A few additional stats on 2015

Lowest CET mean of the year: -0.3C on 2 Feb. This was the only negative CET mean in 2015.

There were two reasonably cool spells early in the year:

CET mean from 17 to 24 Jan was just 1.2C. The highest mean in that period was 2.5C

CET mean from 2 to 7 Feb was just 1.3C. The highest mean in that period was 2.3C

Highest CET mean of the year was 24.6C on 1 July.

Only two other days went over 20C and they were 30 June with 20.2C and 22 August with 20.1C

In November and December a total of 33 days had a CET mean of more than 10C. Only 4 days had a CET mean of less than 5C (21 - 24 Nov).

Not sure if this has been discussed but how does this December compare in a standard deviation kind of way to other record months? Or, I think I'm asking, how record is it?

I guess another way of looking at it is that a tm airstream persisted for much of the month. A tm airstream perhaps a degree warmer than we might expect? Or brought to us more directly from the far SW than is usual? We didn't smash maxima much we smashed persistence of airmass?

It's interesting reading other observation as that few other months could do this since we'd expect less cloud, the odd clear night, some cold night, in other months. I guess we might, next EN, see a November the same, one that averaged 11C now seem possible, or October and 14C plus. But, not later winter or spring months I guess.

I have had a look back to see if I can find other instances where the existing CET record for a month was broken by more than 1C. I have only considered dates after there was already at least 100 years in the data series (i.e. 1760 onwards). I found one:

May 1833 saw a CET of 15.1C. The previous record was 13.8C set in 1758 and 1788. So an increase of 1.3C on the existing record.

Not quite as much of an increase in the record as December 2015 but close. At that point there were almost 200 data points in the series. In December 2015 there were over 350 data points and the movement in the record was greater. So December 2015 is unprecedented and there seems to be only one other instance in the whole of the CET data set that comes close.

Incidentally that May CET figure of 15.1C from 1833 still stands. Since then the highest May CET was 13.9C in 1848. In recent times the warmest May was 13.6C 1992 and 13.4C in 2008. There has never been a May CET in the 14's at all. So that 15.1C figure in 1833 was truly exceptional and does merit being up there with the December 2015 figure I think.

Thanks for the charts and figures along with running the Watch....  Must look at December properly later today. 

Well done to Lanky for being " Top Man" ......     

Nice to see new names.......

And what to say about December 2015 other than....  

Exactly - I am tempted to suggest that this December has been on a par with one of the most remarkable weather events of the past 100 years or so like the summer of 1976 and the winter of 1963.

I have never known a month like this December to break records like this one did.

I'll keep this stickied for at least another week so we can see all these amazing records. Make sure everyone has had time to read and digest this.

 

Exactly - I am tempted to suggest that this December has been on a par with one of the most remarkable weather events of the past 100 years or so like the summer of 1976 and the winter of 1963.

I have never known a month like this December to break records like this one did.

 

In all probability the events above are more like 1 in 1000 year events and its what makes the british climate so special and exciting you just never know what is going to happen unlike many places around the globe.

I'll keep this stickied for at least another week so we can see all these amazing records. Make sure everyone has had time to read and digest this.

I'd find it useful to look back on these threads as I often wonder how previous months performed, especially the ones that throw up record breakers like December did and I'd also like to see how I performed in previous months.  Apart from that, some of our members take the time and effort to post some very useful stats.  So it would be nice if we could keep them all because it's a shame when these threads drop off the radar and disappear for ever and I'm sure other members would agree.

Thanks in advance and anticipation! 

Would it be possible to have an archive of the CET competitions? 

I'd find it useful to look back on these threads as I often wonder how previous months performed, especially the ones that throw up record breakers like December did and I'd also like to see how I performed in previous months.  Apart from that, some of our members take the time and effort to post some very useful stats.  So it would be nice if we could keep them all because it's a shame when these threads drop off the radar and disappear for ever and I'm sure other members would agree.

Thanks in advance and anticipation! 

I'll pas this on to the admin team :-)

I am sure we can work out something. There is some sort of archive system - but its a bit complex.

I'll pas this on to the admin team :-)

I am sure we can work out something. There is some sort of archive system - but its a bit complex.

I was intrigued as to whether the exceptionally mild December could have beaten in terms of the anomaly if any 30 day period was included rather than just the days spanning a calendar month.

In order to find out the daily CET records are required to be analysed rather than the monthly averages.

Fortunately daily CET means exist from 1772 up to the current day and so a sample of 244 years (about 89000 days) is available for analysis rather than the 357 years of monthly data

It turns out that December 2015 was SO exceptional that no 30 period chosen can beat the monthly recods set last week

My list of results is shown below for both hot and cold records covering every 30 day period on the file

The cold records are far more extreme than the hot records mostly reflecting the occasional very severe winters we get with a blocking northern high

(for those interested the daily CET data was "detrended" by gradually increasing the temperature of older data to reflect the overall gradual warming in the last 250 years. This detrending totalled just under 1C applied across the whole data incrementally. Anomalies were calculated based on the average temperature for that day in the year)

I also had a go at calculating the "return period" for such an exceptional month using statistical analysis. Because the data is highly skewed towards colder temperatures at the extreme ends it is not easy. To get a 30 day anomaly that fits exactly into 1 calendar month is itself a 1 in 30 chance and to allow for this and the relative rarity of extreme warm periods I got  a return period of 1850 years - although I'm not convinced about this

I was intrigued as to whether the exceptionally mild December could have beaten in terms of the anomaly if any 30 day period was included rather than just the days spanning a calendar month.

