Met Office Hadley     15.1c.       Anomaly     -0.7c.    Provisional to  25th.

Metcheck                  15.18c.    Anomaly     -1.04c.

N-W                         15.61c.    Anomaly     -0.59c.

Mount Sorrel             15.40c.    Anomaly    -0.80c.

My  Mean                 15.9c.       Anomaly    -1.6c.

One thing I have noticed of late is that temperatures have become very volatile. Recently of course we have had some very warm months until this month. But not so long ago we had a run of very cold months.

In fact only one of the last 20 months has finished within 0.5C of the 1971-2000 mean. That is very unusual. The last time we had a longer run than this was 1985-1987. From August 1985 to August 1987 we had 25 consecutive months with only 1 coming in within 0.5C of the 1971-2000 mean. This was of course a very cool period. 8 months were more than 0.5C above the 1971-2000 mean but 16 were more than 0.5C below the 1971-2000 mean.

The stats below show some figures for recent years. As you would expect if we take the 1971-2000 period we get roughly 4 months each year within 0.5C of the mean, 4 months more than 0.5C warmer and 4 months more than 0.5C cooler. There is a slight bias towards the warmer months.

The period from 1997-2009 stands out as being much warmer with an average of 6.7 months per year more than 0.5C above the mean and only 1.5 months below with about 4 being average.

The period from 2010-2013 was of course much cooler and we see the warm months fall back to the 4.3 figure that is in line with the 1971-2000 mean. However there were 5.5 months on average more than 0.5C below the mean.

The last 2 years from Sept 2012 show a different trend again. This time we have a significantly higher number of warm and cold months with 5.5 and 5 respectively on average each year. Only 1.5 months per year on average were close to the mean.

If we look at the number of months that have seen a CET 1C or more higher or lower than the 1971-2000 mean there again has been an increase in volatility recently. Going back to 1960 we find that only in 5 years have there been 8 or more months in the year where the temperature has been greater than 1C or more either side of the mean. All of those years have occurred since 1989. Even 1962 and 1963 only saw seven months.

The years in question are 1989, 1995, 1999, 2006, 2011 and 2013. Both 1989 and 2013 saw 9 months with a temperature at least 1C warmer or colder than the mean. All the other years had 8 months.

So far in 2014 we have had 6 months out of 8 with an anomaly of 1C or more.

These anomalies have not increased due to a rise in average temperature as the table below shows. The middle column shows the average number of months with a temperature anomaly of 1C or more. The final column shows the mean CET for the whole period.

The mean temperature for the years 2010-2013 was very similar overall to the 1971-2000 mean but there were an average of 7.3 months with an anomaly of 1C or more compared to the mean of 5.6 months. The table shows that trend has been increasing gradually in recent years but seems to have accelerated significantly since 2011. The figure for 2011-2013 is 7.7 months. That is though largely down to two exceptional years falling close to each other (2011 and 2013). Both 2010 and 2012 saw only 6 months each. However, if 2014 also comes in at either 8 or 9 months that will be 3 years out of the last 4 with 8 months or more having 1C+ temperature anomalies which would be unprecedented.

So temperatures do seem to be becoming more volatile. Will this trend continue over the next few years?

I was interested in these stats and decided to do a similar analysis from the database of daily CET means rather than your analysis of monthly CET means to see whether a similar volatility could be detected on these.

I was surprised to see that the daily figure gave very different findings,

The 1st graph shows for each decade since 1801-10 what the standard deviation is between the actual day's CET mean and the 30 year rolling average for that same day.

The second graph shows the number of days/year on average in each decade the CET was either 5C higher or 5C lower than the average daily CET based on a rolling 30 year avergae for that date - The total of these two is also shown

Both these graphs are a measure of volatility on a day by day basis

Both show the 1940's being the most volatile of recent times based on daily data

For the current decade it shows most of the volatility has been generated by additional LOW temperatures rather than high even though the majority of months overall have had  higher than average CET on a monthly basis

This is because on a daily basis there are two very cold periods to report (Nov/Dec 2010 and March 2013) which contributed many records to the 5C lower volatilty whereas the warm periods have been long in duration but not very extreme so add a lot to the monthly averages but very little to the daily volatility

As always with stats it depends how you define the measurement !

