Jump to content
Holidays
Local
Radar
Snow?

Archived

This topic is now archived and is closed to further replies.

Recommended Posts

Hi folks.

This thread is basically going to be for people to post up their analysis on how different teleconnections influence winter weather in the UK, so hopefully some of the more technically proficient members in here will contribute.

The pieces of analysis can then be used as a reference for anyone considering putting together a LRF. So for example, if they want to see how the PDO may influence the weather, they can check out the different analysis on it here before adding it to their forecast.Likewise with the NAO, AO, ENSO, etc.

This thread would also be useful just for learning more things about teleconnections in general, as well as sourcing data and helping with analytical and forecasting techniques and things of the sort.

A few rules to start off with.

-post your own analysis or links to other studies

-back up anything you do with facts and figures

-no general chat

-requests for particular analysis or data should be done through pm or on a more open thread (to make finding what you want in here easier)

Here's a few data sources for anyone that wants them

NAO

AO

PDO

MEI

Arctic Sea Ice Extent 78-10 &10-present

Antarctic Sea Ice 78-10 & 10-present

AMO

QBO

Sunspots

Global Annual Temperatures

CET

Daily Composite Charts

Monthly/Seasonal Composite Charts

Monthly Times Series

We'll see how this goes anyway. Perhaps after a while, the different pieces of analysis can be grouped together in a pinned thread for future reference. Thanks to bobbydog for the coming up with the idea for the thread and the title!

I'll start off by posting up the PDO and ENSO pieces I did in the winter thread and hopefully some others will join in.

Do feel free to tell me if I've mucked up somewhere, as I'm only learning this stuff and would like to learn from any mistakes!

Share this post


Link to post
Share on other sites

From the winter thread... PDO & Winter CET

As I mentioned at the end of last weekend, I started doing some correlations and stuff between some of the different teleconnections (MEI, PDO, etc), things like sea ice, sunspot numbers and of course, with the winter CET (Dec-Feb). I’ve gotten through quite a bit of it now, and thought I’d post up some results. So today will be a look at the PDO.

I’ll start off by saying that these are all very simple analysis, from someone who’s just learning the stuff, so don’t take it all too seriously and feel free to point out any mistakes I’ve made!

Anyway, the first things I looked at in detail were correlations between the PDO (Pacific Decadal Oscillation) and the Winter CET.

The PDO is defined by principal component analysis, and is recognised by a reverse "C" shape of -ve sea surface temperature anomalies in the north Pacific stretching from the Bering sea, down the US west coast and into the tropics when the PDO index is -ve, with the opposite occuring when the PDO index is +ve. Each phase appears to last several decades at a time, though going back before 1900, things are a little less clear.

Posted Image

http://jisao.washington.edu/pdo/

Data for the PDO can be found here http://jisao.washing.../pdo/PDO.latest

Annual PDO

Posted Image

The correlations I looked at were for PDO annual values and CET winter values from 1900-2011.

What is clear is that individual annual PDO values have little to no relevance when it comes to winters here. There is hardly any correlation at all, whether it be lagged, averaged over a few years, or whether the CET is de-trended or not. None of the correlations were significant.

Where the PDO does appear have an effect though is when you view it on a cyclic scale (in this case, as 30 year mean). There is then a clear relationship between a –ve phase of the PDO and an increased likelihood of colder than average winters. How one would wish to factor that into a long range forecast is up to them, but it would seem as though using the strength of the PDO at any given time is pointless, only the longer term phase that we’re currently in matters.

1 Year PDO & Winter CET ....................... 10 Year PDO & Winter CET ..................... 30 Year PDO & WInter CET

Posted Image Posted Image Posted Image

The reasons for this relationship? Well, plotting the winters following the lowest PDO value years, shows a reduction in upper zonal winds across the British Isles, as well as a tendency for stronger heights to the west and north of the UK, with very low heights across mainland Europe and into the Iberian Peninsula.

