The "new normals" for 1991-2020 (CET) now updated through December

[Relativistic]

On 20/04/2020 at 12:27, Thundery wintry showers said:

One sobering fact: after the predominantly cold Februarys of 1978 to 1987, February 1988 stuck out for many at the time as a mild month, with a CET of 4.9 and, as the RMetS Weather Log noted, a marked absence of easterly winds unlike many previous Februarys (there were northerlies in the last third of the month, but not particularly potent ones).  But compared with the 1991-2020 normals, February '88 was very average temperature wise...  Also that whole winter quarter of 1987/88 was only about 0.4C above the 1991-2020 average, taking the December 1991-2019 figure of 5.0C.

February's a weird one. IIRC for periods during the LIA the average February was warmer than the average December, with 20th-century Februaries generally on the cool side (this is partly reflected by the respective mild and cold records set in 1779 and 1947).

The 1961-90 average for January was also markedly cooler than some of the earlier averages from the 20th century, with it previously peaking comfortably in the 4.xC region.

Apologies for not providing the full stats, but I have all this data on my old laptop.... which is stuck in my office.... from which I am currently forbidden to go 

[Roger J Smith]

You're right about February, I have the running 30-year averages in my file and February had a warmer 30-year average than December in these intervals:

1759-1788 (a one year wonder, after being equal for two intervals 1713-42 and 1714-43).

then continuously for a long period, 1762-91 to 1811-40 during which time Feb averages were as much as 1.0 higher than Dec (1790-1819).

After falling behind for a while, February again moved ahead of December from 1846-75 to 1879-1908 managing another relative peak of 0.9 for 1865-94.

After the 1880-1909 interval which was equal (in single digit rounded data) February never again caught December and instead took the opposite track, becoming equal to January (in one-digit rounded data) for several 30-year intervals around 1901-30 (and 0.1 lower for 1909-38), and then colder than January for the 30-year averages for both 1946-75 and 1947-76, and again 1954-83, and for several intervals 1962-91 to 1969-88, the greatest differential it achieved being 0.4 warmer than January for 1967-96. Some of these facts are less widely appreciated than they might have been, had some of the differentials hit on the standard 30-year intervals (notice how the anomaly between 1962-69/1991-98 missed both the 1961-90 and 1971-2000 normal values in which February was equal to January.

After a long period of being well behind, reaching a deficit of almost a full degree during its earlier stint as coldest month mid-20th century, February again moved ahead of December for the interval 1988-2017 (4.96 to 4.94, so when rounded looking like 0.1 although really only .02), but that has not been the case for the last three updates which all see a statistical tie in one decimal.

Although February was considerably milder than January in the Dalton (differential reached 1.8 when January hit its lowest 30-year average 1766-95), the Maunder differentials are slight and sometimes notably around 1665-94 the two months had equal averages. 

December has never been the coldest month but came quite close with differentials of 0.1 for these two 30-year intervals relative to coldest month January in each case: 1869-98 and 1870-1899. In both cases the means rounded to one decimal were 3.4 Jan and 3.5 Dec. ... The coldest 30-year average for December was 3.09 for 1783-1812, but at that point January's average was 2.76. 

Edited September 4, 2020 by Roger J Smith

[Relativistic]

Thanks for your input Roger, confirms my thoughts.

[Roger J Smith]

In post number one of this thread I have updated the new normals through November, 1991-2020 will be 7.4 for November.

The next items of business are the new December normal, and the new annual average which is locked in now at 10.2. 

As shown in post number one, December will finish 5.0 unless it's quite a cold month in which case 4.9 is on the table, or if it did a 2015 then 5.1 comes into play (but no forecast models are showing that as a possibility). 

The year 2020 can be the warmest on record beating out 2014 if December averages 7.6 or higher (for both a one and two decimal win, as 2014 was 10.95 so 2020 needs to reach 10.96 for a win). This currently seems unlikely as is second place (currently 2006 10.87) which requires 6.8. For a 4.7 finish (consensus in our forecast contest) the year will land pretty close to third place 2011 and fourth place 2018 around 10.7.

