Teleconnections and long range forecasting

[BornFromTheVoid]

Thought I'd best get this one finished before I start back in Uni tomorrow. This post turned out much longer than I'd anticipated, so apologies for that!

In this piece, I'm going (to attempt) to have a look at the relationship between Arctic Sea Ice Extent and the usual Winter CET.

A little on Arctic sea ice.

The Arctic sea ice grows to a maximum extent (extent measured as ocean surface covered by at least 15% ice) of around 16 million km2 each March before reducing to an average of 6 million km2 in September. In recent years, the maximum has declined by a small amount (Figure 1) while the minimum has been experiencing an accelerating decline (Figure 2)

Fig 1: Arctic Sea Ice Extent Maxima from 1979-2012

Average... 79-00... 15.9 million km2

Average... 01-07... 15.2 million km2

Average... 08-12... 15.1 million km2

Fig 2: Arctic Sea Ice Extent Minima from 1979-2012

Average... 79-00... 6.7 million km2

Average... 01-07... 5.6 million km2

Average... 08-12... 4.4 million km2

The rapid loss during for the minima is what most likely influences the weather. As the sea ice (especially when covered in snow) has a very high albedo (reflectivity), the sunlight during the 24 hours in summer is mostly reflected back into space. In recent years the, the ice has been reduced by several million km2, allowing large portions of the Arctic Ocean (which has a very low albedo) to absorb that solar energy and accumulate a lot of heat (which melts the ice even more).

When the Arctic enters Autumn and the air begins too cool, the energy which has built up in the ocean surface gets released into the air as heat, causing very large +ve surface air temperature anomalies in Autumn and early winter (Figure 3).

Fig 3: Surface Air Temperature for Sep 1st to Oct 31st, 2007-2011

This warming has several possible ways it can effect the weather in the northern hemisphere mid-latitudes.

(i) The increase air temperature causes the air to expand upward, increasing the 500mb geopotential height during Autumn and Winter. This reduced the meridional (south to north) atmospheric thickness gradient, which causes the jet stream to have much higher amplitude waves and move slower. This causes slow moving or "stuck" weather patterns, as well as the higher amplitude waves carrying warmer air further north, and colder air further south than normal.

More details can be found in the study by Francis and Vavrus here

(ii)Another possible mechanism, relates again to the increased air temperatures in the Arctic and Siberian Autumn snow cover. The rate of increase of Siberian snow cover south of 60N (the SAI) during October is strongly related the the AO phase in winter. As you can see in figure 3, most of the very high temperature anomalies occur along the Eurasian coastline area, where the ice has melted back. This access to newly open ocean will allow the atmosphere to hold much more moisture, but the +ve temperature anomaly may be enough to reduce snowfall in September and early October, Later in October, while the air will still be milder and more moist than usual, the temperature may be low enough anyway to allow for snow to fall, causing a rapid increase in snow cover during October, giving a high SAI value, resulting in a -ve AO pattern later in Winter.

More details and the possible mechanisms for the SAI/Siberian snow cover and AO link can be found in these papers

http://web.mit.edu/j...dsetal_JC12.pdf

http://web.mit.edu/j...Jones_GRL11.pdf

http://web.mit.edu/j...henetal2007.pdf

Despite the above, there are several issues with with correlating sea ice changes and winter weather here. The main one is the short number of year. Consistent very reliable sea ice data only spans from 1979 to present, which doesn't offer up a long time for comparisons. Included with that, is the fact that the sea ice may only have reduced enough in the last decade for the mechanism described earlier to become clear, leaving even less time for correlation. We also have to consider the moderating influence of the open ocean, especially to the north and north west of Scandinavia. With so little time, the interference of the -ve PDO and the low solar activity, there is a lot of noise in the data, but I'll give the correlations a go anyway!

Right so. To take away the warming of the CET since 1979, the temperature data was detrended. Now I used the NSIDC sea ice extent data in the links in the opening post of this thread. From that, I used the September minimum extent to correlated with the winter CET.

The first correlation (Sea ice minimum with detrended winter CET), gave a weak but positive correlation of +0.13 (more ice=higher CET), which isn't statistically significant though, given only 33 years to compare with. Below is a graph of the Winter CET detrended and the sea ice extent minima.