500hPa Heights, -ve PDO Years ........................... 300hPa Zonal Winds

Posted Image Posted Image

This basically indicates a slightly southerly tracking jet with an increased likelihood of northern blocking, which is what you want for cold spells. While other factors will be at play to cause these patterns, I think it's safe to assume that the -ve PDO phase will also contribute to them. The fact that there is no correlation on a year to year basis suggests that PDO's influence in the North Atlantic is easily over-ridden by other factors, and that it's influence can only be discerned by looking at the longer term anomalies.

So, the PDO, in my opinion, not much good for short range forecasts, but handy for working out decadal trends! So considering we're currently 5 years into a -ve PDO phase, it's likely a good sign for cooler winters!

Tomorrow, if time permits, I may have a look at the relationship of ENSO and our winters, by using MEI values.

Share this post


Link to post
Share on other sites

Also from the winter thread ENSO & WINTER CET (using MEI values)

So after looking at how the PDO may influence our winters, today I’m going to look at the influence of the El Nino Southern Oscillation (ENSO).

ENSO is characterised by periodic changes in the Sea Surface Temperatures (SSTs) in the equatorial Pacific. El Nino conditions are said to be present when the SSTs are anomalously high, as occurred in 1998 one of the warmest years on record (depending on the source, the warmest). La Nina represents the opposite, anomalously low SSTs, which have been more common and strong since the turn of the century.

Posted Image

ENSO variations can affect weather across the globe and its oscillations are among the primary divers in inter-annual global temperature variation.

There is loads of info on the net about how ENSO influences weather across the world. A simple google search will find all you need, so I won’t say too much about it, but as is nearly always the case, it’s changes in the jet stream from tropical convection variations that fuels most of ENSOs weather effects.

Here are a few links to get you started

http://www.esrl.noaa.gov/psd/enso/

http://www.elnino.noaa.gov/

http://kids.earth.na...nino/intro.html

http://en.wikipedia...._Oscillation

For the correlations with the winter CET, I used the Multi-variate Enso Index (MEI) which you can find more information on herehttp://www.esrl.noaa.gov/psd/enso/mei/

Each El Nino and La Nina is different from those that preceded it, with different strengths of anomalies and slightly different locations, all having slightly different effects on the atmosphere. The simple nature of this analysis doesn't really take that into account, and like the last PDO analysis, shouldn't be taken too seriously!

Anywho, lets get to it!

First off, here are graphs of the Annual MEI values and the de-trended winter CET.

MEI

Posted Image

Detrended Winter CET

Posted Image

I tried multipe variations with the MEI values for correlating with the winter CET, such as annual MEI, winter MEI and July-December MEI. All of these yielded low +ve correlations (all around ~0.1). This would indicate a very low influence, but with a slightly better chance at a below average winter with La Nina.

I then tried the same correlations, only this time I divided all the -ve MEI values by -1, to give me just +ve values so I could work out if the strength of the ENSO phase is more important than whether it's -ve or +ve.

Doing this improved the correlations by quite a bit!

The annual MEI value correlation (just looking at the strength and not the phase) is +0.27

The winter MEI correlation is +0.24

The July to December mean MEI correlation is +0.28

All this suggests that the strength of the MEI/ENSO values is more important than what phase it's in. So whether we have El NIno or La Nina, once it's weak, or even better, neutral, our chances of a below average winter are somewhat improved!

Here's a graph of the de-trended Winter CET and the July-Dec mean MEI value (all converted to +ve values)

Posted Image

Now, to get an idea of how the low to neutral ENSO phase effects the weather patterns over the northern hemisphere and particularly the North Atlantic region, I created a composite December-February, 500hPa geopotential height anomaly chart (which basically shows the upper level pressure patterns, and where high and low pressure are more likely to form at the surface) using the 20 lowest July-December mean MEI/ENSO values.

Posted Image

On the chart above, there is a very clear tendency for stronger heights over Greenland and to the north in general (tendency for a Greeny high and northern blocking). There is also a clear signal for lower heights over Europe and extending across to the British Isles, so increased likelihood of easterlies.

What about the jet stream?

Posted Image

The chart above shows a clear negative upper zonal wind anomaly right across the north of the British Isles and up into Scandinavia, with a clear positive anomaly going through Iberia and into the Mediterranean. This is a clear signal for a southerly tracking jet-stream once more.

