2. You are quite famous within the meteorological community with the findings on the correlation between the October snow cover in Siberia and the winter AO. Since your first publication with Entekhabi ‘Eurasian snow cover variability and Northern Hemisphere climate predictability’ you have published more than 40 articles. Could you elaborate on the developments of your research, since that first publication? E.g. the step from SCE to SAI?


One correction, now it is closer to 60 publications.


Cohen and Entekhabi (1999) demonstrated a strong statistical relationship between Eurasian snow cover extent (SCE) anomalies in the fall and the Arctic Oscillation in winter. They hypothesized that a possible dynamical mechanism linking Eurasian snow cover anomalies and Northern Hemisphere climate variability, is through the strength and position of the Siberian high. Using a proxy index for the AO, Cohen et al. (2001) showed that the winter surface AO originates as a lower tropospheric regional height anomaly in Siberia during the fall. Persistent positive sea level pressure (SLP) anomalies and negative surface temperature anomalies in the region of Siberia, as early as October, were found to precede a negative winter AO. This provided a further link of the hemispheric scale AO with the inception of the Siberian high in the fall.


At the time that we were arguing that anomalies in Eurasian snow cover preceded large AO events, other scientists were arguing that anomalies in the stratosphere preceded large AO events in the troposphere and at the surface. So we tried to reconcile the two competing theories. It was well known that variability in the stratosphere is forced by propagating wave energy from the troposphere known as Eliassen Palm Flux or Wave Activity Flux (WAF). We were able to show that anomalies in Eurasian snow cover influenced the amount of WAF that propagated in the fall from the troposphere to the stratosphere. Changes in the WAF resulted in circulation changes in the stratosphere that eventually propagated down to the troposphere, influencing the mean winter climate of the Northern Hemisphere. We demonstrated this link between snow cover, WAF, the stratospheric circulation and the winter AO in both the observations (Saito et al. 2001) and in global climate models (GCMs; Gong et al. 2003). So, for example, if Eurasian snow cover is more extensive than normal, this favors increased energy propagation or WAF from the troposphere to the stratosphere. More energy absorbed in the stratosphere forces a stratospheric warming. Then the circulation anomalies associated with a stratospheric warming propagate to the troposphere including high pressure over the Arctic, a southward displaced jet and colder weather in the mid-latitudes. All of these circulation anomalies are most closely associated with the negative phase of the AO. Finally, in Cohen et al. 2007, we tied all the steps together where we argued for a six-step process starting with variability in snow cover and ending with large winter AO events.


The six steps for when snow cover is more extensive than normal are: 1) Snow cover increases rapidly in the fall across Siberia. 2)When snow cover is above normal- diabatic cooling helps strengthen the Siberian high and leads to below normal temperatures. 3) Snow-forced diabatic cooling in proximity to the high topography of Asia increases upward flux of wave activity from the troposphere, which is absorbed in the stratosphere. 4) Strong convergence of WAF leads to higher geopotential heights, a weakened polar vortex, and warmer temperatures in the stratosphere. 5) Zonal mean geopotential height and wind anomalies propagate down from the stratosphere into the troposphere all the way to the surface. 6) Dynamic pathway culminates with strong negative phase of the Arctic Oscillation at the surface in mid-winter. The pathway is linear and the opposite happens when snow cover is more sparse than normal resulting in a positive AO.
Until now all the relationships demonstrated between Eurasian snow cover and the winter AO used the monthly mean Eurasian snow cover extent for October. However in the winter of 2009/10 the AO achieved record negative values. October SCE was above normal, consistent with the negative winter AO but not exceptional. But what I did notice that was exceptional was the rate at which the snow cover advanced during the month of October, especially in the latter half of the month. So instead of correlating the monthly mean snow cover extent with the AO, we created a new index that measures the rate at which the daily snow cover expands over the entire month of October and correlate it with the winter AO (Cohen and Jones 2011a). This new index we refer to as the snow advance index or SAI and it is correlated with the winter AO at much higher values than the more traditional SCE index. Since its discovery we have been using the SAI in producing our winter forecasts instead of the SCE index.


Question 3. There are six steps in the troposphere-stratosphere-troposphere model you introduced. You made it clear that it takes some time for the October snowfall to take effect, not only on regional scale, but also over larger parts of the Northern Hemisphere. We understood from your six-step model that only in January do the stratospheric circulation anomalies propagate downwards into the troposphere and that only then or later this may cause the AO to become (more) negative. So we consider that it is not wrong to predict the winter (DJF) AO, if the effects of the Eurasian snow cover do not occur before January, February and perhaps March. In short: is SAI useful for predicting December temperatures as well? E.g. was the cold December 2010 forecast using the SAI?