All averages and lags I tried made the correlation even lower than +0.13. Correlations just using 10 and 15 year groupings did improve the correlation, especially before 1998 (the 1987-1996 correlation was +0.65, but still not passing statistical significance with so few data points.)

The one method that did generate a statistically significant correlation, of +0.41 significant at p<0.2, was when I also detrended the sea ice extent minima, and then performed the correlation. The relationship between the 2 can be seen in the graph below.

So it seems there is some relationship present between the SIE minima and the following Winter CET

To see what kind of winter pattern follows low sea ice years, I picked the lowest detrended sea ice years, as well as the lowest without the detrending, and created some composite maps.

Lowest Detrended SIE Minima, 500 hPa Geopotential Height Winter Anomaly

Same as above, but based on actual SIE minima, not detrended

A clear signal for northern blocking on both those charts, but hints at the ridges being centred more towards Scandinavia, suggesting more of a easterly or northeasterly component to the upper flow. Perhaps a signal for a stronger Siberian high pressure extending west into Scandinavia?

[EDIT: For comparison the high sea ice years.

Both with a strongly +ve AO and NAO, the opposite of what you want for a cold winter.]

In conclusion, there does seem to be weak correlation between lower sea ice minima in September and below average winters in the CET zone. With the short timespan available for correlations and the rate of change in the sea ice observed, the strength of the correlation is not as strong as it might be if we had a longer time series. As it is though, it would appear to be another factor that's leaning towards a colder winter for the UK, along with the -ve PDO and weak ENSO signal

Edited October 2, 2012 by BornFromTheVoid

[bobbydog]

another excellent post mate! so, as we have discussed before, less sea ice can lead to colder winters and more snow, especially in the areas most important to colder snowier UK winters (in simple terms).

as you have shown, it leads to a warmer arctic, which reduces the strenth of the PV. funnily enough, i was looking into 'rossby waves' last night, (meanders in the jet stream- for anyone who's wondering) many of the points you have raised were mentioned, in particular, the part about reduced temp gradients and stuck weather patterns (northern blocking being the one we want!)

yet another piece of the jigsaw in place! (it all goes wrong when mother nature loses a piece down the back of the sofa!)

keep up the good work and don't let uni get in the way of more important things like this!

[muppet]

Here's a question. How do you know when you have a good model fit?

I looked at regressing CET with data from the previous one and two months for AO and also geo pot thickness 1000-500hPa. From 1950-2012 I got an Rsq of 94% but an average absolute error of 1.1C.

Is this any good? If not, what is good?

Thanks.

Edited October 3, 2012 by muppet77

[bobbydog]

Global average temperature time series

interesting..... whilst not looking for anything in particular, i came across this last chart on the met office website.

i previously mentioned the apparent correlation of the first two, now i couldn't help noticing the similarity of the temperature graph, with regard to timings. if we were to smooth the graphs, i think we may find that the temp graph is a very similar opposite to the others. considering that the first two may be reversing their trend, could this mean that global temps will start to fall? maybe this is a natural cycle?.... maybe.....

[ZONE 51]

I've taken two of the above charts(global average temps+arctic sea ice extent) and done some chopping and merging. What we can see is that as global temperatures go up then the arctic sea ice extent becomes less.

Please note that the global temperature chart spikes are not as tall as the original chart, this is only a guide and not an exact merge.

Edited October 5, 2012 by ElectricSnowStorm

[BornFromTheVoid]

I've taken two of the above charts(global average temps+arctic sea ice extent) and done some chopping and merging. What we can see is that as global temperatures go up then the arctic sea ice extent becomes less.

Please note that the global temperature chart spikes are not as tall as the original chart, this is only a guide and not an exact merge.

I think what's also interesting to note is how the sea ice extent didn't seem to react to global temperatures during the first half of the century.

This is where measuring with extent can be an issue, as the volume may have been dropping at the time (i.e., getting thinner) but the actual surface area covered may have stayed the same.

The Arctic temperatures certainly reacted to global temperatures, even in a more sensitive manner.

