2023 CET - all 12 months above 1961-1990 average?

[Summer8906]

On a knife-edge I guess depending on whether this month is Atlantic-dominated or whether we get another easterly in; models suggest either could occur.

As I said above it would certainly be ironic if such a dismal, wet year takes this record. Given that much of the summer was cool by day and dull, and July/Aug were almost completely devoid of 30C+ temps with even 25C+ rare, it shows the power of mild nights in influencing the CET. Or perhaps it just shows how anomalously cold the 60s were (which are of course part of the 1961-90 period).

Edited December 3, 2023 by Summer8906

[WYorksWeather]

4 hours ago, Summer8906 said:

On a knife-edge I guess depending on whether this month is Atlantic-dominated or whether we get another easterly in; models suggest either could occur.

As I said above it would certainly be ironic if such a dismal, wet year takes this record. Given that much of the summer was cool by day and dull, and July/Aug were almost completely devoid of 30C+ temps with even 25C+ rare, it shows the power of mild nights in influencing the CET. Or perhaps it just shows how anomalously cold the 60s were (which are of course part of the 1961-90 period).

To be fair, even if we look at mean maxima, the anomaly for the year is 1.78C. June was a very warm month for the CET region, average max of 22.6C, and the only months that were near average were July (20.1C, anomaly -0.2C) and August (20.5C, anomaly 0.4C). All other months this were at least 1C above 1961-1990 for max CET.

I think it's more the timing - the months that have done the heavy lifting were not the summer months, with the exception of June. The three largest monthly anomalies were June, September and February, which gives the year quite an odd feel.

I think if you swapped September with July and February with August, it would feel much more like a year deserving of breaking this record.

To be honest though - it is kind of inherent in the way this one works. It rewards a consistently mild year, rather than an exceptionally hot one. You only need one cold month to scupper it.

What would feel like more of a travesty would be if this year dethrones last year for the highest mean CET on record, but it's very unlikely as we'd need a 7.6C December.

[Summer8906]

4 hours ago, WYorksWeather said:

To be fair, even if we look at mean maxima, the anomaly for the year is 1.78C. June was a very warm month for the CET region, average max of 22.6C, and the only months that were near average were July (20.1C, anomaly -0.2C) and August (20.5C, anomaly 0.4C). All other months this were at least 1C above 1961-1990 for max CET.

I think it's more the timing - the months that have done the heavy lifting were not the summer months, with the exception of June. The three largest monthly anomalies were June, September and February, which gives the year quite an odd feel.

I think if you swapped September with July and February with August, it would feel much more like a year deserving of breaking this record.

To be honest though - it is kind of inherent in the way this one works. It rewards a consistently mild year, rather than an exceptionally hot one. You only need one cold month to scupper it.

What would feel like more of a travesty would be if this year dethrones last year for the highest mean CET on record, but it's very unlikely as we'd need a 7.6C December.

Hard to believe the maxima were consistently above average as certainly a good few months "seemed" average or cool, notably March, April, and (based on second hand info and reports of many sub-20C days in my area; I was away) July and August. What was notable about all four of those months is that they almost completely lacked any notably warm days - I guess there were a lot of days in March/April around a degree above average - not really enough to notice the mildness - and very few below.

It looks like July and August were especially bad in my area though, with the temp maps for those months showing a pronounced cool, cloudy and wet anomaly for central southern England, more so than some other areas, particularly near the coast.

Also the constant cloudiness doubtless pulled the "feel like" temps down in the March-August period, so many days might have felt cool when they were actually slightly milder than average.

I guess this year has been very anomalously southwesterly - perhaps one of the most southwesterly years on record, which would mean mild but also very dull and, for much of the country, wet.

 

 

 

 

 

 

Edited December 3, 2023 by Summer8906

[WYorksWeather]

4 minutes ago, Summer8906 said:

Hard to believe the maxima were consistently above average as certainly a good few months "seemed" average or cool, notably March, April, and (based on second hand info and reports of many sub-20C days in my area; I was away) July and August. What was notable about all four of those months is that they almost completely lacked any notably warm days - I guess there were a lot of days in March/April slightly, but not much, above average and few below.

