Jump to content
Sign in to follow this  
Roger J Smith

November 2018 C.E.T. forecasts and optional EWP contest

Recommended Posts

Sorry about the confusion DS, I got mixed up between who said what in the quotes, now I see why you were asking me why I mentioned your forecast CET. Anyway, whatever the provisional value for today, when they adjust it to calendar day as I think they do, it will likely go up a bit, so will need to check that record value of 12.2. 

Share this post


Link to post
Share on other sites

Sunny Sheffield at 7.4C +0.3C above normal. Rainfall 82.3mm 103.7% of average. I'm wondering if there's a chance of 100mm being reach now. 

The final temp looks to be 7.4C unless tomorrow is a lot milder or a lot colder than forecast. 

Share this post


Link to post
Share on other sites

Hadley EWP tracker was at 87 mm through 27th, then add on whatever they make of last 48 hours and a top up on 30th, could go to 105 mm now, possibly 110. Rainfall quite heavy in west last 36 hours. EW average probably about 15-20 mm.

Singularity will end on a high note easily winning the contest either way we scored it. 

Share this post


Link to post
Share on other sites

8.3c to the 29th

1.7 above the 61 to 90 average
1.1 above the 81 to 10 average

___________________________________

Current low this month 6.8c to the 1st & 2nd

Current high this month 10.0c to the 16th *No data received for the 10th*

Share this post


Link to post
Share on other sites
On ‎29‎/‎11‎/‎2018 at 10:33, Scorcher said:

 

Completely irrelevant, all I've said is it wasn't going to end up below average. All a lot of you have been posting is about 'how low it's going to go' in this thread for 2 weeks now. Well we had a rise yesterday! Quite impressive so late in the month.

 

So you didn't make a forecast then!

Anyway, I went for above average(7.3) with a cold last third bringing the CET down, which is what has happened, final figure could be 7.9, so not a bad forecast.

(had far worse)😁

Edited by DAVID SNOW

Share this post


Link to post
Share on other sites
On 01/11/2018 at 12:08, Quicksilver1989 said:

A bit late but I'll go for a CET estimate of 8.4C a very mild month with a brief cold snap in the second half.

Close but not close enough.... seems like my guess shall be a tad too high....

Share this post


Link to post
Share on other sites

Sunny Sheffield is at and will finish on 7.4C +0.7C above normal. Unless the mild air arrives much quicker than shown. Rainfall won't make 100mm but should finish in the 90's. At the moment it's at 86.6mm 109.1% of the months average.  

Share this post


Link to post
Share on other sites

Spreadsheet
Nov 2018 CET.xlsx

Monthly Results
No one got it spot on this month, however

Roger J Smith, syed2878, vizzy2004 and Quciksilver1989 were each 0.1c out.

image.thumb.png.2502488eb9b8ce847e3048a699e64f28.png

Seasonal Results
A large mix around with a mostly new Top 3 .                      

Melatron was 1st, Summer blizzard 2nd and DR(S)NO 3rd.

image.thumb.png.fc8400cf0e663e9662632b3166e21efe.png

Overall Results
Overall no changes in the Top 4.

Man with Beard with an emphatic win, Don 2nd, Norrance 3rd, metaltron 4th.

image.thumb.png.2b5c14d46817357f4fbc4df945cbbfb5.png

Edited by J10

Share this post


Link to post
Share on other sites

Well done and Congratulations to all the winners and Thanks to everyone taking part.

Yearly Summary

Monthly Winners and Correct Predictions

image.thumb.png.cdcbdeafefe201dbfa8efe61f82fd26e.png

image.thumb.png.9cf1188ad2402c4a7f422b9b0def25fc.png

Seasonal Winners

image.thumb.png.6f68c77d31bcdbc2769c0f43d52ad839.png

Edited by J10

Share this post


Link to post
Share on other sites

Happy with my result this month!

November finished 7.6c here, 1.1c above the 1981-2010 average.

Rainfall finished at 54.8mm which is 84% of the average.

Share this post


Link to post
Share on other sites

_ Report on the Three Robotic Forecasters for November 2018 _

___ The final results for the contest year ___

 

The two normals made a little further progress on consensus in November, but fell short of catching consensus, which means that AS A GROUP WE SHOWED SKILL. Well a little bit anyway. Details follow ...  

