November 2021 C.E.T. and EWP contests -- final month of the 2020-21 competitions

[Walsall Wood Snow]

Wow only 0.1c off being correct this time.

Anyway at 7.3c November 2021 shares the same mean CET with Novembers:

1839

1863

1900

1906

1948

1960

2007

 

 

[damianslaw]

3 hours ago, Summer Sun said:

Confirmed as 7.3c

 

https://www.metoffice.gov.uk/hadobs/hadcet/cetml1659on.dat

That is a significant drop! Was not expecting such an adjustment.. mind we are beginning to see regular downward adjustments of up to 0.5 degrees it seems. Not far off a near normal mean compared to last 20 years.

[Roger J Smith]

I always trusted their process until recently and now I have to say this looks bogus to me and there would be one or two obvious reasons for tampering with the numbers to get a certain outcome. Look at the 30th for example, that can't be right. This will create a result that some people will find bogus and it also makes the tracking of the values through the month pointless. 

[Scorcher]

That's a ridiculous adjustment. I'd like to hear an explanation for that. There seems to be a big bias these days towards making months significantly cooler at the end of the month.

A final figure that doesn't anywhere near reflect what a mild first 3 weeks or so we had.

Not only did it feel unusually mild, it was also very dry and benign. A rather rare combination for November.

[Dancerwithwings]

On 27/11/2021 at 22:57, Don said:

My 7.0C is dead in the water.  Punted too low yet again!  Hopefully this won't happen for December!

With this massssive correction you wasn’t that far off in the end

Anyway Roger.....some drop and bogus is taking it likely.....

At this rate we’ll be starting a new competition....GUESS THE CORRECTION

 

[Roger J Smith]

I suppose the overall correction and the value they assign to the 30th are two different issues, would say that the legit value for the 30th is around 8 degrees. That is basically about half of the correction problem explained. The other half would be a bit closer to what we're used to seeing. I doubt that they would correct that value at least for a while, and if they ever do correct it and the monthly average changes then the whole thing will be long forgotten as we get into new contests. I wish they would develop a more transparent system like every other weather contest has available to it, data bases that don't get adjusted, only updated for new daily data, so you can track it without wondering if you're wasting your time, and the process has integrity. Anyway, whatever results this generates will be meaningless for quite a few people here, I would suspect. 

[Don]

1 hour ago, Dancerwithwings said:

With this massssive correction you wasn’t that far off in the end

Anyway Roger.....some drop and bogus is taking it likely.....

At this rate we’ll be starting a new competition....GUESS THE CORRECTION

 

No I wasn't so far off was I!  Like you, I was also surprised with the large downward correction, but not complaining! 

Hope I'm not too low with December.....

[North-Easterly Blast]

November 2021 has seen the first 7s finish for the CET since 2008.  13 years between a November in the 7s is a significant long gap, when you consider that it is a pretty common number for a November CET.  7.3 is actually fairly average for November by recent standards.

Edited December 1, 2021 by North-Easterly Blast

[J10]

The correction from 7.9c to 7.3c has changed the winner in the overall competition this year.

[J10]

I would like to apologise for an error in the original figures, which incorrectly allocated penalty points this month.
It was also unfortunate that this made such a big difference to the overall scores.

As ever if there are any issues with the scores please let me know.

Monthly

PDF Nov 2021 Summary.pdf

Excel November 21 CET.xlsx

This month seven players got it spot on Bobd29, SLEETY, Kirkcaldy Weather, swfc, emmett garland, mulzy and Quicksilver1989.

In total 31 were within 0.5c.

Seasonal

Some changes in the Top 3, Bobd29 (1st from 3rd), General Cluster (2nd from 1st) with Quicksilver1989 (3rd from 12th)

Overall

Some changes in the Top 4,

Summer18 (1st from 2nd)
Quicksilver1989 (2nd from 5th)
Leo97t (3rd from 7th)
reef (4th from 1st)

A big congratulations to Summer18 for winning and a big thanks to everyone taking part this year.