I need to check how the met office derive their annual averages as ranked on their website because it appears they weight the months by number of days, my excel file method is what I'm used to over here, an average of the twelve monthly values (unweighted) which gives February a bit of a disproportionate influence, that being the case, 2020 might be running a bit behind where I have them. Will try to solve that before end of December. 

In any case the year is likely to be around 4th or 5th overall as 5th place is currently 10.65 (tied 1990, 1999). 

[Roger J Smith]

Does anyone know when the Met Office might be releasing their official "new normals" for 1991-2020? 

I trust my math but several of the results are very close to the rounding boundaries and I know they take an average of max and min while I derived all these numbers from averages of monthly CET values. Unless you're within say .01 of the rounding boundaries, the differentials should be small. Jan at 4.66 (4.7) is one that I wouldn't swear will be 4.7, it could be officially 4.6. (I think their method is more likely to reach a lower two-digit average than my method).

Anyway, December is heading straight for the middle of the projected 5.0 zone, any value 4.3 to 7.2 will bring that result and the current end of month projections are in the range of 5 to 6 depending on how cold it turns for how long, after the 22nd. Interesting that 5.0 was also the mean for 1971-2000, a colder mean arrived for 1981-2010 thanks to 2010. The bookends there were a lot different from the books.

The annual mean will come out 10.2 from all indications. 

See the table in the original post of this thread for the various estimates on all other months. 

[Roger J Smith]

Daily Average CET daily means 1991-2020

The December values here include provisionals for Dec 2020 and forecast values 19th-31st. They will be updated at end of the month.

To the level of precision of one decimal place, a value would need to be quite close to a rounding boundary (e.g. 3.54 to 3.56) to be

subject to any revisions. 