___ Past months ____

____________DEC _________________________ JAN ________________________ FEB

"Forecaster" _ Error _ Pts (range) _ rank (range) __ Error _ Pts (range) _ rank (range) __ Error _ Pts (range) _ rank (range)

1987-2016*__ +0.2 __ 89.4 - 93.4 __ 6 to 9 (88-17)>--0.6 __66.7 - 74.1 _ 15 to 19 ______ +1.9 _ 31.2 - 35.1 _ 51 to 54

1981-2010 __ --0.2 __ 89.4 - 93.4 __ 6 to 9 ______ --0.9 __ 50.0 - 55.6 _ 25 to 28 ______ +1.5 _ 48.1 - 49.4 _ 40 to 41

Consensus __--0.6 __ 71.0 - 74.9 __20 to 23 _____--0.9 __ 50.0 -55.6 _ 25 to 28 _______+1.1 _ 57.1 - 58.4 _ 33 to 34

* becomes 1988-2017 from JAN onward.

____________ MAR ___________________________________ APR _____________________ MAY ____________ 

"Forecaster" _ Error _ Pts (range) _ rank (range) __ Error _ Pts (range) _ rank (range) __ Error _ Pts (range) _ rank (range)

1988-2017 __ +1.9 __ 20.8 - 22.1 __ 60 to 61 ____ --1.1 __ 72.2--72,2 __ 18 to 18 _____ --1.3 _ 28.6 - 33.3 _ 43 to 46

1981-2010 __ +1.7 __ 26.1 - 26.1 __ 57 to 57 ____ --1.3 __ 57.4--65.6 __ 22 to 27 _____ --1.5 _ 19.1 - 20.5 _ 51 to 52

Consensus __--0.3 __ 78.9 - 85.5 __12 to 17 ____ --1.4 __ 50.8--55.8 __ 28 to 31 ______ --1.0 _ 46.1 - 54.0 _ 30 to 35

 

______________JUN ________________________JUL ___________________________ AUG _____________

"Forecaster" _ Error_ Points (range)_rank (range) __Error _ points (range) _ rank (range) __Error _ points (range) _ rank

 

1988-2017 __ --1.5 __ 12.0 - 12.0 __ 57 to 57 ____ --2.4 __ 03.1 - 03.1 __ 63 to 63 ____ --0.1 ___ 95.3 - 98.4 __ 02 to 04

1981-2010 __ --1.6 __ 04.1 - 10.4 __ 58 to 62 ____ --2.4 __ 03.1 - 03.1 __ 63 to 63 ____ --0.2 ___ 93.7 - 93.7 __ 05 to 05

Consensus __--0.3 __ 74.9 - 82.7 __ 12 to 17 ____ --0.9 __ 51.6 - 53.2 __ 31 to 32 ____ +1.3 ___ 53.0 - 55.2 __ 29 to 31

 

 

______________SEP ________________________OCT ___________________________ NOV _____________

"Forecaster" _ Error_ Points (range)_rank (range) __Error _ points (range) _ rank (range) __Error _ points (range) _ rank

 

1988-2017 __ +0.4 __ 83.3 - 83.3 __ 11 to 11 ____ +0.4 __ 67.4 - 73.9 __ 18 to 22 _______ --1.0 __ 65.1 - 66.5 __ 24 to 25

1981-2010 __ +0.3 __ 85.0 - 88.3 __ 08 to 10 ____ +0.1 __ 89.2 - 96.9 __ 03 to 08 _______ --1.2 __ 56.3 - 59.2 __ 29 to 31

Consensus __+1.2 __ 43.3 - 51.7 __ 30 to 35 ____ --0.1 __ 89.2 - 96.9 __ 03 to 08 _______ --1.4 __ 49.0 - 52.0 __ 34 to 36

 

________ Current Month __ Nov 2018 ________________ Contest year averages to date (12 mo)___

"Forecaster" _Nov Error_Nov Points (range)_rank (range) // mean abs error_mean points_mean rank

1988-2017 ___ --1.0*_____ 65.1 - 66.5 ____ 24 to 25 ________1.06*_____ 53.8 - 56.3 __ 31 to 33 

1981-2010 ___ --1.2*_____ 56.3 - 59.2 ____ 29 to 31________ 1.07*_____ 53.0 - 55.2 __ 31 to 33

Consensus ___--1.4*_____ 49.0 - 52.0 ____ 34 to 36 ________ 0.88*_____ 60.5 - 64.5 __ 24 to 28 

______________________________________________________________

So for the contest year, our consensus was about 0.2 deg better than the two normals, and our consensus ranked about seven spots higher in the raw ranking (based only on temperature error).