Edited December 1, 2021 by J10

[Roger J Smith]

Just for anyone dropping in to find out about the EWP contest, scroll back to early 29th Nov and all the scoring summaries are there, the outcome for NOV was below all forecasts (looks to be around 40-42 mm) so I was able to score the contest before the month ended. The winner was Feb1991blizzard with noname_weather, snowray, bobd29, Midlands Ice Age the rest of the top five (annual). The PIT won Nov and moved to 10th place. From 6th to 9th we have Godber 1, Mr Maunder, Don and Reef. February1978, davehsug, Kirkcaldy Weather, Polar Gael, Frigid, seaside60, virtualsphere, Mulzy, SteveB and Ed Stone round out the top 20. 

Congrats Reef and stargazer on a great effort in the CET, you have probably been robbed of a legitimate 1-2 finish (which way I'm not sure, from 7.7 or so would say it would have been a virtual tie). It should be kept in mind that the results from 7.3 follow on from a value that is bogus because of the obvious error on the 30th and other considerations. Pershore on the 30th had a max of 13.4 and a min of 3.5, Rostherne was 10.5 and 3.6. The mean of those four would be 7.9 C. So how do they come up with 4.7 from that? Well, the min is listed on their site as -2.0, same as 28th had been. Makes me wonder if it's a clerical error. There was no place in England with a min of -2.0 on the 30th.

My take away from the combination of forecasts and probable "real" monthly value is that we have more or less a multiple photo finish with several people scoring within a very close range if that were the case. We can't do anything about it but if they change it later, then what? Probably doesn't matter that much in the grand scheme of things, much bigger problems in the world. 

 

Edited December 2, 2021 by Roger J Smith

[J10]

A quick look this month, and the final CET meant there could have been 4 different winners.

Edited December 1, 2021 by J10

[Dancerwithwings]

16 minutes ago, J10 said:

A quick look this month, and the final CET meant there could have been 4 different winners.

Look away Reef.....you was robbed

I’m happy with my overall 7th, another top 10 finish for me again.

but I do wonder what my overall position would of been if Nov would of been 7.6c

[Roger J Smith]

Thanks to J 10 for going to the trouble of showing the potential outcomes at various CET values and of course for all the work on scoring the CET contests through long stretches of modern civilization. That was about what I had guessed to be the case although I thought the differential was so slight that Stargazer probably had a chance to win from 7.9 and maybe 7.8 as well. Evidently not. Dancerwithwings, I think you might have gotten up to about 3rd or 4th place at 7.6, but all these scores are so close it would take ages to figure it all out. The more important factoid is that overall it was a much closer top ten than usual no matter what the final value chosen.

I'm thankful that the EWP contest had such a low maintenance finish, the only way Feb1991Blizzard could have lost would have been if he forgot to enter or said 200 mm perhaps. 

Going to keep at it, my scoring turned much better after a very poor start, don't think I lost ground at any point from March to November. If the contest was scored off end of month provisional values I feel like I might have come out ahead too. And I hope they fix that value for the 30th, that is just plain wrong. I'm aware of one or two errors in the max-min tables of the CET, there's one day (in Dec 1885) when the max is lower than the min. Now we know that can't be. 

Edited December 2, 2021 by Roger J Smith

[Roger J Smith]

This feature will continue to appear in this November thread until the 3th of December just to keep it separate from the busy first few days of contest entry material. Then it will go over there for the 4th of December on.

 

Almanac for 2nd December

The highest daily mean (1772-2020) was 12.2 in 1985. This replaced 11.5 from 1939. Other mild 2nd of Decembers include 11.4 (1779), 11.3 (1974), 11.1 (1857), 10.9 (1979), 10.8 (2018), 10.7 (2015), 10.6 (1876 and 1898), 10.5 (1873) and 10.4 (1918).

The lowest daily mean (1772-2020) was -4.7 in 1796. Since then the coldest values have been -4.2 in 1879, -2.7 in 1856, -2.4 in 1874 and 1893, and -1.4 in 2010. 

The highest maximum (1878-2020) was 14.4 in 1985 which replaced 14.2 from 1954. Other warm values include 13.6 (1948), 13.3 (2018), 13.2 (1979), 13.1 (1974), 13.0 (1953) and 12.8 (2015). 

