September CET: can we get below par from the mid way point?

[Stratos Ferric]

The inaugural post in this partition, and in response to the debate over in the CET thread.

Can we return a below par CET from a mid-point level of 15.2C?

The bald facts are these: of 129 years for which I have daily CET values, 5/33 at 15.2C or above mid-month have fallen to 13.7 or less. On the other hand, of 79 that finished at 13.7C or below, only 5 were 15.2C at mid way. What does this tell us? First of all, first halves as high as this one are in the minority historically, only around 25%; and whilst 1/6th of these have fallen away sharply, cold months (sub-par by current standards) tend not to have started warm.

Crudely, therefore, we might suggest around a 15-20% chance of subpar from here, but:

Warm first halves have predominated in the last few years; five of the last eight have been as warm or warmer. The 25 years prior to this yielded just two.

Secondly, and arguably most critically, the five occasions when the requirement this year was previously met were - wait for it: 1886 (15.5-13.6); 1889 (15.8-12.8 ); 1899 (15.3-13.7); 1914 (15.8-13.3); and, 1919 (15.4-12.7).

Near misses in recent years include 2002 and 2003, which both fell from 15.1, to 14.4 and 14.3 respectively.

Eleven of the years from 1983 to 1996 came in below 13.7, but the highest mid-month starting points were: 14.9 in 1984; 14.7 in 1987; and, 14.8 in 1990.

The ten year rolling mean drop-off is 0.7C; 1.5C has been reached only once in the last 15 years, in 2005, which, interestingly, was warmer on the 15th than 2006 was (and at thatpoint the second warmest in this 129 year data set).

Taking this additional context into account, coupled with warmer waters to our NW (one source of any of the cold required to pull things down), the odds don't look favourable. I wouldn't rule it out absolutely, but the odds are less than 5% of returning a sub-par outturn.

If we do, whilst it wouldn't directly indicate anything regarding the coming winter, it would suggest, by recent standards, an unusual cool down.

[Stu_London]

Stratos - I hope you are not going to use this section as a platform for your mild ramping :lol: :lol:

Seriously, though the figures are interesting and yes in isolation they do make a sub-par month very much an outside bet.

What the stats do not do however is bring into play the immediate outlook (i.e. the reliable timeframe). If we do that then the falls due to occur this week (Monday to Wednesday) are more than enough to bring the 71-00 into the sights, although the warmer (albeit less certain) outlook from Wednesday onwards would appear to scupper the chances.

There is a reasonable chance the CET will come in under the ten year rolling mean which in itself should provide some comfort to the coldies given the warm start

[Stratos Ferric]

Stratos - I hope you are not going to use this section as a platform for your mild ramping :blink:

Seriously, though the figures are interesting and yes in isolation they do make a sub-par month very much an outside bet.

What the stats do not do however is bring into play the immediate outlook (i.e. the reliable timeframe). If we do that then the falls due to occur this week (Monday to Wednesday) are more than enough to bring the 71-00 into the sights, although the warmer (albeit less certain) outlook from Wednesday onwards would appear to scupper the chances.

There is a reasonable chance the CET will come in under the ten year rolling mean which in itself should provide some comfort to the coldies given the warm start

Yes and no. Whether or not this challenge is valid in a way can be assessed by these data themselves. On the face of it there is an allure in the argument that says 'but in two days' time we'll have eaten into the difference' so the requirement will be easier. This is akin to a road race I might run in where, at the half-way point, I am behind the leader; I put the foot on the metaphorical gas and close the gap, perhaps disproportionately, by the 3/4 mark, and I reason, as might a spectator, that if I carry on at this rate I'll cross the line in the lead. The question is whether or not that increased burn rate could be maintained, and therein, I think, lies the rub. At any season in the UK a two week run of cold weather is unusual; yes, we've just had a cool summer, but on average even that degree of coolness is less than we'd need from now in; and in any case, whilst the receny of that data gives it some weight, it is still very unusual. The final contrary argument is that the month is a cooling month. Indeed it is, but so it was on the many other occasions when, in this situation, it still didn't cool enough.