It looks like July and August were especially bad in my area though, with the temp maps for those months showing a pronounced cool, cloudy and wet anomaly for central southern England, more so than some other areas, particularly near the coast.

Also the constant cloudiness doubtless pulled the "feel like" temps down in the March-August period, so many days might have felt cool when they were actually slightly milder than average.

I guess this year has been very anomalously southwesterly - perhaps one of the most southwesterly years on record, which would mean mild but also very dull and, for much of the country, wet.

 

 

 

 

 

Yep, there were parts of the country that were below average. The other thing is 1961-1990 is quite a cool average. I think most of us in our heads when thinking about weather compare to the last few years as a rule, and using that comparison, July and the first half of August would probably come out quite poor.

And as you say, other factors like above average rainfall, and low sun hours don't help.

[Summer8906]

4 hours ago, WYorksWeather said:

What would feel like more of a travesty would be if this year dethrones last year for the highest mean CET on record, but it's very unlikely as we'd need a 7.6C December.

Could happen I guess (e.g could be achieved with a mean max of around 10C and  min 5C, conceivable if we get a prolonged southwesterly spell) but also still some good signs of blocking returning on the models.

[WYorksWeather]

Forgot to do the update, so based on today. I still think any mild spell is too uncertain, so no update to the short-range forecast idea.

Using the data we have:

Today's average: -1.0C (2 days)

10-year average for rest of December: (-1.0 * 2 + 5.9 * 29) / 31 = 5.5C

30-year average:  (-1.0 * 2 + 4.9 * 29) / 31 = 4.5C

We're now in a position where we already need a milder spell than the 30-year average to cancel out the early cold. This pattern is likely to continue for the next day or two, and following that it will probably become possible to have a better guess at where we'll be by mid-month.

Edited December 3, 2023 by WYorksWeather

[LetItSnow!]

2 hours ago, Summer8906 said:

Hard to believe the maxima were consistently above average as certainly a good few months "seemed" average or cool, notably March, April, and (based on second hand info and reports of many sub-20C days in my area; I was away) July and August. What was notable about all four of those months is that they almost completely lacked any notably warm days - I guess there were a lot of days in March/April around a degree above average - not really enough to notice the mildness - and very few below.

It looks like July and August were especially bad in my area though, with the temp maps for those months showing a pronounced cool, cloudy and wet anomaly for central southern England, more so than some other areas, particularly near the coast.

Also the constant cloudiness doubtless pulled the "feel like" temps down in the March-August period, so many days might have felt cool when they were actually slightly milder than average.

I guess this year has been very anomalously southwesterly - perhaps one of the most southwesterly years on record, which would mean mild but also very dull and, for much of the country, wet.

 

 

 

 

 

 

I am willing to make a solid bet with money that when we have average months that feel cool, it's not because they were that cool but that we are used to above average conditions. Like, November 2023 felt like a normal November to me and that makes sense because it was near normal to the 1991-2020 average, but about a degree above the old average (the all time its closer to 1.2-1.3C above). An average November would feel chilly to me, like how I remember November 2019 felt like a pretty cold one at the time but in a wide frame it was about average. A C.E.T. in the low 6s now is equivalent of a C.E.T. in the upper 4s/low 5s in a historical context. The same applies for the summer. The new average July is about 16.8 so July 2023 would be about 0.7C below average. That's the equivalent of 15.3C in the historical context and would be generally a cool feeling month. July 2020 would have come out in this region if not for the last two days. That anomaly in a historical context would be a July of about 14.9C. I find it interesting to put 1991-2020 anomalies onto older anomalies to see what these months may have produced in times gone by. A crude method, but interesting nonetheless.

[WYorksWeather]

Update based on data up to 3rd December

Today's average: 0.1C (3 days)

10-year average for rest of December: (0.1 * 3 + 5.9 * 28) / 31 = 5.3C

30-year average:  (0.1 * 3 + 4.9 * 28) / 31 = 4.4C

Another drop today - we're increasingly in need of a milder pattern if this record is going to go.