Date__ JAN _FEB _MAR _APR _MAY _JUN _JUL _AUG _SEP _OCT _NOV _DEC 

_01 ___ 4.7 _ 4.4 _ 5.0 _ 8.3 _ 10.5 _ 13.8 _16.4 _17.3 _15.2 _13.1 _ 9.6 _ 5.8

_02 ___ 4.5 _ 4.3 _ 4.9 _ 8.5 _ 10.4 _ 14.2 _16.1 _17.2 _15.5 _12.3 _ 9.1 _ 6.0

_03 ___ 4.2 _ 4.3 _ 5.0 _ 8.0 _ 10.1 _ 13.5 _16.3 _17.3 _15.6 _12.3 _ 9.2 _ 6.1

_04 ___ 4.2 _ 4.8 _ 4.8 _ 7.7 _ 10.4 _ 13.3 _16.4 _17.2 _15.2 _11.9 _ 8.7 _ 5.5

_05 ___ 4.6 _ 4.9 _ 5.2 _ 8.0 _ 10.7 _ 13.4 _16.5 _17.6 _15.4 _11.8 _ 8.6 _ 5.6

_06 ___ 5.0 _ 4.8 _ 6.1 _ 8.3 _ 11.2 _ 14.0 _16.2 _17.6 _15.1 _11.6 _ 8.5 _ 5.4

_07 ___ 4.7 _ 4.4 _ 6.6 _ 8.0 _ 11.6 _ 14.2 _16.2 _17.3 _14.9 _11.6 _ 8.7 _ 5.5

_08 ___ 5.4 _ 4.2 _ 6.7 _ 8.1 _ 11.5 _ 14.1 _16.2 _16.9 _14.8 _12.1 _ 8.4 _ 5.4

_09 ___ 5.1 _ 4.1 _ 6.6 _ 8.5 _ 11.3 _ 14.1 _16.4 _16.9 _14.8 _11.9 _ 7.9 _ 5.1

_10 ___ 5.1 _ 4.4 _ 7.0 _ 8.7 _ 11.4 _ 14.3 _16.5 _17.0 _14.7 _12.3 _ 7.9 _ 5.2

_11 ___ 5.0 _ 4.5 _ 6.8 _ 8.4 _ 11.3 _ 13.9 _16.3 _17.0 _14.9 _12.2 _ 8.1 _ 4.8

_12 ___ 5.2 _ 5.0 _ 6.4 _ 8.3 _ 11.2 _ 14.0 _16.2 _17.1 _14.6 _11.8 _ 8.4 _ 4.8

_13 ___ 5.4 _ 4.8 _ 6.7 _ 7.8 _ 11.4 _ 14.2 _16.2 _16.3 _14.3 _11.4 _ 7.6 _ 5.0

_14 ___ 5.2 _ 4.3 _ 6.8 _ 8.4 _ 11.3 _ 14.5 _16.5 _16.5 _14.1 _11.0 _ 7.7 _ 4.6

_15 ___ 5.1 _ 5.1 _ 7.5 _ 8.6 _ 11.3 _ 14.4 _16.6 _16.7 _14.0 _10.9 _ 7.7 _ 5.0

_16 ___ 5.1 _ 5.1 _ 7.8 _ 8.6 _ 11.3 _ 14.8 _17.0 _16.4 _13.8 _10.9 _ 7.4 _ 5.1

_17 ___ 4.7 _ 5.1 _ 7.7 _ 8.8 _ 11.3 _ 15.2 _16.8 _16.7 _13.5 _10.4 _ 6.8 _ 4.7

_18 ___ 4.5 _ 5.0 _ 6.9 _ 8.6 _ 11.7 _ 15.1 _17.0 _16.7 _13.6 _10.1 _ 6.8 _ 4.9

_19 ___ 4.8 _ 5.0 _ 6.7 _ 8.8 _ 12.2 _ 15.4 _17.5 _16.6 _13.8 _10.0 _ 6.4 _ 4.7

_20 ___ 4.6 _ 5.0 _ 7.1 _ 9.2 _ 12.5 _ 15.2 _17.2 _16.6 _14.3 _ 9.8 __6.1 _ 4.0

_21 ___ 4.3 _ 5.3 _ 6.8 _10.0 _12.5 _ 14.8 _16.6 _16.6 _14.0 _10.0 _ 6.2 _ 4.6

_22 ___ 3.9 _ 5.3 _ 7.0 _10.2 _12.8 _ 14.7 _17.1 _16.7 _13.8 _10.2 _ 6.3 _ 5.2

_23 ___ 4.0 _ 5.6 _ 7.0 _10.5 _13.0 _ 15.2 _17.5 _16.5 _13.5 _10.4 _ 6.0 _ 5.6

_24 ___ 4.4 _ 5.8 _ 7.3 _10.6 _13.4 _ 15.3 _17.3 _15.8 _13.0 _10.4 _ 6.3 _ 5.1

_25 ___ 4.5 _ 5.4 _ 7.5 _10.6 _12.8 _ 15.6 _17.5 _16.0 _13.0 _ 9.9 __6.5 _ 4.7

_26 ___ 4.4 _ 5.7 _ 7.2 _10.0 _13.0 _ 15.8 _17.5 _15.9 _13.1 _ 9.9 __6.6 _ 4.5

_27 ___ 4.5 _ 5.5 _ 7.0 _ 9.7 _ 13.1 _ 16.1 _17.4 _15.4 _13.2 _ 9.8 __6.7 _ 4.0

_28 ___ 4.2 _ 4.9 _ 7.4 _ 9.6 _ 13.1 _ 15.8 _17.1 _15.4 _13.5 _ 9.8 __6.4 _ 4.0

_29 ___ 4.4 _ 5.4 _ 7.4 _ 9.7 _ 13.2 _ 16.0 _17.4 _15.0 _13.5 _ 9.3 __6.1 _ 4.0

_30 ___ 4.1 _ ---- _ 8.0 _10.2 _13.5 _ 16.1 _17.2 _15.1 _13.1 _ 9.5 __5.8 _ 4.4

_31 ___ 4.4 _ ---- _ 8.4 _ ----- _13.7 _ ------ _17.4 _ 14.8 _ ----- __9.8 _ ----- _ 4.6