Great work with the stats Lanky, but what does it all mean?

I've noticed the period from 1900-40 doesnt seem to have many colder than normal days? There wasnt a negative CET month from Feb 1895 to Jan 1940. Was that the longest period without a negative month?

Thanks Martin. I think your analysis does just reinforce the point that you can prove almost anything you want by cutting the stats in a particular way. But it is interesting to see that the daily stats do give a somewhat different picture.

One thing that can be quite dangerous when looking at volatility is taking different lengths of time and trying to compare them which is sort of what I was doing in a rather rough and ready way. Clearly when you look at averages over a longer period (30 years say compared to a handful of years) then the volatility is likely to be smoothed out somewhat. Looking at daily data compared to a rolling mean as you have is probably a more valid and rigorous approach to take.

What your last graph does show very well is a big flip in the trend we saw in the 1980's to 2000's when the number of anomalous warm days was steadily increasing and the number of cold days decreasing. So far this decade we have flipped very suddenly to a high number of cold days and fewer very warm days. This is of course as you say due to relatively short but very cold periods at the beginning and end of 2010 as well as early 2013. Although we have just had an extended period of warm weather it was never exceptionally warm unlike the cold conditions a few years ago. It has been persistent rather than exceptional warmth with very few days seeing an anomaly greater than 5C. 

In answer to kendalian's question I'm not sure there is any great meaning one can draw from this. I would not want to suggest there is any sort of clear trend or shift in weather patterns that can be read into this data. I was just interested in what appeared to be an increasing trend for regular and significant deviations from the long run monthly means. The fact that the daily data do not show the same trend suggests that temperature volatility has not really changed in any significant way.

The longest period without a negative CET month is indeed the period that you have pointed to kendalian.

The CET is unchanged today and stands at 14.92C by my calculations. Today's mean was 14.93C.

A small fall is likely tomorrow but then quite a rise is likely on Thursday. Not quite as warm then for the final 3 days of the month but still probably just warmer than the current mean and a bit warmer than was forecast a few days ago.

The sub 15C August mean is now looking less likely. Latest output suggests a final figure of 15.02C. But still too close to call.

 Latest output suggests a final figure of 15.02C. But still too close to call.

That would be a very annoying outcome, LOL!

Met Office Hadley    15.1c.     Anomaly    -0.7c.

Metcheck                15.10c.   Anomaly     -1.12c.

N-W                       15.58c.   Anomaly     -0.63c.

Mount Sorrel           15.30.    Anomaly      -0.90c.

My   Mean               15.5c     Anomaly      -2.0c.                  

As expected a chilly start this morning has resulted in another fall in the CET today down to 14.86C. It was even cooler than expected today.

No change in the forecast for the rest of the month. A warmish day tomorrow due to high minimums largely will push the CET back up and then very small increases through till Sunday.

The cooler day today does mean my prediction for the final figure is just back below 15C again at 14.98C.

So it's either going to be 14.9 or 15.0?

Met Office Hadley     15.1c.     Anomaly     -0.8c.     Provisional to  27th.

Metcheck                 15.16.    Anomaly     -1.06c.

N-W                        15.56c.   Anomaly    -0.64c.

Mount Sorrel            15.37c.  Anomaly    -0.83c.

My  Mean                 15.9c.   Anomaly     -1.06c.

Met Office Hadley   15.2c.      Anomaly     -0.7c.    provisional  to 28th.

Metcheck               15.19c.    Anomaly     -1.04c.

N-W                      15.62c.    Anomaly     -0.58c.     

Mount Sorrel           15.40c.   Anomaly     -0.80c.

My  Mean                15.9c.    Anomaly      -1.6c.

I've just had another look at the predictions table. Is this the worst batch of guesses since March 2013?

It is yes although actually only slightly worse than the December 2013 predictions. A lot of people went for cold conditions in December. Only Martyn got anywhere near close to the correct figure.

CET now standing at 14.98C by my calculations.

No change expected today and a very small fall tomorrow means I currently expect a final figure of 14.96C.

Met Office Hadley        15.2c.       Anomaly     -0.7c.    Provisional to  29th.

Metcheck                    15.21c.     Anomaly     -1.02c.

N-W                           15.64c.     Anomaly     -0.56c.