AER's winter Seasonal Forecasts rely on Siberian snow cover extent in the fall for more accurate winter predictions. Over a decade of research has allowed AER to understand how variability in Siberian snow cover mostly in October can influence the weather in remote regions including the Eastern US and Europe months later. A six-step model with a timeline that begins with the advance of Siberian snow cover in October and ends with either more (less) frequent Arctic outbreaks during the winter in the Eastern US, Europe and East Asia associated with the negative (positive) phase of the large-scale teleconnection pattern the Arctic Oscillation (AO).i

December 2010 was in large parts of the Northern hemisphere very cold

December 2010 was in large parts of the Northern hemisphere very cold.ii


We did predict a cold winter in 2010/11 including a cold December. I included below the winter forecast that I presented at a prediction workshop at the end of October in Bergen, Norway. You are correct that the signal most often propagates from the stratosphere to the troposphere in January. This is the culmination of the six-step process and is when the atmospheric response to the snow cover anomalies becomes hemispheric in scale. However the second step is the strengthening and expansion of the Siberian high. This takes place in November and December and is limited to the Eurasian continent (Cohen et al. 2014). So with the strengthening and expansion of the Siberian high it turns colder across Eurasia including Europe and it may or may not include North America. In December 2010 it turned colder in Europe and in the US, however in December 2012 it turned colder in Europe when the Siberian high expanded but it remained mild across the US. But then in January 2013 there was a stratospheric warming and a downward propagation of the signal. Post the stratospheric warming, the signal became hemispheric in scale and Europe remained cold and now the US was cold as well.

Observed and fortecast temperatures 


Question 4. We are truly astonished at the accuracy of the January-March 2013 temperature forecast for Europe. This detailed winter temperature forecast seems to have been prepared with other data than the SAI alone. Could you tell us more about the procedure? And are these forecasts now (or in the future) freely available?

The temperature forecast January to March 2013.

The temperature forecast January to March 2013.


The new seasonal forecast model has been continuously updated and tested using hindcasts and real-time forecasts. In addition to incorporating data on October Eurasian snow cover extent, sea surface temperature, other inputs into the forecasts include anomalies from the fall season such as sea level pressure anomalies and sea ice concentration anomalies.iii

We use other predictors besides the SAI. We also use El Niño/Southern Oscillation (ENSO) and an index to represent the expansion or contraction of the Siberian high. We showed that the expansion/contraction of the Siberian high is a precursor to stratospheric variability that is at least partially independent of snow cover that can influence the Northern Hemisphere winter climate (Cohen et al. 2001; Cohen and Jones 2011b). Starting last winter we also used sea ice extent anomalies in our forecast model. The forecast model and its skill is presented in Cohen and Fletcher (2007).

For the past several years I have been publishing the winter forecast on the National Science Foundation (NSF) website: And as long as NSF continues to fund me, I will post the winter forecasts on their website.


Question 5. In the article ‘Warm Arctic, Cold Continents’ (2013) you write in the conclusion: “The strong coupling between sea ice and atmosphere, possibly through convective feedback in the dynamical models, may disrupt the coupling between a warm Arctic and cold continents found in observations. These lessons from seasonal prediction indicate that rapid response of sea ice to external forcing, as expressed in past abrupt climate change, may lead to future surprises in the Arctic, thus increasing the uncertainties in future climate projections for the entire Northern Hemisphere“. Please tell us more about your thoughts on future surprises in the Arctic and uncertainties in future climate projections.


To me it is fascinating how winters warmed as predicted due to increasing greenhouse gases up until the 1990’s but then surprisingly turned noticeably colder. This temperature reversal from warm to cold winters was not predicted by the GCMs. Though many scientists argue that the string of cold winters is explained by natural variability, I don’t agree. And I have argued in my papers that the negative trend in the winter AO and the resultant winter cooling is at least partially due to an increasing snow cover trend in the fall (Cohen and Barlow 2005; Cohen et al. 2009; Cohen et al. 2012). Also GCMs are the most relied on tools for seasonal forecasting. The GCMs always predict warm winters across the Northern Hemisphere. This has been true throughout the streak of cold winters over the past six years that we have experienced and for this winter they continue to predict a warm winter across the entire Northern Hemisphere (though of course we don’t know yet and this forecast could turn out to be correct but I am very skeptical). It is my growing belief that a “warm Arctic, cold continents” is a fundamental dynamical feature of the winter atmosphere. Yet in the GCMs a “warm Arctic, warm continents” is the prevailing dynamic and is a fatal flaw in the models. It is my belief that in Nature a warm Arctic (due to sea ice melt) favors cold continents in winter but in the model when the Arctic warms due to sea ice melt, the warmth also extends into the continents. In my opinion the inability of the GCMs to correctly simulate both snow-atmosphere and ice-atmosphere coupling handicaps the ability of GCMs to provide not only reliable seasonal winter forecasts but makes longer term projections highly uncertain, at least in the winter season. I would say in general continue to expect overall colder winters than predicted/projected by the GCMs into the foreseeable future.