__________________________________________________________________________

This 1991-2020 daily data set averages 0.85 C deg above the values for the

entire CET daily data set 1772-2020 (in rounded numbers, the annual average

has risen 0.8 from 9.4 to 10.2), albeit the 9.4 includes the last thirty years, so

from the period 1772-1990 this data set has risen almost 1.0 C deg. 

... The dates that have increased the most over the long-term averages are

12, 13 Jan (both 2.0 above), and 10, 16 Mar (both 2.1 above) 15, 17 Mar 2.0 above. 

Another peak occurs in November with 15 Nov at 1.8 above the long-term. 

Because the 1991-2020 averages are not smoothed (the 1772-2020 tend to smooth out irregular variations),

some days are actually cooler than the long-term averages. These are most frequent in June. The 4th and 5th

of June averaged 0.4 and 0.3 deg below the long-term and the 22nd of June averaged 0.2 below the long-term,

then the only other date below long-term averages was 13th July (0.1 below). 

I will attempt to link or post a graph showing these variations. 

 

variations of 1991-2020 daily CET averages vs 1772-2020 long-term averages

(note: scale of the graph is 10:1, the 5 on the vertical axis is 0.5 C deg, the 10 is 1.0 C deg, etc)

 

 

Edited December 19, 2020 by Roger J Smith

[Relativistic]

Thanks, very interesting. Not surprised to see a sustained dip from May to July, and nor the dip in December. Surprised by the dip from mid-January to mid-February, but not at all by the mild spike at the start of the year (seems to have been a feature of a number of recent Januaries; 2013 is a good example of this). Not surprised to see peaks in early Spring and late Autumn either; there have been plenty of very mild Marches, Aprils, Octobers, and Novembers in recent years. Slightly surprised by the peak in early August, but I guess a few severe heatwaves have coincided in this period which may be pulling the figures up a little (1995, 2003, and 2020 spring to mind).

Edited December 18, 2020 by Relativistic

[Roger J Smith]

Well that interval in late January to early February has generally avoided much sustained warmth in the 1991-2020 period, so it didn't gain much over the long-term (but is still warmer by about 0.5 C). That's an odd little dip in the mean daily temp on 20 Dec, you tend to get longer oscillations like those seen in late April in this data set, but a one-day dip like that is strange. I checked my data to make sure there wasn't an erroneous data point causing the phenomenon. In checking that, found that a one-day dip had occurred several times on that date, 2004 had a large one. 2007 had a similar two-day dip on the 20th and 21st. 2001 contributed with a 2.5 C one-day dip and 1995 showed a 1.9 C deficit there also. The 2010 super-cold spell bottomed out on the 20th. So it's a real data signature, probably just a statistical fluke without much other significance. It may explain why there's a perception that the weather always turns milder just before Christmas, with this colder interval then more frequent mild days that follow it. Not a hard and fast rule but the tendency in recent years. The longer-term data show an opposite trend of a falling temperature series from 24th to 26th (4.0 to 3.5). 

As you say, the first part of August is relatively warm but then the second half settles back towards the long-term average. The particularly cool day you can see there is 17 September which is 0.01 deg warmer than the long-term average, a much lower value than most through the autumn period. 

[Roger J Smith]

The updates are complete now, see the original post for the 1991-2020 averages as calculated from monthly CET values. In a few cases rounding may be different for the metoffice method of taking average of mean max and mean min. The new annual "normal" is 10.2 C. This is a value that was exceeded in only thirteen years before 1991 (two of those were 1989 and 1990) and matched after rounding by three others. During the 30 years 1991-2020, 19 years exceeded 10.2 and we're not very far from the boundary where rounding would make this value 10.3.