Mount Sorrel               15.42c.     Anomaly     -0.78c

My  Mean                    15.9c.      Anomaly      -1.6c.

JFF here are some pattern matching stats to get you thinking about late Autumn and Winter.

These are all based on years where the August CET has been between 14.7C and 15.7C and the July CET has been greater than 17C going right back to the start of the CET series.

There are a total of 25 matches. I am using the 1961-1990 mean for comparison to take out the effect of warming in recent years.

The following November CET was lower than the 1961-1990 mean in 18 of those 25 years. The average November CET in those 25 years was 5.6C compared to the 1961-1990 mean of 6.6C. The November CET was less than 5C in 9 of those 25 years.

The following December CET was lower than the 1961-1990 mean in 20 of those 25 years. The average December CET in those 25 years was 3.8C compared to the 1961-1990 mean of 4.7C. The December CET was less than 3C in 7 of those 25 years and less than 0C in 2 years.

Food for thought there.

If we look at the following January there is less of a trend as one might expect as we move further out. Only 13 of the 25 years had a CET less than the 1961-1990 mean. But there were some very cold Januarys including 2 with a negative CET including the coldest January on record in 1795.

February shows a similar trend with 13 years below average and a few that were very cold including one at 0.1C.

What is perhaps most interesting is that, of those 25 years, 19 had at least one winter month that was 1C or more below average. Only 2 years had all three winter months with an average or above average CET.

So the pattern matching would suggest that a repeat of the exceptional warmth of last winter is very unlikely. There is a very strong probability (>90%) of at least one winter month being below average and about a 75% chance of at least month being 1C or more below average. Just based on pattern matching of course which is in no way scientific.

After our run of very warm months maybe we are now going to move back into a period where we see a number of cooler months over the next 6 months or so perhaps interspersed with some warmer months.

My CET estimate now stands at 15.00C exactly up to today.

It may dip very slightly tomorrow to finish at 14.99C but it looks very likely that something close to 15.0C will be the final number.

JFF here are some pattern matching stats to get you thinking about late Autumn and Winter.

These are all based on years where the August CET has been between 14.7C and 15.7C and the July CET has been greater than 17C going right back to the start of the CET series.

There are a total of 25 matches. I am using the 1961-1990 mean for comparison to take out the effect of warming in recent years.

The following November CET was lower than the 1961-1990 mean in 18 of those 25 years. The average November CET in those 25 years was 5.6C compared to the 1961-1990 mean of 6.6C. The November CET was less than 5C in 9 of those 25 years.

The following December CET was lower than the 1961-1990 mean in 20 of those 25 years. The average December CET in those 25 years was 3.8C compared to the 1961-1990 mean of 4.7C. The December CET was less than 3C in 7 of those 25 years and less than 0C in 2 years.

Food for thought there.

If we look at the following January there is less of a trend as one might expect as we move further out. Only 13 of the 25 years had a CET less than the 1961-1990 mean. But there were some very cold Januarys including 2 with a negative CET including the coldest January on record in 1795.

February shows a similar trend with 13 years below average and a few that were very cold including one at 0.1C.

What is perhaps most interesting is that, of those 25 years, 19 had at least one winter month that was 1C or more below average. Only 2 years had all three winter months with an average or above average CET.

So the pattern matching would suggest that a repeat of the exceptional warmth of last winter is very unlikely. There is a very strong probability (>90%) of at least one winter month being below average and about a 75% chance of at least month being 1C or more below average. Just based on pattern matching of course which is in no way scientific.

After our run of very warm months maybe we are now going to move back into a period where we see a number of cooler months over the next 6 months or so perhaps interspersed with some warmer months.

Very interesting. Thanks GW

CET had updated:

15.2 to 30th

http://hadobs.metoffice.com/hadcet/cet_info_mean.html

We'll find out tomorrow exactly where we finished!

Met Office Hadley  15.2c      Anomaly      -0.6c.     Provisional to 30th.

Metcheck              15.20c.  Anomaly       -1.02c.

N-W                     15.64c   Anomaly        -0.56c.

Mount Sorrel         15.43c.  Anomaly        -0.77c.

My  Mean              15.9c    Anomaly        -1.6c.     

We have our first sub 15c August since 1993

14.9

http://hadobs.metoffice.com/hadcet/cetml1659on.dat