A mean error 0.88 deg (our consensus mean error) was surpassed by 12 of the regular entrants and matched by two others.

A mean error of 1.06 or 1.07 deg (the two normals) was surpassed by about 27 regular forecasters in these contests. This tells us that the top half of the field are showing some skill vs random chance. This past year, the more recent normals did no better really than the 1981-2010 normals, likely a result of the cold months in spring and autumn. 

That mean error of 0.88 deg is an average absolute error, the mean error itself was -0.27 deg. This means that our average forecast was 0.27 lower than the actual outcome, over twelve months. We guessed generally lower than the warm months and generally higher than the cold months (relative to normal) which shows a conservative frame of mind among the forecasters, however, we spotted trends to beat normals. 

The mean of the twelve months was 0.48 deg above the 1981-2010 normals. We cut 0.21 off that with our slightly skilled group consensus. 

By rank of consensus, the order of skill shown was October first and March tied second with June. 

I think our biggest challenge is to spot mid-month reversals of trends. For example, August which reversed trend around the 23rd finished a good deal cooler than most of us expected and this was our worst outing relative to the two normals, September was only marginally better. 

________________________________________________

On a different topic, congratulations to Man With Beard for leading the pack (again). And well done to all in the top ten of the annual contest. Finally, thanks to J10 for putting up with all of us. As to the November win, I seem to remember this happening in 2015 which is not a great thought, is it? 

Edited by Roger J Smith

Share this post


Link to post
Share on other sites

At 8.3c November 2018 is the first ever November going back to 1659 to record this exact figure. However October's 1829, 1799, 1763, 1751 and 1704 did record this same figure. March's 2012, 1990 and 1948 did also, as well as April's 1985, 1975, 1930, 1900, 1828, 1822, 1805, 1801, 1773, 1728, 1704 (same as October the same year) and 1703. This covers every recorded instance I believe. Funny though that November ended up with a CET much more common in April than any other month.

Edited by Walsall Wood Snow

Share this post


Link to post
Share on other sites
5 hours ago, BornFromTheVoid said:

An rare upward correction this time, of 0.03C.

Seems they ignored urban warming this month...

Share this post


Link to post
Share on other sites
1 hour ago, DAVID SNOW said:

Seems they ignored urban warming this month...

Yup. Without doubt, that's the only conceivable explanation.

Share this post


Link to post
Share on other sites
36 minutes ago, BornFromTheVoid said:

Yup. Without doubt, that's the only conceivable explanation.

I wonder why they have done this?

Share this post


Link to post
Share on other sites
4 hours ago, J10 said:

Well done and Congratulations to all the winners and Thanks to everyone taking part.

 

Yearly Summary

Monthly Winners and Correct Predictions

image.thumb.png.cdcbdeafefe201dbfa8efe61f82fd26e.png

image.thumb.png.9cf1188ad2402c4a7f422b9b0def25fc.png

Seasonal Winners

image.thumb.png.6f68c77d31bcdbc2769c0f43d52ad839.png

Thanks J10 and @Roger J Smith for all your hard work and comments! Very surprised to be on top again, got lucky with a few big months early on such as February, and managed to keep it ticking over in the last few months without really doing that well! No silly speech from me this year, but might crack open the vintage port in the cupboard tonight 🙂 Here's to next year and good luck everyone 🙂

oh and @Summer Sun and @BornFromTheVoid, thanks greatly for all your updates every month, wouldn't have been the same without you both. 

Edited by Man With Beard

Share this post


Link to post
Share on other sites
6 hours ago, Don said:

I wonder why they have done this?

Sorry, one of those situations where a "/s" was needed at the end. My comment was entirely factitious
I'm certain they have a perfectly valid explanation for the correction applied, the same as every month.