So, in conclusion, it appears that once ENSO is weak, things look best for a below average winter.

So with the latest ENSO forecast suggesting neutral conditions for the remainder of the year, and with us currently in a -ve PDO phase, things are looking quite positive so far. Plenty more to consider though!

Share this post


Link to post
Share on other sites

i would like to create a graph of uk winter snowfall - basically to show how 'snowy' or not, any given winter has been over the past 100 years. obviously it could be difficult to gauge average 'snowiness' over the uk but i'm trying to create the equivalent of the CET, regarding snowfall. the idea is to attempt to overlay this graph onto the data BFTV is researching to see if there is any correlation.

if anyone has any stats or links which could help, i would be grateful.

Share this post


Link to post
Share on other sites

i would like to create a graph of uk winter snowfall - basically to show how 'snowy' or not, any given winter has been over the past 100 years. obviously it could be difficult to gauge average 'snowiness' over the uk but i'm trying to create the equivalent of the CET, regarding snowfall. the idea is to attempt to overlay this graph onto the data BFTV is researching to see if there is any correlation.

if anyone has any stats or links which could help, i would be grateful.

Rutgers University have a data request page http://climate.rutgers.edu/snowcover/docs.php?target=datareq

You can request snow extent for specific areas going back to '66.

Share this post


Link to post
Share on other sites

I think the problem is that we can make all the measurements we like at this time of year but the best we can do is to make an educated guess at probabilities and quite often then Mother Nature throughs a spanner in the works and introduces something not taken into account.

The feeling in my water is that it will be an average winter but then most winters are average anyway - however I can confidently predict that if you do wish to see a layer of white in December, get in an aircraft during daylight hours, climb through the only layer of cloud on the day, a layer of strata cumulus, then look down and voila, a layer of white and if you climb high enough the temperature will drop below freezing :)

Share this post


Link to post
Share on other sites

Well, today I'm going to attempt some correlations between sunspots and the winter CET.

I'll be using annual sunspot numbers, based on the 12 month mean. The source is from the link to sunspot data in the opening post.

The sunspot data goes back to 1700, giving us plenty of years to find a correlation from.

The main issue with correlating sunspots and goings on in our weather, is that it isn't the sunspots themselves that influence the weather, but more so the ~11 year solar cycle they represent and the variations in the suns output, whether it be total solar irradiance, or fluctuations in a particular light wavelength.

It's long been thought that low solar activity is related to cooler temperatures on Earth and in particular, harsh winters in Europe, especially the UK. This is due to the low sunspot numbers observed during times such as the maunder minimum between about 1650 and 1715, a time when the UK experienced some very harsh winters, and when much of Europe cooled during the little ice age.

More recently, the surprise prolonged minimum experienced between 2007 and 2010 (and continued low solar activity) resulted in plenty of folk discussing the possibility of us entering another period of very low solar and sunspot activity, perhaps even a new grand minimum, with the associated severe winters. The return to cooler winters in 08/09 09/10 and 10/11 has been blamed by many on the sun, but with the rapid changes in the Arctic and the -ve PDO also occurring then, it's difficult to place blame on any one thing.

The latest research points the finger at UV variations as a contributor to cold winters in Europe.

Some papers for anyone interested

Are cold winters in Europe associated with low solar activity?

The solar influence on the probability of relatively cold UK winters in the future

Anyhow, on to the correlations. As with all of these, feel free to point out any errors, as I'm sure there are at least some!

With sunspots there are lots of averaging and lags that can be used, so I'll try a good few here.

First off, just a graph of the Winter (DJF) CET and annual sunspot numbers, which as you might expect, has little correlation (<0.05)

post-6901-0-66757700-1348934486_thumb.jp

Starting from here, I tried loads of different correlations, annual with a 1/2/5/11 year lag. 1/2/5/11 year averages with similar lags (detrending the CET data with the longer averages, as correlations may occur just because both solar and CET have upward trends) but very few produced a correlation coefficient higher than +/-0.1. ("1" being a perfect +ve correlation, 0 being none whatsoever, -1 being a perfect -ve correlation).