The highest minimum (1878-2020) was 9.9 in 1985, which replaced 9.5 from 1974. Before that 8.7 in 1898 was mildest. 

The lowest maximum (1878-2020) was -1.2 in 1879, although 1796 could have been around -2 or so. Since then the coldest value in recent years was 0.4 in 2010 -- other cold values include 1.0 in 1893 and 1.5 in 1962.  

The lowest minimum (1878-2020) was -7.1 in 1879, although it could have been colder in 1796. Since then the coldest value has been -5.8 in 1893, followed by -4.1 in 1925 and 1952, and -4.5 in 1973, then -3.1 in 2010.  

The wettest 2nd of December (1931-2020) in the EWP data was 12.49 mm (1996) followed by 11.27 mm (1992), 10.81 mm (1937), 9.96 mm (2006), and 9.73 mm (2005).

The wettest week ending 2nd December (26th Nov -2nd Dec) was 61.18 mm in 1992, followed by 55.65 mm in 1954, and 52.32 mm (2009).

The least amount falling in a week ending 2nd December was 0.23 mm in 1962, then 0.37 mm in 1957, 0.44 mm in 1989, 1.08 mm in 2016, 1.12 mm in 2013, and 1.64 mm in 1991.

-----------------------------

(map for 02 Dec 1985,  very mild interval) 

https://www.wetterzentrale.de/reanalysis.php?map=1&model=noaa&var=1&jaar=1985&maand=12&dag=02&uur=1200&h=0&tr=360&nmaps=24#mapref

[Roger J Smith]

EWP tracker has reached the end of the road and the value is 43 mm. 

A more precise table value will be posted on the 5th and I will leave any edits of the scoring post (back on 29th Nov) for that, as the differentials

at this point are insignificant. 

[Relativistic]

Apart from the obvious error for the 30th the correction doesn't seem ridiculous to me. There've been many corrections in the past that have been around -0.5C, and they go back quite far (I remember August 2011 having a huge drop).

The data may well be corrected at some point as I've seen figures change in the past. The anomaly for February 2012 was listed as 0.0C for the remainder of 2012, which was corrected to +0.1C at the end of the year, suggesting that perhaps one of the daily values was reviewed and corrected.

[Roger J Smith]

Top combined annual scoring results

These are the top finishers in terms of high ranked results in both contests. I have limited this to regular entrants in both contests. Beyond that you can check the excel files for both contests and see how you finished. 