Share this post


Link to post
Share on other sites

People shouldn't read it as an upward adjustment, it was more or less sideways, so that means they didn't find many cases where the calendar day readings were much different from the partial calendar day preliminary readings. If you have a month with several days of clear, cold evenings that see midnight lows, then you're going to get a big downward adjustment. The one or two cold spells of the month were cloudy and probably didn't behave in that way. The very mild 29th was left with a similar final value to provisional. So were the other mild spells. 

I was expecting a small downward correction too, just because the average of them seems to be 0.2 to 0.3. So it was unusual. Now we have a November at 8.3, and it will be the 21st warmest on record (1659 to 2018, 360 years of data). It was the 13th warmest of the past hundred or also since 1937. It will be the ninth warmest since 1977. And it will be the sixth warmest of the 21st century so far. The shorter the interval for comparison, the more ordinary it appears. It was warmer than 94% of the past 360, 85% of the past 80, 80% of the past 40, and 67% of the past 18. In fact, in the ten Novembers from 2009 to 2018, it was barely above average, and fifth warmest (the average of these ten being 7.5). 

Edited by Roger J Smith

Share this post


Link to post
Share on other sites

The thing about the 'adjustment' is that the final figure is taken from a different set of stations than the running figure.

Generally the set of stations used for the final figure produce a cooler CET than the running CET- in many ways the running figure is a very close estimate.

Nothing to do with urban warming. I know this because I emailed the Met Office about it a few years ago.

Interesting to get the rise after David Snow being so brazen and confident that there was going to be drop again.

8.3C is a lot warmer than many were predicting mid-month.

Share this post


Link to post
Share on other sites
11 hours ago, Scorcher said:

The thing about the 'adjustment' is that the final figure is taken from a different set of stations than the running figure.

Generally the set of stations used for the final figure produce a cooler CET than the running CET- in many ways the running figure is a very close estimate.

Nothing to do with urban warming. I know this because I emailed the Met Office about it a few years ago.

Interesting to get the rise after David Snow being so brazen and confident that there was going to be drop again.

8.3C is a lot warmer than many were predicting mid-month.

Their data is adjusted to allow for urban warming, have a look at their site. (so it says anyway);)

Share this post


Link to post
Share on other sites
12 hours ago, Scorcher said:

The thing about the 'adjustment' is that the final figure is taken from a different set of stations than the running figure.

Generally the set of stations used for the final figure produce a cooler CET than the running CET- in many ways the running figure is a very close estimate.

Nothing to do with urban warming. I know this because I emailed the Met Office about it a few years ago.

Interesting to get the rise after David Snow being so brazen and confident that there was going to be drop again.

8.3C is a lot warmer than many were predicting mid-month.

Thanks, I was sure that someone had done this at some point and shared it on here - now I know I wasn't going insane! But David Snow is correct that they subtract 0.2C off after all other adjustments due to urban warming (which presumably is why all corrections seem to be centred around -0.2C, more or less).

Edited by Relativistic

Share this post


Link to post
Share on other sites

The Hadley EWP tracker chart has reached its conclusion, 103 mm is the estimate for November.

http://hadobs.metoffice.com/hadukp/charts/charts.html

 

We'll get some official numbers in a day or two and reveal the scoring, but I am sure that Singularity will be our first contest year (for EWP) champion in both the scoring alternatives used. 

Share this post


Link to post
Share on other sites

I would love to be able to reveal the final results of the EWP contest (using NCIC this past contest year, we'll change back to Hadley only for 2018-19) but neither NCIC nor Hadley have updated tables yet. The Hadley tracker was showing 103 mm as a final value. If I go with 103.9 (NCIC is usually a bit higher than Hadley, it seems) then I have a final table including annual scoring. So assuming only small changes required in the November scoring, I can post it as soon as I know the outcome. 

It's easy to confirm however that Singularity has won the contest as there are no outcomes, even preposterous ones, that alter the scores enough to bring anyone up the table to within range. I think I know who finished second but we'll keep that a secret until we know for sure, however, our consensus moved up the table to second place as well, and the two normals made upward progress to finish 10th (1981-2010) and 12th (1988-2017). That won't bump any forecaster ranks down, there are 10th and 12th place human participants too. 

Also the Hadley numbers will likely give the same general order of finish. It may be that I'll have both of these alternate scoring tables ready within minutes of each other and we can tie a ribbon around this contest. Sorry about the delay but it's out of my hands. 

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

×
×
  • Create New...