Question 6. The KNMI released new climate scenarios in 2014 for the coming decades. In terms of wind flow there are two options. In option one the wind flow will not change (or only slightly change) and in option two the wind flow will change, with more westerly winds during the winter.iv Some scientists suggest (e.g. Jennifer Francis: ‘Arctic amplification’) there will be more Jetstream amplification in the future, whereas according to the KNMI scenarios, the AO will remain the same or turn more positive. If we understand Francis correctly, the AO will slowly become extremely negative. Is that your view?


See the text for an explanationion

In 2014 the KNMI released an update of their climate scenarios. On the horizontal axis the worldwide temperature. Two options; moderate or warm. On the vertical axis the change of wind flow. Two possibilities, no or just a little bit change or far more western winds in the winter.


I cannot speak for Jennifer Francis but she has not really addressed the AO in her papers and presentations. Her idea is that as the Arctic warms relative to the mid-latitudes and the meridional temperature gradient weakens, the zonal component (west-east direction) of the jet stream will weaken but the meridional component (north-south) will strengthen. Together this will result in more amplified waves. Also waves in the atmosphere will move more slowly or become blocked and their associated weather will be more persistent. You could extrapolate out and say the positive AO is related to more zonal flow and the negative AO to more amplified and blocked flow. But as far as I know Jennifer has never said this.


I am very skeptical of the KNMI forecasts that you discuss. For two decades now there have been projections of a trend towards more positive winter AO conditions. I am very proud that as early as in 2005 I published a paper showing that the AO is not in a long term positive trend and in fact is in a negative trend, in part due to an increasing trend in Eurasian snow cover. It is now a decade after that paper was published and it has turned out to be quite prescient with the trend in the AO turning even more negative since the paper was published. The models also have consistently projected a poleward shift in the jet stream and a strengthening of the zonal flow on the poleward side of the jet stream. So far neither of these has been observed, and in fact, just the opposite has been observed. But the AO and the position of the jet stream have been variable or even cyclical and I would expect for that to continue.


You might be interested to know that Jennifer and I among other scientists just published a review paper going into much detail on this subject titled “Recent Arctic amplification and mid-latitude extreme weather.” You can download the paper from the following link:


Question 7. You suggested in 2012 (‘Arctic warming, increasing snow cover and widespread boreal winter cooling’) that sea ice loss has contributed to humidity in the Arctic, resulting in more extensive snow cover in the fall, that in turn forced a dynamical response in the atmosphere favourable to a negative winter AO. The enhanced snow accumulation is also consistent with the idea that low sea ice and higher temperatures lead to enhanced snow accumulation and possibly the beginning of another ice age.

JFM Season Standardized AO index

The January to March AO according to NOAA.v


NOAA published this graph of the January to March AO. Although the October snowfall has increased significantly since 2000, even compared to pre 1988 levels, the AO did not respond accordingly. Could there also be long term factors involved, e.g. the AMO or the phase of the PDO, which contribute to the anomalies of the Arctic Oscillation?


Not sure what you are referring to as the AO has been in a strong negative trend since 1988 as shown in the graph that you present from NOAA. This also nicely matches the increasing trend in October snow cover extent since 1988 (see chart below snow ver anomalies are multiplied by -1 for ease of comparison). Over the longer term (since 1950 as shown in the graph) the AO’s trend is neutral. But since 1950 the snow cover extent trend is also neutral (as you show from my graph below) so the trend of the winter AO and October snow cover match very nicely on decadal time scales. Certainly other factors can influence the AO such as the AMO or maybe even the PDO but I have not studied these relationships enough where I feel that there is much that I can say intelligently on the subject.

DJF AO and Oct SCE Anomaly 1949-2012

Question 8. In 2011 you wrote with Justin Jones in ‘A new index for more accurate winter predictions': "An important unanswered question is why the October SAI is more highly correlated with the DJF AO than the October SCE index. One likely reason is that the SAI is limited to latitudes south of 60°N while the SCE index includes all of Eurasia, which has a significant amount of snow cover north of 60°N. Assuming that the high albedo of snow cover is one if not the most important snow characteristic influencing the overlying atmosphere, this would favour the SAI, which is limited to regions that are exposed to a higher sun angle more so than the SCE, in predicting the atmospheric response to snow cover variability. Another possibility is that the SAI is sensitive to the timing of snowfall, where snowfall at the end of the month contributes to higher values of the SAI and snowfall at the beginning of the month contributes to lower SAI values, while the monthly mean SCE is insensitive to the timing of snowfall". Have you developed any new insights since then? Moreover: in October 2013 there occurred a unique situation. With a high SCE and simultaneously low SAI, partly a result of September snowfall, the high SCE suggested a negative AO and the low SAI implied a positive one. What lessons can be learned from this kind of situation?