Update based on short term projection up to 10th December

Looking at the GFS tables today from Meteociel, I now feel that there's enough confidence to extend the short term projection out to the 10th December.

I'm going to make an initial estimate of around 4C today (4th Dec), 2C tomorrow, 4C on the 6th, then I'm going to go out on a limb and say that the major warm up will come off over the latter half of this week, with 9C averages from the 7th-10th. If you put that all together, you get an average of 46.3 / 10 or 4.6C for the first 10 days of December.

So, using this, we get the following figures, based on 21 remaining days after the 10th.

10-year average for rest of December: (4.6 * 10 + 5.9 * 21) / 31 = 5.5C

30-year average:  (4.6 * 10 + 4.9 * 21) / 31 = 4.8C

Summary

In short, based on the data we have, the average is continuing to drop. However, there is increasing confidence that a significantly milder spell will occur towards the end of this week, and hence we expect the CET to recover to more normal levels. I would say it is still too uncertain to say either way beyond the 10th, so I'll hold my subjective probability of the record going at 60% for now.

[SollyOlly]

Accepting that people have their own views and preferences, and I know it's my hobby horse, but the WMO say: "WMO uses 1961-1990 as the internationally agreed base period for measuring long-term climate change."

I find it useful and interesting to see the difference, and as others have eloquently pointed out, the rolling average in a warming world is only going to go one way...

Definitely agree that "I am willing to make a solid bet with money that when we have average months that feel cool, it's not because they were that cool but that we are used to above average conditions", the recency effect is entirely natural and that's why having some form of constant baseline is useful, at least in a scientific sense.

Very interesting thread, btw!

[LetItSnow!]

5 hours ago, SollyOlly said:

Accepting that people have their own views and preferences, and I know it's my hobby horse, but the WMO say: "WMO uses 1961-1990 as the internationally agreed base period for measuring long-term climate change."

I find it useful and interesting to see the difference, and as others have eloquently pointed out, the rolling average in a warming world is only going to go one way...

Definitely agree that "I am willing to make a solid bet with money that when we have average months that feel cool, it's not because they were that cool but that we are used to above average conditions", the recency effect is entirely natural and that's why having some form of constant baseline is useful, at least in a scientific sense.

Very interesting thread, btw!

Not sure why anyone would disagree with this. I’ve previously stated that I believe that two averages should be used concurrently - A 1901-2000 average and the most recent 30 year average. It keeps things in perspective. 

[TillyS]

My brief take on the topic of baselines is that once it’s been done, it’s done. You can’t really have a moving baseline otherwise it doesn’t measure climate change (hot or cold). It becomes pointless, which is why I personally find summer sun’s posting of the 1990-2010 mean a little confusing for the proper stats.

The real issue isn’t the fact that we’re using 1961-1990 now.

It’s why they chose that particular 30 year period in the first place? That’s where the thing plays into conspiracists’ hands because you could hardly have chosen a colder 30 year mean benchmark in the last hundred years. It almost looks like they deliberately picked the coldest possible 30 year mean to prove their preconceived belief in AGW (which I do think is genuine by the way but that’s not the point).

It looks well fishy and I’m not referring to Michael.

Edited December 5, 2023 by TillyS

[reef]

3 hours ago, TillyS said:

My brief take on the topic of baselines is that once it’s been done, it’s done. You can’t really have a moving baseline otherwise it doesn’t measure climate change (hot or cold). It becomes pointless, which is why I personally find summer sun’s posting of the 1990-2010 mean a little confusing for the proper stats.

The real issue isn’t the fact that we’re using 1961-1990 now.

It’s why they chose that particular 30 year period in the first place? That’s where the thing plays into conspiracists’ hands because you could hardly have chosen a colder 30 year mean benchmark in the last hundred years. It almost looks like they deliberately picked the coldest possible 30 year mean to prove their preconceived belief in AGW (which I do think is genuine by the way but that’s not the point).

