Given all the discussion about more favourable conditions for cold weather this winter I have been doing a bit of analysis of my own to try and get a better understanding of what the past might tell us about the probabilities for a warm / cold / average winter this year from a temperature perspective in the UK. This is not in any way intended to be a scientific analysis.
The table below summarises the results of my analysis. I have to a large degree only populated the data relevant to cold years in order to show more clearly what the various parameters were showing in those years. We can then draw our own conclusions as to whether the set up this year provides potentially favourable conditions for cold weather.
The first column in the table shows the QBO index figure for October. I have only populated those years where the figure was lower than -10 (i.e. in a strong easterly phase). I have also included the mean QBO index figure for December, January and February in the next column, focusing only on negative numbers.
The next column shows the OPI index (which is only available from 1976). Again I have only included negative years (the OPI purports to be a predictor of the winter AO index). I then show the ENSO index figure (the 3 month running mean for Aug, Sept, Oct), followed by the mean monthly number of sunspots for December, January and February.
The next column shows the October Eurasia snow anomaly and only includes years where the anomaly is positive by more than 1m sq km.
I then show the mean NAO and AO index figures for Dec, Jan, Feb, again focusing on negative years only. The next columns shows the winter CET mean and the Dec-Mar CET mean. I only include data for those years where the mean temperature was below the 1971-2000 mean. I then list for each year the individual months between Dec and Mar where the CET mean was more than 0.5C below the 1971-2000 mean.
I have also included the Nov CET mean figure, the EWP rainfall total for Oct, Nov and finally the dates on which the first cold snap of winter occurred (this is defined for these purposes as at least 3 consecutive days with a CET mean of less than 4C).
Commentary
El Nino Southern Oscillation (ENSO)
Firstly let me comment briefly on the ENSO (Nino) index. To my eye there seems to be very little correlation whatsoever between this index and the UK winter temperature. Cold winters occur when the ENSO is both positive and negative. So no real pattern at all there. The current 3 month running mean index for ASO 2014 is 0.2.
The ENSO is represented graphically here http://www.esrl.noaa.gov/psd/enso/mei/ts.gif
The rolling three monthly data for the index is shown here http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml
Snow cover
Snow cover figures for October are suggested to impact the weather patterns for the winter. The data I have covers the whole of Eurasia (from 1967). However there is no particular pattern based on the overall October snow cover anomaly. Indeed some of the largest positive anomalies occurred in 2006 and 2013 which both saw very warm winters.
Snow cover anomaly data for Eurasia in October can be found here http://www.ncdc.noaa.gov/snow-and-ice/extent/snow-cover/eurasia/10
Sea ice cover
Reduced levels of sea ice cover in the Arctic are expected to have a number of impacts on the weather. High levels of sea ice loss in the summer reduce the temperature differences between the poles and the tropics which tends, based on recent evidence, to reduce winter precipitation in Northern Europe. It may also delay the onset of winter.
Sea ice reflects most of the sun’s energy. With the loss of sea ice this will mean a much warmer arctic. A warmer arctic will tend to weaken the polar vortex and promote high pressure in the arctic region. When the polar vortex weakens sufficiently it can send cold air spilling southwards out of the arctic. Higher levels of sea ice cover tend to help maintain a strong polar vortex in the winter.
As with all other variables sea ice cover alone does not directly correlate with winter temperatures. There have been years with low sea ice cover and very warm winters such as 2007/8.
In 2014 the sea ice cover at the end of October was actually very similar to 2013 having been below 2013 for much of the month. At present ice cover on the Atlantic side of the Arctic is near average whereas it is well below average on the Pacific side.
Comparing sea ice cover extent over the last 60 years with winter temperatures is rather meaningless because sea ice cover has been gradually reducing over time and is now much lower than it was a few decades ago.
A chart of sea ice cover for October can be found here http://www.ncdc.noaa.gov/snow-and-ice/extent/sea-ice/N/10
A comparison of ice cover in recent years and other related news can be found here http://nsidc.org/arcticseaicenews/
Sea surface temperatures (SST)
This is another variable that is difficult to directly correlate with winter temperatures. The May SST is often used as a possible predictor of the winter NAO.