Why the SAI is more highly correlated with the AO than SCE is an important question that I still don’t have a good answer for. The two points that you write are ideas that I have shared and I think make sense. Another idea that I can add is that if you correlate the SCE of different regions in Eurasia with the winter AO, almost all the skill comes from eastern Eurasia and much less from western Eurasia. If you also look at the variability in October snow cover across all of Eurasia it is south of 60°N in eastern Eurasia but north of 60°N in western Eurasia. So again because the SAI is limited to snow cover variability south of 60°N, it is better at isolating the signal from the noise in snow cover since by construct it weighs more heavily snow cover extent variability in eastern Eurasia. But I still consider all of these answers incomplete.


October 2013 was very interesting. It is unique in the observations in that SCE and the SAI diverged so strongly. Because of this divergence in the two indicators, I had very low confidence in the forecast. Looking back, the SAI was a better predictor of the winter AO and NAO and because of their strong relationship to European weather, also a better predictor of the mild winter across Europe. Still I do believe that the high SCE contributed to a very active WAF season last winter that weakened the polar vortex and resulted in a cold winter in the US. Below I have included a plot of the WAF anomalies from October 1, 2013 through March 31, 2014. And you can see how active the WAF is from the end of December through the end of March. I have studied WAF anomalies very carefully and this is unique in the records since at least 1950.

Vertical wave Activity Flux

The very active and prolonged WAF continuously perturbed the polar vortex causing it to elongate repeatedly throughout the winter. And where the vortex was elongated or stretched out, that is where there were repeated Arctic outbreaks. I think the very active WAF and the continual perturbing of the vortex resulted in the cold winter across the US to be so persistent. You can see an animation of the polar vortex from last winter from the following link:


The contours are the full values of the geopotential heights at 10 hPa and the shading are the anomalies. In the free atmosphere the height anomalies are an excellent proxy for temperature anomalies. You can see in the animation how the cold or blue anomalies continuously extend over the US throughout the winter; in contrast, Europe is persistently in the red or warm anomalies.


After last winter, one lesson for me was not to discount the SCE in favor of the SAI. The SAI has been so skillful at predicting the winter AO that I have started to neglect the SCE in making winter forecasts. I won’t do that again.


Question 9. Peings et al. (2013) corroborated the high correlation between the SAI and the winter since the 1970’s, but earlier in the century they found the correlation to be much lower and therefore they argue that the relationship is changing. They suggest the relation between Siberian October snow cover and following winter AO is possibly just temporary. What is your opinion about these findings? And what is your view in general about the relationship of the Siberian October snow cover and winter AO for the future?


There was a recent paper that argued that there are large and significant errors in our satellite observations of snow cover extent in the most recent period.. If there are large errors in the recent snow cover data then how can we trust snow cover observations from before 1950 and when we did not have satellite to provide extensive sensing of snow cover? I cannot tell you that Peings et al.’s findings are wrong but I don’t lose any sleep over it. It is plausible that the snow-AO relationship is non-stationary as they argue but what is most important is that the relationship continues to be very robust and over the course of my career has been incredibly stable. In the short term I don’t anticipate that the relationship will change. Still, I feel that we have made much progress in understanding the dynamics of the winter circulation and, even if the relationship between snow cover and the AO broke down, we could adjust for that.


Question 10. The SAI and SCE are based on October snowfall in Siberia, which in turn is the result of particular (regional) atmospheric patterns. Might it be relevant to look for particular air pressure patterns for October, in order to find clues for the winter AO & NAO? We wonder if you and/or AER are exploring this field. A group of Italian weather amateurs have developed a new index, the OPI, which they presented last year. We believe you visited the OPI creators this spring in Italy. Can you tell us more about these developments?


As I discussed earlier, we do use an index of the lower tropospheric pressure patterns in our winter forecast. The theoretical basis for the predictor goes as far back as Cohen et al. (2001) and Cohen et al. (2002) and is further supported by Cohen and Jones (2011b) and Cohen et al. (2014).


You certainly have done your homework if you know that I met with the Italian scientists that are developing the OPI. For now there isn’t much for me to say. The idea is interesting but still needs to be vetted and I hope that we can collaborate on this potentially significant advancement in seasonal prediction.


Question 11. In your work you focus on the AO. In Europe, scientists (e.g. UK Met. Off.) are more concerned with forecasting the NAO. We wonder if the SAI is important for the winter NAO as well? Based on Brands (2013): ‘Is Eurasian snow cover in October a reliable statistical predictor for the wintertime climate on the Iberian Peninsula?’ we consider that the SAI might be relevant for parts of Europe as well, especially southern and northern Europe. Do you think that the SAI is useful for northwestern Europe as well?