It looks well fishy and I’m not referring to Michael.

I think in an ideal world you'd use 1901-2000 and then the most recent 1991-2020 mean alongside it. The issue is, there aren't really enough local station records going that far back.

I think looking at this chart of the CET, 1961-1990 is a pretty decent representation of the 20th century. It was a bit cooler from 1900-30, slightly warmer for the next 30 years, cooler from 1960-80 and then starting to climb much more after 1990.

[TillyS]

38 minutes ago, reef said:

I think in an ideal world you'd use 1901-2000 and then the most recent 1991-2020 mean alongside it. The issue is, there aren't really enough local station records going that far back.

I think looking at this chart of the CET, 1961-1990 is a pretty decent representation of the 20th century. It was a bit cooler from 1900-30, slightly warmer for the next 30 years, cooler from 1960-80 and then starting to climb much more after 1990.

Yes, but I ‘think’ without going into depth that 1961-1990 is the coldest 30 year mean of the past 100 years? Perhaps longer. I could be wrong. And I’m not a statistician.  

@Roger J Smith ?

 

Edited December 5, 2023 by TillyS

[reef]

3 hours ago, TillyS said:

Yes, but I ‘think’ without going into depth that 1961-1990 is the coldest 30 year mean of the past 100 years? Perhaps longer. I could be wrong. And I’m not a statistician.  

@Roger J Smith ?

 

I think that would be 1951-80 by a small amount, but I'm not certain.

[SollyOlly]

7 hours ago, TillyS said:

My brief take on the topic of baselines is that once it’s been done, it’s done. You can’t really have a moving baseline otherwise it doesn’t measure climate change (hot or cold). It becomes pointless, which is why I personally find summer sun’s posting of the 1990-2010 mean a little confusing for the proper stats.

The real issue isn’t the fact that we’re using 1961-1990 now.

It’s why they chose that particular 30 year period in the first place? That’s where the thing plays into conspiracists’ hands because you could hardly have chosen a colder 30 year mean benchmark in the last hundred years. It almost looks like they deliberately picked the coldest possible 30 year mean to prove their preconceived belief in AGW (which I do think is genuine by the way but that’s not the point).

It looks well fishy and I’m not referring to Michael.

I see your point, but in honesty the conspiracists are going to latch on to anything, they'd find something to support their world view whatever the situation. On a personal note, I don't really have an entrenched view, but if the WMO recommend it then I'm happy enough to go along with it - they are more expert than I am on this! Equally, using a century's data, or having the WMO baseline as well as looking at the 30 year rolling average is all good...in any case, the UK is getting warmer and wetter, sometimes very much so. Which on a personal level I do lament a little bit, as I remember 'proper' seasons in my own lifetime and can see and feel the change all around.

[WYorksWeather]

14 hours ago, TillyS said:

My brief take on the topic of baselines is that once it’s been done, it’s done. You can’t really have a moving baseline otherwise it doesn’t measure climate change (hot or cold). It becomes pointless, which is why I personally find summer sun’s posting of the 1990-2010 mean a little confusing for the proper stats.

The real issue isn’t the fact that we’re using 1961-1990 now.

It’s why they chose that particular 30 year period in the first place? That’s where the thing plays into conspiracists’ hands because you could hardly have chosen a colder 30 year mean benchmark in the last hundred years. It almost looks like they deliberately picked the coldest possible 30 year mean to prove their preconceived belief in AGW (which I do think is genuine by the way but that’s not the point).

It looks well fishy and I’m not referring to Michael.

My assumption on the 1961-1990 average is that this was set as the baseline back in the early 90s, when the CET daily was published. See this link to the first paper on daily CET: https://hadleyserver.metoffice.gov.uk/hadcet/Parker_etalIJOC1992_dailyCET.pdf. I don't think there's any reason to suspect any other reason - Parker et. al. would simply have used the latest data available to them at the time.