Rank ____ FORECASTER _______ CET ___ EWP _____ Avg 

_ 01 _____ Feb1991Blizzard ______ 6 _____ 1 _______ 3.5 

_ 02 _____ Reef ____________________ 4 _____ 9 _______ 6.5

_ 03 _____ noname_weather ____ 13 _____ 2 _______ 7.5

_ 04 _____ Bobd29 ________________19 _____ 4 ______11.5

_ 05 _____ Mr Maunder __________ 17 _____ 7 ______12.0

t 06 _____ davehsug ______________14 ____ 12 _____ 13.0

t 06 _____ February1978 _________15 ____ 11 _____ 13.0

_ 08 _____ Mulzy __________________10 ____ 18 ______14.0

_ 09 _____ Polar Gael _____________16 ____ 14 ______15.0

_ 10 _____ Ed Stone (Gen Cluster) _11 ____ 20 ______15.5

_ 11 _____ Don ____________________24 _____ 8 ______16.0

t 12 _____ Leo97t __________________3 _____33 ______18.0

t 12 _____ Godber 1 ______________30 _____ 6 ______ 18.0

_ 14 _____ Roger J Smith __________ 8 _____32 ______ 20.0

t 15 _____ J 10 ____________________ 18 ____ 23 ______20.5

t 15 _____ The PIT ________________t31 ____ 10 ______20.5

_ 17 _____ Weather26 ____________ 22 ____ 22 ______22.0

_ 18 _____ DR(S)NO _______________21 _____24 ______22.5

_ 19 _____ Midlands Ice Age ______42 _____ 5 ______ 23.5

_ 20 _____ weather-history ______ 12 _____36 ______ 24.0

t 21 _____ Summer18 _____________1 _____ 48 ______ 24.5

t 21 _____ seaside60 _____________t33 _____16 ______24.5

t 21 _____ snowray _______________ 46 _____ 3 ______ 24.5

_ 24 _____ Stargazer _______________5 _____45 ______ 25.0

_ 25 _____ Frigid __________________ 36 _____15 _____ 25.5

_ 26 _____ Kirkcaldy Weather ____ 40 ____ 13 _____ 26.5

_ 27 _____ daniel* ________________ 29 _____28 ______28.5

_ 28 _____ Steve B ________________ 39 _____ 19 _____ 29.0

_ 29 _____ syed2878 ______________38 _____21 ______29.5

_ 30 _____ Jeff C __________________t33 _____27 ______30.0

t 31 _____ Emmett Garland ______23 _____38 ______ 30.5

t 31 _____ virtualsphere _________ 44 _____17 ______ 30.5

_ 33 _____ Stationary Front ______ 37 _____25 ______ 31.0

_ 34 _____ Summer Blizzard _____ 27 _____39 ______ 33.0

_ 35 _____ shillitocettwo _________t31 _____37 ______ 34.0

_ 36 _____ I Remember Atl 252 __ 28 _____ 41 _____ 34.5

_ 37 _____ jonboy _________________ 47 _____ 26 _____ 36.5

t 38 _____ DiagonalRedLine ______45 _____ 29 _____ 37.0

t 38 _____ Walsall Wood Snow ___ 43 _____ 31 _____ 37.0

_ 40 _____ Neil N __________________ 48 _____ 42 _____ 45.0

_ 41 _____ Earthshine _____________49 _____ 50 _____ 49.5

=======================================================

Thanks for playing both contests ... these are the CET-only ranks missing from the above ...

Quicksilver1989 2nd, dancerwithwings 7th, Summer Sun 9th, Mark Bayley 20th, Typhoon John 25th

damianslaw 26th, Duncan McAlister t33rd, Kentish Man 41st. 

_ Note that rwtwm joined at the halfway point, playing seven months, looking at 12/7 points and

considering average error etc, their ranks would have been around 7th CET and 17th EWP, avg 12 _

so rwtwm scoring at about the pace of 5th in the table above. 

[Roger J Smith]

 

 

Report on Consensus and the two Normals in CET contest

For the CET contest, we do not officially score consensus or the two recent normals. (In the EWP contest they are tracked). So here's an update which shows how our consensus (median rather than mean) and the two normals have been doing. 1991-2020 played December as 1990-2019. All I can report here is their rank in each contest, as the other scoring parameters are a bit too complex for me to recreate after the fact. The rank shown is equivalent to the highest scoring forecaster at that temperature level, some others would have ranked lower (higher rank numbers). 

The number in brackets beside the rank is the number of forecasts submitted. 

 

Robotic forecaster ____ DEC rank _ JAN rank _ FEB rank _ MAR rank _ APR rank _ MAY rank

Consensus ____________ 27th (97) _ 16th (78) _38th (80) _ 22nd (62) __ 30th (62) _ 31st (64) __ cont'd below

1981-2010 normal ____ 27th (97) __51st (78) _ 5th (80) __ 36th (62) __ 37th (62) _ 48th (64) 

1991-2020 normal* ___ 4th (97) ___56th (78) _ 3rd (80)__ 33rd (62) __ 48th (62) _ 51st (64) 

Consensus error _______ --0.4 ______ --0.3 ______ --2.1 ______ --0.3 _______ +1.9 _______ +1.0 ____ cont'd below

1981-2010 error _______ --0.4 _______ +1.3 ______ --0.7 ______ --0.6 _______ +2.1 _______ +1.6 ____

1991-2020 error _______ --0.1 _______ +1.6 ______ --0.2 ______ --0.5 _______ +2.6 _______ +1.8 ____

 

Robotic forecaster ____  JUN rank_  JUL rank_ AUG rank _ SEP rank_ OCT rank _ NOV rank ____  average rank (12 mo)

Consensus ____________  14th (66) _ 25th (58) _ 26th (56) _ 25th (55) _ 28th (59) __ 1st (63) _____ 23rd-24th (67)