Here are some graphs from the ones I tried with.

11 Year Sunpot average and 1 Year Winter CET: Correlation +0.08

post-6901-0-40929900-1348935787_thumb.jp

11 Year Sunspot average and detrended 11 year winter CET with an 11 year solar lag: Correlation -0.03

post-6901-0-01135900-1348936500_thumb.jp

30 Year Sunspots and 30 Year Detrended WInter CET: Correlation: -0.08

post-6901-0-69222800-1348937297_thumb.jp

11 Year Sunspots and 1 Year Winter CET since 1900: Correlation -0.06

post-6901-0-00305600-1348939151_thumb.jp

11 Year Sunspots and 5 Year Average Winter CET since 1900: Correlation -0.12

post-6901-0-57434900-1348939474_thumb.jp

Same as above but with the CET detrended: Correlation -0.27

post-6901-0-79458300-1348939990_thumb.jp

So, what can we say from that? Well, over the long term, it seems there is little to little correlation. When the sunspot and CET data is averaged out for more than a decade, the correlation appears to turn weakly negative. This certainly goes against what I'd expected!

A negative correlation in this sense means the more sunpots, the lower the winter CET!

This relationship becomes stronger during since 1900, with increasing sunspot numbers from the lows of the 20s, up to the 60s, correlating negatively with a drop in winter CET values during the same time.

Perhaps, as suggested earlier, I've done something wrong! Or maybe in this case, the -ve PDO, -ve AMO and suspected increase in aerosols from pollution early to mid century caused more cooling here than the solar activity did warming? Or maybe internal natural variability just over-rides the solar influence occasionally? Perhaps I need some other measure such as total solar irradiance or UV output to get a better correlation? Hard to say.

Anyway, if anyone wants to suggest some more correlations for me to try, or tell me where I've messed up, please feel free to do so!

Share this post


Link to post
Share on other sites

Well, today I'm going to attempt some correlations between sunspots and the winter CET.

I'll be using annual sunspot numbers, based on the 12 month mean. The source is from the link to sunspot data in the opening post.

The sunspot data goes back to 1700, giving us plenty of years to find a correlation from.

The main issue with correlating sunspots and goings on in our weather, is that it isn't the sunspots themselves that influence the weather, but more so the ~11 year solar cycle they represent and the variations in the suns output, whether it be total solar irradiance, or fluctuations in a particular light wavelength.

It's long been thought that low solar activity is related to cooler temperatures on Earth and in particular, harsh winters in Europe, especially the UK. This is due to the low sunspot numbers observed during times such as the maunder minimum between about 1650 and 1715, a time when the UK experienced some very harsh winters, and when much of Europe cooled during the little ice age.

More recently, the surprise prolonged minimum experienced between 2007 and 2010 (and continued low solar activity) resulted in plenty of folk discussing the possibility of us entering another period of very low solar and sunspot activity, perhaps even a new grand minimum, with the associated severe winters. The return to cooler winters in 08/09 09/10 and 10/11 has been blamed by many on the sun, but with the rapid changes in the Arctic and the -ve PDO also occurring then, it's difficult to place blame on any one thing.

The latest research points the finger at UV variations as a contributor to cold winters in Europe.

Some papers for anyone interested

Are cold winters in Europe associated with low solar activity?

The solar influence on the probability of relatively cold UK winters in the future

Anyhow, on to the correlations. As with all of these, feel free to point out any errors, as I'm sure there are at least some!

With sunspots there are lots of averaging and lags that can be used, so I'll try a good few here.

First off, just a graph of the Winter (DJF) CET and annual sunspot numbers, which as you might expect, has little correlation (<0.05)

post-6901-0-66757700-1348934486_thumb.jp

Starting from here, I tried loads of different correlations, annual with a 1/2/5/11 year lag. 1/2/5/11 year averages with similar lags (detrending the CET data with the longer averages, as correlations may occur just because both solar and CET have upward trends) but very few produced a correlation coefficient higher than +/-0.1. ("1" being a perfect +ve correlation, 0 being none whatsoever, -1 being a perfect -ve correlation).