The SST anomalies for the north Atlantic in May 2014 can be seen here http://www.ospo.noaa.gov/data/sst/anomaly/2014/anomnight.5.29.2014.gif
If we compare that to a recent winter with consistent cold throughout (2009 / 10 being the best example) we can see some similarities with an area of cold water to the south of Greenland. But we don’t see the cold water off the coast of Portugal in 2009. http://www.ospo.noaa.gov/data/sst/anomaly/2009/anomnight.5.28.2009.gif
If we look at the position at the end of October we see 2014 shows a pool of cold water in mid Atlantic with very warm water off the Canadian coast and fairy warm water in Europe as well. That is not too dissimilar to 2013 although the cold was further south and further west in 2013. I don’t feel able to draw any particular conclusions about what the SSTs might be telling us for this winter.
http://www.ospo.noaa.gov/data/sst/anomaly/2014/anomnight.10.30.2014.gif http://www.ospo.noaa.gov/data/sst/anomaly/2013/anomnight.10.31.2013.gif
Quasi-Biennial Oscillation (QBO)
I now move on to the QBO which is much more interesting. The first thing to say is that a negative QBO does not in and of itself mean we will have a cold winter. Equally a positive QBO does not prevent a cold winter. So again the QBO by itself does not tell us a great deal.
However, one point of note is that when you have a winter with a negative QBO and a negative NAO and a negative AO then you will almost always get a cold winter (perhaps not surprisingly). There are 14 years in my table below (included within the years highlighted in blue) where the three parameters all line up and the winter CET is below the mean.
There is only one year where all three parameters line up and the winter CET was not especially cold (1997). However, the QBO was only marginally negative that year.
So we can I think conclude that when we get a significantly negative QBO and a negative NAO and AO we are almost certainly going to see a colder than average winter (some of the years above were only fractionally colder than average so it does not necessarily mean a very cold winter).
This year of course the QBO is currently heavily negative. The index for October is -23.9 which is the fourth lowest since 1950 and very similar to 2012.
You can find QBO data at the following link http://www.esrl.noaa.gov/psd/data/correlation/qbo.data
The QBO is also represented diagrammatically at this link http://www.geo.fu-berlin.de/met/ag/strat/produkte/qbo/qbo_wind_pdf.pdf
October Pattern Index (OPI)
This index purports to be a good predictor of the Arctic Index (AO) for the winter. Data is available back to 1976 and can be found here:
https://forum.netweather.tv/topic/81494-october-pattern-index-opi-monitoring-winter-season-2014-2015/?p=3051011
A negative index is suggested to predict a negative AO for the winter. If we look at recent years we can see from the table below that every year with an OPI less than minus 1 had a winter CET below average (some not much below average). That might suggest that the OPI is a good predictor of winter temperatures.
However the other point to note is that the NAO index was also negative in all of the years mentioned above which is also very important for cold blocked weather. There were some years with an OPI between 0 and -1 where the winter temperatures were not that cold. In those years the NAO was generally positive.
This year the OPI is expected to come in below -2C. So that could be helpful in promoting cold outbreaks during this winter.
Sunspot numbers
For Europe generally low solar activity tends to promote colder conditions. This can be seen quite clearly in the table below (which shows mean sunspot numbers averaged over the three winter months). Most of the years highlighted in blue where there were negative QBO, NAO and AO readings also had very low sunspot numbers. So low solar activity is clearly helpful.
But once again it is not absolutely necessary to have low solar activity to get a cold winter. 1958, 1967, 1968 and 1978 all had solar sunspot numbers averaging over 100 in the three winter months but still saw a cold winter.
In 2014 we are coming off a solar maximum. In recent months mean sunspot numbers have generally been around 70-80. There was a slight increase in September but the mean figure fell back in October to 60.6 (varying between 20 and 93 on the daily readings) which is somewhere between the low figures seen in many cold winters and the above 100 readings in a few cold winters.