[bobbydog]

The Arctic temperatures certainly reacted to global temperatures, even in a more sensitive manner.

that appears to be true-

however, note the cycle of arctic temps and compare to the PDO- (its a shame the last 10 years of your chart are missing)

yet another rough correlation as regards timescale. if we put all these factors we have looked at together, we should be (if it is an interconnected cycle) entering a cooling phase

[BornFromTheVoid]

that appears to be true-

however, note the cycle of arctic temps and compare to the PDO- (its a shame the last 10 years of your chart are missing)

yet another rough correlation as regards timescale. if we put all these factors we have looked at together, we should be (if it is an interconnected cycle) entering a cooling phase

Have you a link to the first graph?

The overall trend for the PDO is flat, where as the Arctic air temperature anomaly is positive.

Here's a graph of surface air temperature anomalies for north of 70N, taken from here

The average monthly anomaly so far this year is +2.3C, with September being the mildest on record. And with SAT anomalies like this so far in October...

We would probably expect Arctic cooling given the PDO trend since the late 90s. But it's looking like the Arctic isn't following the PDO yet at least...

Edited October 6, 2012 by BornFromTheVoid

[bobbydog]

Have you a link to the first graph?

The overall trend for the PDO is flat, where as the Arctic air temperature anomaly is positive.

Here's a graph of surface air temperature anomalies for north of 70N, taken from here

The average monthly anomaly so far this year is +2.3C, with September being the mildest on record. And with SAT anomalies like this so far in October...

We would probably expect Arctic cooling given the PDO trend since the late 90s. But it's looking like the Arctic isn't following the PDO yet at least...

here's the link for the graph-http://www.appinsys....g/rs_arctic.htm

i haven't had time yet to read through the site but it looks like there's some interesting info on there

Edited October 6, 2012 by bobbydog

[bobbydog]

this is a better example, taken from that site-

"The following figure shows the AMO+PDO (black line above changed to red below) superimposed on the Arctic average annual temperature shown at the beginning of this document.

The above figures show the clear correlation of the Arctic temperature cycles to the oceanic oscillations – no CO2 connection."

[BornFromTheVoid]

here's the link for the graph-http://www.appinsys....g/rs_arctic.htm

i haven't had time yet to read through the site but it looks like there's some interesting info on there

I'm not too familiar with that site, but just careful reading through that. There are links to handy papers and data in it, but a lot of it seems to be mis-represented or mis-understood by the author. Lots of cherry picking from bits and pieces of different studies and many false conclusions drawn. Has a very strong denier/skeptic slant to it...

EDIT: This stuff is beginning to border on being more suitable for the climate and environment section.

Edited October 6, 2012 by BornFromTheVoid

[bobbydog]

I'm not too familiar with that site, but just careful reading through that. There are links to handy papers and data in it, but a lot of it seems to be mis-represented or mis-understood by the author. Lots of cherry picking from bits and pieces of different studies and many false conclusions drawn. Has a very strong denier/skeptic slant to it...

yes, i'm always suspicious when 'global warming' is mentioned but a lot of the info included in it seems useful.

[bobbydog]

EDIT: This stuff is beginning to border on being more suitable for the climate and environment section.

maybe, but it's all about finding the links between teleconnections to find the best 'ingredients' which go together to make a colder winter. i suppose we have to wander off on a tangent occasionally!

[BornFromTheVoid]

this is a better example, taken from that site-

"The following figure shows the AMO+PDO (black line above changed to red below) superimposed on the Arctic average annual temperature shown at the beginning of this document.

The above figures show the clear correlation of the Arctic temperature cycles to the oceanic oscillations – no CO2 connection."

Just as an example, the graph you posted there. They excluded the PDO + AMO data after 1998 because it went the opposite direction.

[bobbydog]

Just as an example, the graph you posted there. They excluded the PDO + AMO data after 1998 because it went the opposite direction.

i get what you're saying, they are not going to include data which doesn't support their cause but the chart does show a clear correlation before that time. any theories on why things changed so dramatically?

[BornFromTheVoid]

i get what you're saying, they are not going to include data which doesn't support their cause but the chart does show a clear correlation before that time. any theories on why things changed so dramatically?

I'd only guess that the AMO and the PDO are no longer the main influence.