1981-2010 normal ____ 28th (66) _ 30th (58) _ 24th (56) _ 42nd (55) _ 42nd (59) __12th (63) ____32nd-33rd (67) 

1991-2020 normal* ___ 31st (66) _ 28th (58) _ 26th (56) _ 36th (55) __ 35th (59) ___ 8th (63) ____30th (67)

Consensus error _______ --0.2 ______ --0.8 _______ +0.7 ______ --1.4  ______ --1.0 ______ 0.0 _______ avg abs 0.84

1981-2010 error _______ --1.0 ______ --1.0 _______ +0.6 ______ --1.9 ______ --1.4 _______-0.2 _______  avg abs 1.07

1991-2020 error ________ --0.8 ______ --0.9 ________+0.7 ______ --1.7 ______ --1.2 ______+0.1 _______ avg abs 1.02

mean bias of consensus is --0.24, mean bias of 1981-2010 is --0.13, mean bias of 1991-2020 is +0.12 _ these are the average errors with sign utilized. It basically says that the average month was slightly warmer than 1981-2010 normals and slightly cooler than 1991-2020 normals, and that our consensus forecasts were overall 0.24 C too low (we have a slight group cold bias, go figure). 

ANALYSIS: Our consensus made yet another small gain on the two normals, and our group consensus in October had zero error, but the two normals also did well. The end result shows slight "group forecaster" skill overall. This was not the case as recently as the Dec to July averages. The EWP values for the same three "robotic" forecasters are all higher (lower rank numbers, better performances), in fact group consensus in the EWP is ended up with more points than the 5th place finisher, and the two normals would have both finished 11th. Our consensus in the EWP contest had an average contest rank of 22nd, while the two normals were both between 24th and 25th. This is about the same slight improvement in group forecasting over random chance as demonstrated by the performance of normals. 

One other metric for the CET results, the average absolute error of consensus finished equal to 7th lowest average error among the frequent or "regular" entrants, and the 1991-2020 average error was bettered by 20 of the top 21 finishers, while the 1981-2010 average error was behind the first 22, and 24th, and tied with 27th place in the contest (out of 49 regular or frequent entrants). Thus the top half of the contest entrants often exceeded random measures, beating consensus is more difficult in most contest situations and in my experience, the number who do so is often about 10 to 20% of the total number of entrants. The math involved pretty much guarantees that consensus will beat two-thirds of contest entrants on a regular basis.  In other contests on other weather forums, at this stage, similar numbers are ahead of consensus.

You can see quite clearly that a good strategy in these contests is to consider what might be a blend of the most recent average with the likely group consensus. Anyone who waited to the last moment and worked those out would probably guarantee a top five finish. (this is not what I do by the way, but people smarter than me do this, I am pretty certain). 

--------------------------------------------

[ItNeverSnowsInWindsor]

On 27/11/2021 at 22:57, Don said:

My 7.0C is dead in the water.  Punted too low yet again!  Hopefully this won't happen for December!

I went way low!

[Frigid]

On 23/10/2021 at 10:29, Frigid said:

7.3C and 75mm. I don't see an overly cool November which would make Autumn 2021 one of the warmest. 

If only I stuck with my original guess...

[noname_weather]

Thanks for hosting the competition!! @Roger J Smith. Congratulations to the winners.

I went too far ahead with the cold wave, but hey I promised some snow and I guess that -6C 850 hPa 980mb on the 28th probably delivered some for higher ground.

Probably broke the record for the biggest downfall. From 1st to being out of the top 10 :P. Risk it for busquit. I guess that I will starve for now.

[reef]

I cant believe that downwards correction. Thought I had it 

It must be one of the largest adjustments surely?

[Summer18]

Wow can’t believe I ended up finishing top on my first ever attempt at this, guess all those years lurking have paid off  Shame it’s in such suspect circumstances, I had written myself off for most of the month! Anyway well done all you guys who came close! 

[Scorcher]

2 hours ago, reef said:

I cant believe that downwards correction. Thought I had it 

It must be one of the largest adjustments surely?

Yes really surprising. It basically means the month will be looked at in the record books as an average month. Yet for me it's one of the most memorable mild Novembers- certainly the first 3 weeks.