So in summary I would say that the solar position at the moment is not strongly conducive to a cold winter but if the other parameters do all line up then it is clear that the level of solar activity at present is not high enough to prevent a colder than average winter.
Monthly mean sunspot data can be found at this link http://www.sidc.be/silso/INFO/monssninfocsv.php
Arctic Oscillation (AO)
This is linked to the discussion on the OPI earlier. A negative AO (particularly when aligned with a negative NAO) tends to promote blocking and hence drier and often colder than average conditions.
Looking at the table below we can see that almost all the years since 1950 where the winter CET has been below average saw a negative AO. The AO figures in the table represent the average AO index figure over the three winter months.
Only 4 out of 33 below average winters since 1950 did not see a negative AO (being 1982/3, 1983/4, 1990/1 and 2008/9). In all these 4 years the AO index was between 0.2 and 0.4 so only just above zero.
So a neutral or negative AO is pretty much a necessary condition for a cold winter but by no means a sufficient condition.
The current AO index for October was -1.1 so we are already negative. As noted above the OPI is likely to come in below -2 indicating the negative AO conditions may well continue through the winter.
The AO figure for last winter was 0.2 so fairly neutral. January was at -1 and February was 0. It was the run up to winter when the index was very positive representing the very stormy conditions we experienced with an index of 2 in November and 1.5 in December.
If instead we look at autumn 2012 in the run up to that cold winter the AO was at -1.5 in October and remained negative all the way through until March. So the fact that October 2014 was at -1.1 is certainly encouraging.
The AO index figures can be found at this link http://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/monthly.ao.index.b50.current.ascii.table
North Atlantic Oscillation (NAO)
This is a very important variable. Basically a positive NAO is occurs when there is low pressure over Iceland and high pressure in the Azores which is a typical winter set up. This gives us westerly winds and hence relatively mild conditions. When the reverse is true we can get an easterly drift.
Most of the below average winters in the table below have an NAO that is either neutral or negative. As for the AO the table shows the mean NAO index for the three winter months. Very cold winters (CET less than 3C) since 1950 all had a NAO index of less than -1. However an NAO less than -1 does not necessarily mean the CET will be close to or less than 3C. Nevertheless the warmest winter since 1950 when the NAO was less than -1 was 3.5C in 1963/4 which still 1C below average. So a strongly negative NAO almost certainly would give a colder than average winter.
Even a neutral NAO can still be sufficient to give a cold winter. 2012/13 is a good example where the NAO was exactly zero averaged over the winter as a whole.
If we look at cold winters in very recent times all have seen an October NAO close to or lower than -1C as follows:
2005 -1.0 2009 -1.0 2010 -0.9 2012 -2.1
For 2014 the October NAO has come in at -1.3. So that is interesting.
You can find the NAO data at the following link http://www.cpc.ncep.noaa.gov/products/precip/CWlink/pna/norm.nao.monthly.b5001.current.ascii.table
Temperatures in the Stratosphere
Cold temperatures in the Stratosphere (particularly around 30hPa (around 25km up) can help to maintain a strong polar vortex in the winter months. A strong vortex and a large temperature gradient between the poles and the tropics helps to maintain a strong jet stream with consequent mild and often wet weather for the UK.
This can be illustrated well if we look at data for last winter.
This chart shows the mean temperatures in the stratosphere for 2013. http://www.cpc.ncep.noaa.gov/products/stratosphere/strat-trop/gif_files/time_pres_TEMP_MEAN_ALL_NH_2013.gif
At the 20-25km height you can see a core of very cold temperatures from November onwards through into Jan 2014. There is absolutely no disruption to this core of cold at all and that is consistent with the non-stop train of Atlantic storms we saw last winter.
If the temperature contrast between the poles and the tropics is reduced then the jet stream is weakened and that can promote blocked conditions.
Another factor to consider is Sudden Stratospheric Warming (SSW) which can massively disrupt the temperatures in the stratosphere and then lower down in our atmosphere and sometimes results in an outbreak of cold weather. You can see on the chart above a SSW which occurred in early Jan 2013 for example.