Since the PDO turned -ve in 2007, the ice around the Bering sea has seen a recovery, but only in winter and spring, it still melts away completely by summer and doesn't begin to grow again until later in winter.

The AMO is still positive and is likely contributing to the very low ice levels in the Kara and Barents seas.

But every other part of the Arctic is showing a downward trend also, including places away from the known AMO and PDO influences.

Also, the sea ice was still thinning during the entire last 33 years, despite variations in the AMO and PDO affecting the ice extent

Edited October 7, 2012 by BornFromTheVoid

[bobbydog]

I'd only guess that the AMO and the PDO are no longer the main influence.

Since the PDO turned -ve in 2007, the ice around the Bering sea has seen a recovery, but only in winter and spring, it still melts away completely by summer and doesn't begin to grow again until later in winter.

The AMO is still positive and is likely contributing to the very low ice levels in the Kara and Barents seas.

But every other part of the Arctic is showing a downward trend also, including places outside of the away from the known AMO and PDO influences.

Also, the sea ice was still thinning during the entire last 33 years, despite variations in the AMO and PDO affecting the ice extent

so maybe 'global warming' (for want of a better phrase) is over-riding other factors. as nature tries to balance things out, i wonder what will happen to do so. we seem to be in uncharted territory now.....!

[BornFromTheVoid]

But I guess to avoid turning this into a global warming debate and detracting from things, we can leave it at simply... something else is having a large impact on the sea ice, and the current accepted scientific answer for that is rising global temperatures.

What will happen in the next few years? Nobody knows for certain. But there does appear to be a very clear trend, as is observable in the graph...

[Relativistic]

i get what you're saying, they are not going to include data which doesn't support their cause but the chart does show a clear correlation before that time. any theories on why things changed so dramatically?

I think it seems rather odd too, if CO2 was having an effect then shouldn't the temperature line steadily rise above the AMO+PDO line as an added warming effect? Why would it be only now that the warming CO2 effect is kicking into gear?

(Sorry if this is off topic)

[Methuselah]

I think it seems rather odd too, if CO2 was having an effect then shouldn't the temperature line steadily rise above the AMO+PDO line as an added warming effect? Why would it be only now that the warming CO2 effect is kicking into gear?

(Sorry if this is off topic)

Perhaps it would be, if it were not for all that latent heat being spent on melting the Arctic sea-ice? That heat has to be coming from somewhere...

[UV-RAY]

Perhaps it would be, if it were not for all that latent heat being spent on melting the Arctic sea-ice? That heat has to be coming from somewhere...

Oceanic heat content RP, it takes many years for latent heat to flush through the oceans system, well that's my take on it anyway.

[Methuselah]

Oceanic heat content RP, it takes many years for latent heat to flush through the oceans system, well that's my take on it anyway.

I think that we're in broad agreement there, 7...But I don't think that all of that melting can be coming from below? A difficult one...

[Optimus Prime]

Really interesting stuff, a great and informative read. It's a shame people tend to like posts who merely suggest a cold winter with lots of snow is on the way . This is actually very interesting using a scientific format.

In what is a very complicated area you're doing well to understand some of this!

[UV-RAY]

Really interesting stuff, a great and informative read. It's a shame people tend to like posts who merely suggest a cold winter with lots of snow is on the way . This is actually very interesting using a scientific format.

In what is a very complicated area you're doing well to understand some of this!

I agree it's a fantastic thread and would like to thank BFTV for all the hard work he's put into this, and everyone else whose contributed of course.

[ZONE 51]

Perhaps it would be, if it were not for all that latent heat being spent on melting the Arctic sea-ice? That heat has to be coming from somewhere...

Why is it only just melting the Arctic ice so much over the last few years, that i would like to know.

I have read some interesting articles on the Arctic melt, how this is to affect our weather is not exactly answered, and it cannot be as this is a new event, the melt has been happening over the years but the last year it's hit the record, there is most likely a delay of a few months to get the effects here, so this winter then we should find out, the ideas are that we see increased blocking, less westerly winds, with more cold winds due to the high pressures in the north and northeast. How the models react to this will be interesting.

Edited October 8, 2012 by ElectricSnowStorm