There are broadly three things you could be trying to do in setting the baseline, and here's my view on what might work best in each case:

1. To compare change over time as the dataset evolves (in which case 1961-1990 makes sense, as it's the original)
2. To look at the impact of AGW (in which case 1851-1900 makes sense, for consistency with various global temperature datasets, and to give a much broader outlook on the pre-industrial, and compare to today)
3. To look at temperatures in comparison to a recent baseline, to minimise any warming impacts and just look at deviation from the recent mean (in which case I'd say you could use an IPCC-style 20-year climate average, and update it every year, so we'd currently be on 2003-2022, and update to 2004-2023 on January 1st).

 

 

[WYorksWeather]

Just catching up with this post after a few days away. My short-term projection based on an average of 4.6C up to the 10th still looks sensible, and too early for any clarity beyond the weekend, or maybe early next week. I'll wait a couple more days before trying to predict where we'll be by mid-month, hence this update will just use the data we have.

Update based on data up to 6th December

Today's average: 1.9C (6 days)

10-year average for rest of December: (1.9 * 6 + 5.9 * 25) / 31 = 5.1C

30-year average:  (1.9 * 6 + 4.9 * 25) / 31 = 4.3C

Forecasted prediction still dropping, but a little more slowly. That will start to change over the next few days. I will be watching to see if my benchmark of 4.6C up to the 10th is reached or exceeded, and if we start to see a clearer way forward with the temperatures beyond that.

Edited December 7, 2023 by WYorksWeather

[Roger J Smith]

If we accept that the GFS 16-day is likely to be within 1 C, then by 24th the CET would be 4.5 give or take. That would leave the question in the hands of the currently unknown outcome of the last week of December. It seems unlikely that the question will be settled before Christmas, it could go either way.

[WYorksWeather]

11 hours ago, Roger J Smith said:

If we accept that the GFS 16-day is likely to be within 1 C, then by 24th the CET would be 4.5 give or take. That would leave the question in the hands of the currently unknown outcome of the last week of December. It seems unlikely that the question will be settled before Christmas, it could go either way.

Definitely still quite uncertain. But generally I do feel that GFS tends to under-read. Today, for instance, GFS was going for a temperature map like this by 1pm. Pretty much as close to a nowcast as you can get, only 6 hours ahead:

And yet if we look on Meteociel for live obs, as of my time of posting (12.20pm):

Clear that the band of double digit temperatures intrudes into more of the south, and instead of temperatures near 5C for Newcastle etc. it's more like 7C.

However, under certain circumstances, GFS can get it wrong the opposite way, for example in exceptionally dry conditions. Some GFS runs near to July 19th 2022 were throwing out 43C, for instance. But I think generally the tendency in most cases is to underdo the temperatures.

 

[WYorksWeather]

Update based on data up to 7th December

Today's average: 2.2C (7 days)

10-year average for rest of December: (2.2 * 7 + 5.9 * 24) / 31 = 5.1C

30-year average:  (2.2 * 6 + 4.9 * 24) / 31 = 4.2C

Very small additional drop today, less than 0.1C. All eyes now on how much the next three or four days is able to raise the CET by, before we probably go into a holding pattern (if the forecast UK high does take hold). Still sticking with 4.6C up to the 10th, and will provide a further short range prediction after that.

[WYorksWeather]

CET today rising rapidly as projected, up to 3.6C as of the 9th. I think we may not be too far off from my 4.6C when data for the 10th is released tomorrow, so all in all pretty happy with my short-term projection.

Update based on data up to 9th December

Today's average: 3.6C (9 days)

10-year average for rest of December: (3.6 * 9 + 5.9 * 22) / 31 = 5.2C

30-year average:  (3.6 * 9 + 4.9 * 22) / 31 = 4.5C

Definitely showing the predicted rise - even a 30-year average December now close to getting us to the required 4.6C.

In terms of the short-term projection, where we go from here is still up for grabs. I need to have a guess though, so let's do short-term up to the 14th, as the position of the UK high is still quite uncertain. I'll assume we reach 4.6C tomorrow as per my previous prediction, and then reasonable guesses based on UKV would be 7C for Monday, 6C Tuesday, 6C Wednesday, possibly as low as 3C on Thursday. Averaging that out, we'd get (4.6 * 10 + 7 + 6 + 6 + 3) / 14, which gives us a prediction of 4.9C up to the 14th.