Brands (2014) showed that the SAI is as highly correlated with the NAO as it is with the AO. Brands (2013) showed that the SAI has positive skill predicting surface temperatures for many parts of Europe including northwestern Europe. Finally, Brands et al. 2012 also showed that the SAI even provides skillful forecasts of precipitation for parts of Europe, something that the GCMs have had no success with whatsoever.


I have attached the skill of the SAI from temperature hindcasts (cross-validated) using just the SAI (on the left, only positive skill is shown). The SAI has skill across northern and southern Europe. On the right I show the same skill for our model and it has positive skill over most of Europe. This is a significant improvement of the skill compared to the GCMs that have negative skill for most, if not all, of Europe. I also did share with you our individual hindcasts and real time winter temperature forecasts for Europevi. I think that you will agree that the hindcasts and forecasts are impressive. But what is probably most remarkable is that the real-time forecasts have been more skillful or accurate than even the hindcasts! As far as I know that is unique in climate forecast modeling where the opposite is true and is to be expected.

Anomaly Correlation of Snow Index and DJF T, 1997-2010ACC of AER forecast model and DJF T, 1997-2010

Question 12. In ‘Linking Siberian Snow Cover to Precursors of Stratospheric Variability’ (2014) you differentiate between low and high snow cover years. In your presentation at MIT in November 2013 you used the winters of 1988/89 and 2009/10 as examples of low and high snow cover years.vii These years are very well defined and have a high rate of predictability. Many years however don’t have such an extreme (negative or positive) snow cover, meaning that the SAI or SCE are close to their normal values. Take for example the winter of 1962-63, the coldest for many years in large parts of Europe. We reconstructed from your presentation at MIT a quite normal SCE for October 1962.viii It appears that SCE doesn’t always apply. Could it be that in October 1962 the SAI was above normal? And how do you cope with standard SCE and more importantly SAI years, bearing the six steps in mind? Put simply, is a moderate value for the SAI a prelude to a moderate winter AO?

An image of the Judah Cohen presentation in November 2013

An image of the Judah Cohen presentation in November 2013.


As I think of it, the SCE/SAI first and foremost is best as a predictor of the winter AO. The usefulness of the SCE/SAI to predict temperature, precipitation or any other weather or climate variable is dependent on the relationship between the AO and the sensible weather variables such as temperature. If large temperature anomalies are not related to variability in the winter AO then I would not expect the SCE/SAI to perform well predicting large temperature swings not related to the AO.


I have attached the figure you show, updated through 2012/13 (and is the same as chart above).

DJF AO and Oct SCE Anomaly 1949-2012

The SCE, at least according to the graph, very accurately predicted the moderately negative AO during the winter of 1962/63. Again, according to the winter AO, there is no indication that winter temperatures across Europe of that year should have been exceptionally cold. To explain the extremely severe temperatures across Europe that winter, more is needed than simply to point out the negative AO. But as I discussed, SCE likely has large errors the further back in time you go and certainly before the satellite era, which includes October 1962. I would be hesitant to claim failure or success in correctly predicting the winter climate in the earlier part of the record and certainly before 1972.


Question 13. You write in your articles that SAI determines 75% of the winter AO variance, so 25% depends on other factors. Do you expect any improvement of the SAI? Or will you focus on the other 25%?


Using the proxy for the lower troposphere circulation, we can achieve even higher correlations with the winter AO when combined with the SCE/SAI, something we even showed as early as Cohen et al. 2002. Our focus is not trying to achieve perfect skill in predicting the winter AO. Though the AO is the dominant climate mode of the Northern Hemisphere, it explains 30% or less of the variance. Instead our focus is to predict temperature and precipitation variance not related to or dependent on AO variability. We feel that this has more potential in improving our winter forecasts than increasing our ability to predict the winter AO. The problem is that finding a boundary condition that has as much skill as SCE/SAI in predicting winter climate has been elusive.


Question 14. Professor Labitzke of the Institut für Meteorologie, Freie Universität Berlin gave a presentation in 2011 highlighting the importance of the QBO and of the Sun for understanding the variability of the stratosphere in the winter. See below.

In your work you stress the importance of the SCE and SAI and the six steps in the troposphere-stratosphere-troposphere for the winter AO. We assume that the power of the polar vortex depends on the Siberian snow cover in October. On the other hand Labitzke argues that the strength of the polar vortex depends on the ENSO, QBO and solar activity. Are these two theories of yours and Labitzke compatible? Is the SAI/SCE simply the result of the QBO, ENSO and solar activity or does the SAI/SCE account for 75% of the winter variance and QBO, ENSO and solar activity for the other 25%?