In order to find out the daily CET records are required to be analysed rather than the monthly averages.

Fortunately daily CET means exist from 1772 up to the current day and so a sample of 244 years (about 89000 days) is available for analysis rather than the 357 years of monthly data

It turns out that December 2015 was SO exceptional that no 30 period chosen can beat the monthly recods set last week

My list of results is shown below for both hot and cold records covering every 30 day period on the file

The cold records are far more extreme than the hot records mostly reflecting the occasional very severe winters we get with a blocking northern high

(for those interested the daily CET data was "detrended" by gradually increasing the temperature of older data to reflect the overall gradual warming in the last 250 years. This detrending totalled just under 1C applied across the whole data incrementally. Anomalies were calculated based on the average temperature for that day in the year)

I also had a go at calculating the "return period" for such an exceptional month using statistical analysis. Because the data is highly skewed towards colder temperatures at the extreme ends it is not easy. To get a 30 day anomaly that fits exactly into 1 calendar month is itself a 1 in 30 chance and to allow for this and the relative rarity of extreme warm periods I got  a return period of 1850 years - although I'm not convinced about this

 

 

 

 

.

Why detrend the data? Or, in what way does the undetrended data result differ I guess I'm asking?

I see Tamino has done a similar analysis, though just limited to calendar months - unsurprisingly, the same result.

Interesting 74/5 is in there - very similar by my recollection, if less tm dominated in December. But a little surprising that a 30 day spell in 76 didn't beat it - many days must have had maxima well past 5C above normal, cooler nights maybe. Indeed, I just wonder how high the minima anomaly for December was, I bet it's here somewhere....

Many thanks for all your efforts throughout the year. Despite my dismal effort this year I love doing this and will continue to do so for as long as you are running the comp.

Happy new year

 

.

Why detrend the data? Or, in what way does the undetrended data result differ I guess I'm asking?

I see Tamino has done a similar analysis, though just limited to calendar months - unsurprisingly, the same result.

Interesting 74/5 is in there - very similar by my recollection, if less tm dominated in December. But a little surprising that a 30 day spell in 76 didn't beat it - many days must have had maxima well past 5C above normal, cooler nights maybe. Indeed, I just wonder how high the minima anomaly for December was, I bet it's here somewhere....

Dev:

I Just detrended to get to a "level playing field". The early data was at the end of the LIA so getting a 30C day then was more impressive than today and similarly colder winters in the 18th and 19th century were more prevalent. The adjustment on the oldest data was about +0.95C gradually reducing to zero on current data.

The average adjustment was therefore about 0.48C so without this detrending but still using date averages across the whole file, you would subtract about 0.5C from the oldest anomalies (e.g. 1779) and add 0.5C to the newest (e.g 2015).  I don't think this is such a fair comparison and would show December 2015 as even more remarkable

As for summer 1976, I just had a look and the main blast of the real hot weather was 23/06 up to 09/07 so a lot less than 30 days. The remaining days up to the end of July were more moderate and this reduced the overall average somewhat

Many thanks for all the work you put into this, GW - we all really appreciate it.

I was intrigued as to whether the exceptionally mild December could have beaten in terms of the anomaly if any 30 day period was included rather than just the days spanning a calendar month.

In order to find out the daily CET records are required to be analysed rather than the monthly averages.

Fortunately daily CET means exist from 1772 up to the current day and so a sample of 244 years (about 89000 days) is available for analysis rather than the 357 years of monthly data

It turns out that December 2015 was SO exceptional that no 30 period chosen can beat the monthly recods set last week

My list of results is shown below for both hot and cold records covering every 30 day period on the file

The cold records are far more extreme than the hot records mostly reflecting the occasional very severe winters we get with a blocking northern high

(for those interested the daily CET data was "detrended" by gradually increasing the temperature of older data to reflect the overall gradual warming in the last 250 years. This detrending totalled just under 1C applied across the whole data incrementally. Anomalies were calculated based on the average temperature for that day in the year)

I also had a go at calculating the "return period" for such an exceptional month using statistical analysis. Because the data is highly skewed towards colder temperatures at the extreme ends it is not easy. To get a 30 day anomaly that fits exactly into 1 calendar month is itself a 1 in 30 chance and to allow for this and the relative rarity of extreme warm periods I got  a return period of 1850 years - although I'm not convinced about this

 

 

 

 

Excellent post Lanky and just goes to show how exteme December was, but also highlights how few 30 day periods in recent times have been exteme regards anomoly.

A different question for you or Gw how far back or forth from december do you have to go to find a higher 30 period in actual cet as opposed to an annomaly difference.