If that were to come off, here's where we'd be:

Prediction based on 4.9C up to 14th December

10-year average for rest of December: (4.9 * 14 + 5.9 * 17) / 31 = 5.4C

30-year average:  (4.9 * 14 + 4.9 * 17) / 31 = 4.9C

This would of course be enough that even a 30-year average December would see the record go, and the 10-year average would beat it quite handily.

Subjectively, I think this is enough to raise my probability of the record being broken this time to 70%. We are now very much reliant on a cold end to the year - anything approaching standard fare will mean we cruise to the record.

 

[WYorksWeather]

Hm... data up to the 10th came out a bit lower than I thought at 4.1C. I'll still stick with 4.9C by the 14th even though it probably won't happen, since I've committed to it now and I want to be transparent with my guesses.

In any case, based on today's data, here's the latest.

Update based on data up to 10th December

Today's average: 4.1C (10 days)

10-year average for rest of December: (4.1* 10 + 5.9 * 21) / 31 = 5.3C

30-year average:  (4.1 * 10 + 4.9 * 21) / 31 = 4.6C

Now enough gains made on the CET for even a 30-year average rest of the month to take us above the 1961-1990 average. Still some uncertainty around a late cold spell so it's not over yet.

[al78]

On 05/12/2023 at 02:12, LetItSnow! said:

Not sure why anyone would disagree with this. I’ve previously stated that I believe that two averages should be used concurrently - A 1901-2000 average and the most recent 30 year average. It keeps things in perspective. 

A rolling 30-year average better represents the current climate when there is a long term warming trend. Using a 1901-2000 average the probability of above an above-average monthly temperature is likely well over 50%, which leads to the "what are you complaining about the CET was above average" responses to complaints about a summer month with double the normal rainfall. For my seasonal forecasting work (tropical cyclones) I am using 1991-2020 climatologies.

[WYorksWeather]

4 hours ago, al78 said:

A rolling 30-year average better represents the current climate when there is a long term warming trend. Using a 1901-2000 average the probability of above an above-average monthly temperature is likely well over 50%, which leads to the "what are you complaining about the CET was above average" responses to complaints about a summer month with double the normal rainfall. For my seasonal forecasting work (tropical cyclones) I am using 1991-2020 climatologies.

To refer back to my earlier post though, it depends what the purpose of your baseline is. Deviation from recent norm, measuring the effect of AGW, or measuring from when you started the dataset are all reasonable ways of doing it.

In terms of probability, in the last ten years or so at least 80% of months have been above the 1961-1990 average almost every year. Depending on your perspective, that either makes the average useless (if you're interested in temperatures relative to recent norms) or makes it very useful (to show how much our climate has changed since the 60s, 70s and 80s).

I would note also that things are moving quickly enough that even in comparison to the 1991-2020 average, there's probably still about a 2:1 ratio of warmer and cooler than average months in recent years.

[LetItSnow!]

7 hours ago, al78 said:

A rolling 30-year average better represents the current climate when there is a long term warming trend. Using a 1901-2000 average the probability of above an above-average monthly temperature is likely well over 50%, which leads to the "what are you complaining about the CET was above average" responses to complaints about a summer month with double the normal rainfall. For my seasonal forecasting work (tropical cyclones) I am using 1991-2020 climatologies.

That’s all well and good but people complaining on a forum like that isn't relevant to me for a climate analysis. As stated, I think both should be used. Our climate is changing so fast. A 3.0C January is about 1.5C below average nowadays, but that doesn’t make it a cold January in the wider scheme of things. Exactly the same as a 16.0C July being about 1C below average - it may feel “cool” but it isn’t in a recent historical context. It can be used by rampers with an agenda to paint a picture that a month has been abnormally below normal when it hasn’t.

A 1991-2020 average is a great tool for recent perspective but old values should remain in place to drive home the changing climate.