I am familiar with the ideas of Labitzke and colleagues on the QBO and I am familiar with the ideas that ENSO variability also influences stratospheric variability. Let me start out by saying that I am not an expert on the QBO or ENSO. However I do feel that my own threshold of what can be used to predict winter climate is higher than for most other scientists. I cannot be satisfied with simply demonstrating a statistically significant relationship in the observations or in the models. I have clients that are only interested in accurate temperature and precipitation forecasts and not interested in the state of the stratosphere. If you correlate ENSO, QBO and solar activity with surface temperatures, the relationships are all weak. Also with ENSO at first it was reported that El Niños favour stratospheric warmings. Now more recently it has been shown stratospheric warmings are as likely during La Niñas as El Niños, therefore relying solely on ENSO for predictions of temperatures in the extratropics is very difficult. Take for example the back-to-back winters of 2009/10 and 2010/11. The surface temperature anomalies for both winters were nearly identical with widespread cold across the mid-latitude continents, yet those two winters were of opposite ENSO and QBO phases. I included a provocative advertisement of a poster that I presented at EGU, which probably didn’t help with my popularity. What were common to both winters were high October SCE and a negative AO. Using two winters as proof is overly simplistic but it is a dramatic example and there are other similar examples.

Poster showing NH Temperatures from the last two winters 

Question 15. In ‘Linking Siberian Snow Cover to Precursors of Stratospheric Variability’ you wrote: “Observational analysis shows that high SCE in October across Eurasia favours the building of high SLP across northern Eurasia in the following six weeks (from early November to mid-December). Simultaneously, low pressure is favoured in the northern Atlantic and Pacific Ocean basins forming a tripole SLP pattern preferentially favouring enhanced vertical wave propagation, particularly across eastern Eurasia and the North Pacific”.


Our readers can monitor October snowfall via Are there any other clues to the development of the winter AO that our readers can easily monitor? What advice can you give them on tracking developments in the Northern Hemisphere from October to December or even later?


The Rutger’s website is the best and easiest place to monitor SCE. Other enthusiasts can read my papers Cohen et al. 2014 and Cohen and Jones 2011b to try to understand tropospheric precursors to stratospheric variability and AO variability. My hope is to launch a blog where I discuss variability in the AO present and an outlook of anticipated weather and climate variability. In the blog I would hope to discuss SCE, the SAI, tropospheric precursors and all the related content as discussed here. It is only in the very early discussion stages but the hope is to launch it this year but the likelihood is still very uncertain and will require a lot of preparation.



Question 16. What can we expect in the future from you or AER?

We hope to continue to improve seasonal forecasting. We are also very much interested in sub-seasonal forecasting. We have experimented with a model that can predict the daily strength of the polar vortex one month in advance. We have had some success with it but we need more investment to get it fully operational. Given that large AO events in the stratosphere often precede large AO events in the troposphere, we have high expectations that such a model would have immediate and beneficial application to sub-seasonal temperature forecasts on the order of 4-6 weeks.

We would in addition be grateful if you could send us a listing of the annual SAIs from 1997 onwards for our article on this subject.



Question 17.  What's  your forecast of the winter 2014/15? Last year we saw lot's of cold in parts of the North America. What's your expectation of this year, will Europe and Asia get their share as well?

With the high values for both the SAI and SCE, I am expecting a negative AO this winter and our model forecast is consistent with the expectation of a negative AO winter.  The forecast is posted on the NSF website:

Cold air has become established in both North America and northern Eurasia conducive to Arctic outbreaks on both continents if the negative AO materializes this winter.

Question 18. This years SCE/SAI has been very high. According to your theory we can expect a negative (very) AO. How would you describe the developments after 31st of October?  Is everything going according to plan ?
There are things that have transpired as I would expect since the end of October and some that have not.  From the end of October through the third week of November, Eurasian snow cover stagnated for an unusually long period.  Not sure what to make of it but it does give me pause.  Also a persistent and deep polar low has developed in Northwestern Asia that is not typical of high snow cover years and I will discuss more about it in my answer to question 19.

Question 19. In your blog 14 November you wrote that the current atmospheric configuration is somewhat orthogonal to the mean atmospheric response to snow cover with the strengthening and expansion of the Siberian (high) different from composite  analysis based on high minus low snow cover. We wonder in what way the current (14 November)  configuration differ from the expected one and what  might be the cause of that? Was the Siberian high in November in the right place for changing the Jetstream and for WAFz?

The deep polar low in Northwestern Asia that developed in November has interfered with the expansion of the Siberian high.  The WAFz was active in November but I believe that it would have been even more active had the polar low not formed in that unusual location.  That is what I was referring to when we wrote that the atmospheric response is orthogonal to high snow cover extent in the in blog dated November 14.  In Cohen and Jones 2011 (on tropospheric precursors) and Cohen et al. 2014 (J. Climate) we provided plots of the atmospheric anomalies that are optimal for breaking down the stratospheric polar vortex and that are associated with high Eurasian snow cover.  The models have been predicting that low to slide off to the south and east closer to its climatological position and more favorable for further expansion of the Siberian high as shown in the above mentioned papers.  However today the models are predicting that the polar low will redevelop further west.   I am watching this development very closely.  If the low deepens in the location as the models predict and remains persistent that would destructively interfere with triggering further WAFz and weakening of the polar vortex, something more common with low snow cover extent.