Here are some graphs from the ones I tried with.

11 Year Sunpot average and 1 Year Winter CET: Correlation +0.08

post-6901-0-40929900-1348935787_thumb.jp

11 Year Sunspot average and detrended 11 year winter CET with an 11 year solar lag: Correlation -0.03

post-6901-0-01135900-1348936500_thumb.jp

30 Year Sunspots and 30 Year Detrended WInter CET: Correlation: -0.08

post-6901-0-69222800-1348937297_thumb.jp

11 Year Sunspots and 1 Year Winter CET since 1900: Correlation -0.06

post-6901-0-00305600-1348939151_thumb.jp

11 Year Sunspots and 5 Year Average Winter CET since 1900: Correlation -0.12

post-6901-0-57434900-1348939474_thumb.jp

Same as above but with the CET detrended: Correlation -0.27

post-6901-0-79458300-1348939990_thumb.jp

So, what can we say from that? Well, over the long term, it seems there is little to no correlation on short time scales. When the sunspot and or CET data is averaged out for more than a decade, the correlation appears to turn weakly negative. This certainly goes against what I'd expected!

A negative correlation in this sense means the more sunpots, the lower the winter CET!

This relationship becomes stronger during since 1900, with increasing sunspot numbers from the lows of the 20s, up to the 60s, correlating negatively with a drop in winter CET values during the same time.

Perhaps, as suggested earlier, I've done something wrong! Or maybe in this case, the -ve PDO, -ve AMO and suspected increase in aerosols from pollution early to mid century caused more cooling here than the solar activity did warming? Or maybe internal natural variability just over-rides the solar influence occasionally? Perhaps I need some other measure such as total solar irradiance or UV output to get a better correlation? Hard to say.

Anyway, if anyone wants to suggest some more correlations for me to try, or tell me where I've messed up, please feel free to do so!

just a wild stab in the dark, but would ozone levels play a part in the effects of solar activity? i wonder if there is a chart for ozone over a similar period. maybe i'm talking rubbish but it was just a thought!

just did a quick search and found this chart-

Total Ozone and its relationship with Effective Stratospheric Chlorine

Posted Image

again, it may be rubbish but think of the harsher winters of the 60's to the 80's, (which come from memories more than stats!) then the seemingly milder ones of the 90's, up to the past few years of a trend to harsher winters. the graph initially stirs some interest. (with me anyway!)

i will keep looking at this angle for now and just add that this thread is not only for the standard type of research but also for 'thinking outside the box' ! - come on people, lets have some input!

Share this post


Link to post
Share on other sites

i thought this might be a good place to put my post from the winter thread-

the NAO is, as you say a reflection of the overall pattern. a negative NAO is what we get in the classic 'greeny high' situation. myself and BFTV have been looking at trends and causes of colder winters and have come up with a few ideas, built upon the findings of others, by trying to find links between various and alternative teleconnections (all credit goes to BFTV for the hard work and technical stuff)

an interesting link, still in research, is the link between polar ice extent and ozone levels. - they seem to correspond. as in the fall in ozone seems to match the timing of the fall in arctic ice levels. (it has been mentioned that less arctic ice, under favourable conditions, can lead to more snow in the N hemisphere) ozone levels have been on the rise again for the past few years. as chionomaniac has explained in the strat thread, more ozone leads to warming of the arctic strat, leading to a disrupted polar vortex. - a main ingredient for colder winters here. add this to the negative PDO (the coldest winters have generally occurred during the neg phase) and we have a few of the major 'building blocks' in place already. G.P. has said that the early signs are favourable for northern blocking this winter, i.e. negative NAO. there are many other factors which need to fall into place and there are never any guarantees. however, as of now, things are looking favourable for a cold (snowy??.....) winter. we shall see.....

Posted ImagePosted Image

just to show what i'm on about!

Share this post


Link to post
Share on other sites

What a fantastic thread. Can't wait until the next instalment.

I have looked at years when the PDO is negative and ENSO is neutral (like now) and the average Winter CET is 0.8C below the 30 year average.