This chart shows temperature anomalies. What you can see is that in winter 2013/14 there was a significant cold anomaly at all levels in the atmosphere which helped to promote a strong polar vortex. If you look at early 2013 the reverse is true with very large positive anomalies at the 20-25km level which certainly had an impact in the resulting bitterly cold late winter in the UK that year.
If we look back at 2009 when the cold weather started earlier in December we also see a significant positive temperature anomaly which started in late October.
http://www.cpc.ncep.noaa.gov/products/stratosphere/strat-trop/gif_files/time_pres_TEMP_ANOM_ALL_NH_2009.gif
So turning to the present and the current temperature at the 20km level is not as cold as one might expect. No sign of the purple colouring appearing as yet.
http://www.cpc.ncep.noaa.gov/products/stratosphere/strat-trop/gif_files/time_pres_TEMP_MEAN_ALL_NH_2014.gif
The anomaly chart confirms temperatures are currently above average although not by much. But certainly a very different position to this time last year.
http://www.cpc.ncep.noaa.gov/products/stratosphere/strat-trop/gif_files/time_pres_TEMP_ANOM_ALL_NH_2014.gif
Looking at the current forecast for the stratosphere we can see a fairly clear and stable core of cold temperatures in the polar region
http://www.netweather.tv/index.cgi?action=stratosphere;sess=
However if you run the animation the position does change quite a lot over the next couple of weeks. The cold core of temperatures is pushed towards Greenland and an area of very warm air develops over the Siberian side of the Pole. That might start to have an impact on surface conditions as we move into the start of December.
November Central England Temperature (CET)
As shown in the table below, the majority of years since 1950 with a below average winter CET saw either an average or below average November CET. However there are plenty of exceptions. So if 2014 sees a warmer than average November which seems likely then this is not necessarily a bad sign for the coming winter. A November CET between 7.7C and 8.7C has occurred 8 times since 1950 and been followed by a colder than average winter.
CET data can be found here http://www.metoffice.gov.uk/hadobs/hadcet/cetml1659on.dat
October / November precipitation (England and Wales Precipitation [EWP] series)
I have also looked at the precipitation figures for October and November prior to winters that were colder than average. I have particularly focused on those years highlighted in blue where a number of key parameters for winter prediction are all favourable.
What we see is that there is no particular pattern. It is about 50/50 between drier and wetter than average late autumn periods. A very wet autumn does not seem to preclude in any way a cold winter. It certainly looks like Oct / Nov 2014 will be very wet.
EWP data can be found here http://www.metoffice.gov.uk/hadobs/hadukp/data/monthly/HadEWP_monthly_qc.txt
First cold period of winter
Finally I looked at when the first cold snap of the winter occurred. This was defined for these purposes as three consecutive days with a CET less than 4C.
For the years highlighted in blue many saw the first cold snap around the middle of November. Others saw the first cold period at the end of November or right at the beginning of December.
Only 3 of the 19 years saw their first cold snap in the middle of December or later (1955, 1986 and 2009).
Summary
Based on the analysis above I am now fairly confident that the likelihood of a repeat of last winter temperature wise is virtually zero. There is a good chance of seeing at least some significant cold outbreaks this year possibly with sustained cold at times. The key parameter that I think will make the difference between close to average and well below average conditions is the NAO. Given how all the other parameters are lining up if we can get a neutral or negative NAO then we could well see a winter CET below 4C, possibly around 3.5C.
The closest analogue year looking at all the parameters does actually seem to be 2012/13. The QBO, OPI, ENSO, and sunspot data are all looking very similar this year to what they did two years ago. Worth noting that 2012/13 was at its coldest in the latter part of winter. December was average, January a fair bit below as was Feb but March was really cold. Maybe something similar this year?
1984/5 is also quite a good analogue across most of the parameters except sunspot data. Again that year the cold weather was focused on Jan – March.
We will have to wait and see what happens but I would be very surprised if we don’t see some decent cold spells during the course of this winter. We may have to be a bit patient though. The start of December is I think the earliest we might possibility see significantly cold conditions. But I am not currently expecting any severe cold this side of Christmas.