Question 20. The United States and Canada experienced a massive cold outbreak in November. Lot's of snow e.g. in Buffalo. Was this outbreak related to SCE/SAI?

I would not say that the cold air outbreak in North America was forced by the high snow cover across Eurasia but was associated with it.  The rapid advance of snow cover in November contributed to very cold temperatures in Siberia in October.  So when the cold air mass in Siberia was dislodged and traveled over the Pole into North America it was unusually intense for this time of year.  We posted on our Arctic Oscillation blog:

A temperature anomaly animation that was created by compositing high snow cover years minus low snow cover years from September through February.  You can see in the animation a strong Arctic outbreak into North America in mid November.  So I do think that a rapid advance of Eurasian October snow cover and an Arctic outbreak into the United States a couple of weeks later are related.


In the article we used the detrended index but here are both:

1997-2013 SAI

[-0.41, -0.12, -0.68, 0.68, -0.07, 0.62, 0.16, -0.55, -0.37, -0.50, -1.44, -0.12, 2.30, -0.50, -0.17, 2.24, -1.07]

[-0.16, 0.10, -0.50, 0.83, 0.05, 0.72, 0.22, -0.52, -0.37, -0.53, -1.50, -0.21, 2.17, -0.66, -0.36, 2.02, -1.31]





Brands, S., R. Manzanas, J. M. Gutiérrez, and J. Cohen, 2012: Seasonal Predictability of Wintertime Precipitation in Europe Using the Snow Advance Index, J. Climate, 25, 4023-4028.

Brands, S., 2013: Skillful Seasonal Predictions of Boreal Winter Accumulated Heating Degree-Days and Relevance for the Weather Derivative Market, J. Appl. Meteor. 52, 1297-1302.

Brands, S., 2014: Predicting average wintertime wind and wave conditions in the North Atlantic sector from Eurasian snow cover in October, Environ. Res. Lett. 9, 045006 doi:10.1088/1748-9326/9/4/045006.

Cohen, J. and D. Entekhabi, 1999: Eurasian snow cover variability and Northern Hemisphere climate predictability, Geophys. Res. Lett., 26, 345-348.

Cohen, J., K. Saito and D. Entekhabi, 2001: The role of the Siberian high in Northern Hemisphere climate variability, Geophys. Res. Lett., 28, 299-302.

Saito, K., J. Cohen and D. Entekhabi, 2001: Evolution in atmospheric response to early-season Eurasian snow cover anomalies, Mon. Wea. Rev., 129, 2746-2760.

Cohen, J., D. Salstein and K. Saito, 2002: A dynamical framework to understand and predict the major Northern Hemisphere mode, Geophys. Res. Lett., 29(10), 10.1029/2001GL014117.

Gong, G., D. Entekhabi and J. Cohen, 2003: Modeled Northern Hemisphere winter climate response to realistic Siberian snow anomalies, J. Climate, 16, 3917- 3931.

Cohen, J., and M. Barlow, 2005: The NAO, the AO, and Global Warming:  How Closely Related? J. Climate, 18, 4498–4513.

Cohen, J., and C. Fletcher, 2007: Improved Skill for Northern Hemisphere winter surface temperature predictions based on land-atmosphere fall anomalies, J. Climate, 20, 4118-4132.

Cohen, J., M. Barlow, P. Kushner, and K. Saito, 2007: Stratosphere-Troposphere coupling and links with Eurasian Land-Surface Variability, J. Climate, 20, 5335–5343.

Cohen, J., M. Barlow, and K. Saito, 2009: Decadal fluctuations in planetary wave forcing modulate global warming in late boreal winter, J. Climate, 22, 4418–4426.

Cohen, J. and J. Jones, 2011a: A new index for more accurate winter predictions, Geophysical Research Letters, 38, L21701, doi:10.1029/2011GL049626.

Cohen, J. and J. Jones, 2011b: Tropospheric precursors and stratospheric warmings, J. Climate, 24, 6562-6572.

Cohen, J., J. Furtado, M. Barlow, V. Alexeev and J. Cherry, 2012: Arctic warming, increasing fall snow cover and widespread boreal winter cooling, Environ. Res. Lett., 7, 014007 doi:10.1088/1748-9326/7/1/014007.

Cohen, J., J. Furtado, J. Jones, M. Barlow, D. Whittleston and D. Entekhabi 2014: Linking Siberian snow cover to Precursors of stratospheric variability. J. Clim., 27, 5422-5432.