What does de trended mean and how do you do it?

Share this post


Link to post
Share on other sites

What a fantastic thread. Can't wait until the next instalment.

I have looked at years when the PDO is negative and ENSO is neutral (like now) and the average Winter CET is 0.8C below the 30 year average.

What does de trended mean and how do you do it?

It's basically getting rid of the trend, in this case, the general rising temperature trend in the CET values. It's so you can can a better idea of the variability and the different influences on the CET, without the global warming effect kinda interfering and skewing things.

The method I used was from this http://flylib.com/books/en/2.22.1.97/1/

Just scroll down a little until you get to the "detrending a time series" section, where it's explained how to do it on excel.

Share this post


Link to post
Share on other sites

Great link. Thanks. That really makes sense.

Which factor are you looking at next?

I'm hooked.

Share this post


Link to post
Share on other sites

http://t.co/3vn9J9oB these give tabulated dam thickness. ⛄

I'm sure the closer to 5220 the more chance of snow.

Geopotential height of 500 mb correlated with a range of factors would be superb.

I may give one a go when time.

Share this post


Link to post
Share on other sites

Do feel free to contribute your own stuff, I like this to be a place where people can learn some analysis techniques (and hopefully I can pick up a few tips too!).

Myself and bobbydog are working on some stuff with regard the ozone changes, how that may have influence the AO, the sea ice, the polar vortex and essentially the weather here.

I'll be posting up some stuff on the Arctic sea ice and the statistical link the winter CET this week.

Share this post


Link to post
Share on other sites

No problem.

I will start correlating some stuff to geopotential 500mb height, which I think is a better indicator of snow potential than 2m air temp.

Share this post


Link to post
Share on other sites

The NOAA site does some superb correlation maps.

Here is geopotential 500mb height correlated with Northern Hemisphere snow cover.

post-15692-0-73106200-1349118833_thumb.j

Share this post


Link to post
Share on other sites

A couple more Winter correlations to Winter NH snow cover.

All three look interesting.

post-15692-0-47494600-1349119778_thumb.j

post-15692-0-32405200-1349119792_thumb.j

Share this post


Link to post
Share on other sites

Good stuff!

The 500hPa geopotential height is closely related to the source and temperature of the airmass, so the -ve correlation (lower geopot height=more snow) with snow cover there makes sense.

The same is true with zonal winds and surface air temperature, all to be expected. The difficulty arises in trying to work out what teleconnections cause the reduced zonal winds, lower surface air temperature and drop 500hPa geopotential height. That's where the long range forecasting part comes into play and the complex interactions of variables from all around the world!

Share this post


Link to post
Share on other sites

i found this article earlier- http://dotearth.blogs.nytimes.com/2010/12/28/putting-a-siberian-snow-connection-to-the-test/

i'm not sure how much weight it holds but its another interesting theory. the doubts i have with it are, that they are saying autumn siberian snow cover can affect the AO but surely the AO would affect the amount of snow in the first place. ( a "chicken and the egg" type situation)

i looked back at the last 3 years, covering 2 cold winters and a 'mild' one (bearing in mind its only really a small snapshot and would take a lot more work to get a more accurate picture) the thing i noticed, was it was not the amount as such but the placement of snowcover which seemed to have more influence. i posted some images in the snow and ice thread for comparison- http://forum.netweather.tv/topic/74146-snow-and-ice-in-the-northern-hemisphere-201213/page__st__400#entry2376270

Share this post


Link to post
Share on other sites

It's quite an accepted theory from J Cohens work and has been quoted a number of times over the last few years.

post-4523-0-90843800-1349122520_thumb.gi

Share this post


Link to post
Share on other sites

I think that AO plotted against geop 500hPa is worth a look when time.

Share this post


Link to post
Share on other sites

It's quite an accepted theory from J Cohens work and has been quoted a number of times over the last few years.

post-4523-0-90843800-1349122520_thumb.gi

From this paper it seems the rate of snow cover increase during October (SAI), rather than the overall snow snow cover extent, that correlates better with the AO.

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×