28-11-2014 | WS_Vrij_Diversen | 1325
  • 17/5 09:38 John:
    Oss nu een beetje regen. De vijfde dag "al" deze maand met neerslag, maar het maandtotaal staat nog op 1,8 mm...
  • 16/5 19:00 Hans:
    Enschede-W: 0.8 mm regen en om ongeveer 15.00 de maxtemp van maar liefst 27.2 in de bebouwde kom. Temp na de regen gezakt naar 18.4 momenteel.
  • 16/5 17:02 Erik:
    Erg tegenvallend vandaag. Hier in Bennekom Noord is het vanmorgen net even nat geworden maar daar bleef het bij. Uiteindelijk nog wel 2mm in de regenmeter. Tot mijn verbazing is heeft Sylphidestation Bennekom ZO (ruim 1km naar het zuiden) 5,2mm gemeten. Lokaal flinke verschillen dus.
  • 16/5 14:29 Bert:
    Het is vrij snel voorbij getrokken met slechts wat druppels.
  • 16/5 14:22 Bert:
    In Eerbeek rommelde het wat in het zuiden maar aan de lucht te zien lijkt het weer op te lossen.
  • 16/5 12:10 John:
    Het rommelt en dondert best lekker in Rijen. Regen wil niet echt lukken. Zwaartepunt ligt westelijk van Rijen.
  • 16/5 11:38 Herman:
    Wouw Om 11:30 was de onweerbui voorbij en liet 3,5 mm regen achter.
  • 16/5 11:17 Herman:
    Wouw Om 10:58 de eerste donder en om 11;13 de eerste regen met matige westelijke wind
  • 16/5 09:19 Johan:
    Het regent in Ede Veldhuizen, 18 Gr Celsius
  • 16/5 00:03 Jaap:
    Hier in De Bilt nipt de zomerse dag gehaald. De bewolking bleef beperkt. De droogte wordt nu merkbaar. Neerslagtekort al zo groot, dat een paar onweersbuien het niet zullen oplossen. Een lange herfstige periode is nodig, maar komt die er nog? Ja, in oktober, maar dan is het te laat.
  • 15/5 19:11 Pieter:
    Veel zon en max 24 C , door de stevige wind voelde niet niet warm aan .
  • 15/5 16:52 Johan:
    Eerste zomerse dag in de Bilt
  • 12/5 06:50 Hans:
    Enschede-W: gisteravond laat na een mooie dag (24.3 als maxtemp) een korte stevige bui die 2.5 mm bracht. De spreekwoordelijke druppel...
  • 11/5 20:39 Ted:
    Heerlijk. Even compleet herfst. ZW-5 tot 6, sinds 19.40 soms ZW-7, mat soms een vlaag van kracht 8. Enkele takken breken zelfs af. Af en toe lichte regen, maar sinds 10 minuten zet de regen flink in. Koud. Brrr.
  • 11/5 18:58 Pieter:
    Na een paar uurtjes zon trok de lucht dicht ,en nu een beetje regen. Met een harde kille wind is het zomergevoel weer aardig verdwenen
  • 10/5 19:19 Hans:
    Enschede-W: volle 26 graden als maximum! Zwoel-warm met veel ZZW-wind. Veel bewolking maar een buitje heb ik tot nu toe niet gehad.
  • 10/5 18:20 Ted:
    Tot nu toe slechts 0,7 uur zon. Max.temp. 21,5 graden om 13.30 uur. Nu (18.05 uur) nog maar een zielige 14,7 graden. Wat regen rond 17.30 uur. Neerslag 0,2mm.
  • 10/5 18:09 Pieter:
    Weinig zon vandaag, maar wel een krachtige en zwoele zuiden wind en ca 20 C , later een klein rbeetje regen en dalende temperaturen
  • 10/5 13:08 Ted:
    Het zweetseizoen is begonnen. Klef bij 21,5 graden. WZW-5 tot 6 met vlagen van 7.
  • 10/5 11:38 Hans:
    Enschede-W: zachte meinacht met 13.1 als minimum. Vanaf vanochtend stak een vrij warme ZZW-wind op met vrij veel bewolking. Actueel 22.9 al!
  • 10/5 09:56 Johan:
    Ede Veldhuizen zwaar bewolkt, 18,3 Gr Celsius
  • 9/5 20:54 Hans:
    Enschede-W: met 25.3 de eerste zomerse dag vandaag! Actueel fraaie merelzang en wat overtrekkende wolkensluiers. Nog erg zacht met 19.8 graden.
  • 9/5 19:02 Ted:
    Eerst fraai zonnig. Nu forse hoge bewolking. Bisamratten knagen krachtig aan het hogedrukgebied. Regen komt nu al boven de westelijke Noordzee.
  • 9/5 17:37 Pieter:
    Eeste zomese dagen in Nederland met 25,1 C in Ell en 25,3 C in Eindhoven, de Bil met 22,7 C nog geen zomerse dag
Bekijk het archief
Di 17 mei 2022
Bezoekers online
Er zijn 2 gasten en 3 leden aanwezig: Bonne, Dick, Harma