I would say that overall what these data say is that there's a lot of thermal inertia; if the month is set during a warm period it is harder to cool down to what are below long range average levels.

So, it's not impossible, just very unlikely still. All of that said, if we do get there, then it might presage a more interesting autumn than we've seen for three or four years.

[Stu_London]

Yes and no. Whether or not this challenge is valid in a way can be assessed by these data themselves. On the face of it there is an allure in the argument that says 'but in two days' time we'll have eaten into the difference' so the requirement will be easier. This is akin to a road race I might run in where, at the half-way point, I am behind the leader; I put the foot on the metaphorical gas and close the gap, perhaps disproportionately, by the 3/4 mark, and I reason, as might a spectator, that if I carry on at this rate I'll cross the line in the lead. The question is whether or not that increased burn rate could be maintained, and therein, I think, lies the rub. At any season in the UK a two week run of cold weather is unusual; yes, we've just had a cool summer, but on average even that degree of coolness is less than we'd need from now in; and in any case, whilst the receny of that data gives it some weight, it is still very unusual. The final contrary argument is that the month is a cooling month. Indeed it is, but so it was on the many other occasions when, in this situation, it still didn't cool enough.

I would say that overall what these data say is that there's a lot of thermal inertia; if the month is set during a warm period it is harder to cool down to what are below long range average levels.

So, it's not impossible, just very unlikely still. All of that said, if we do get there, then it might presage a more interesting autumn than we've seen for three or four years.

I can't disagree with what you have said especialy the highlighted bit.

As for the bit in italics, the models are hinted at another retrogression event / possible northerly which would keep the cool theme right until near the end of the month, however it is not in the reliable timeframe just yet.

[West is Best]

There is a reasonable chance the CET will come in under the ten year rolling mean which in itself should provide some comfort to the coldies given the warm start

Better than evens I'd have thought given that the 10 yr rolling mean is 14.8C. We should reach that in the next 24 hours, and the only way I can see it rising in the last week requires an absolute pinpoint precise positioning of an HP to our east drawing up southerlies.

I think the trend has been set: a cool summer and now a much cooler September than we've been used to. Whether we break the 1971-200 mean (which is all but identical to the 1961-1990 mean) remains to be seen: I'd say looking at the synoptics it's about 50:50, although my money is just on the below average punt.

This is akin to a road race I might run in where, at the half-way point, I am behind the leader; I put the foot on the metaphorical gas and close the gap, perhaps disproportionately, by the 3/4 mark, and I reason, as might a spectator, that if I carry on at this rate I'll cross the line in the lead. The question is whether or not that increased burn rate could be maintained, and therein, I think, lies the rub.

I'm a little surprised to see you writing this. Although it can be true that we often have yo-yo weather (a week warm a week cold), I've normally associated you much more with pattern change posts: where you argue for the sort of change that can give progressively mild or cold prolonged spells. I think since June we've been in a cooldown and that's continuing: these synoptic patterns are very similar to what we've been seeing.

So, I'd prefer the analogy of an ocean liner: once set in one direction it takes a long time to shift it.Smell the coffee Stratos? :blink:

Edited September 17, 2007 by West is Best

[Stratos Ferric]

Better than evens I'd have thought given that the 10 yr rolling mean is 14.8C. We should reach that in the next 24 hours, and the only way I can see it rising in the last week requires an absolute pinpoint precise positioning of an HP to our east drawing up southerlies.

I think the trend has been set: a cool summer and now a much cooler September than we've been used to. Whether we break the 1971-200 mean (which is all but identical to the 1961-1990 mean) remains to be seen: I'd say looking at the synoptics it's about 50:50, although my money is just on the below average punt.

Certainly agree that we'll be around about the 10 year mean.

Re the trend, another cut would be very cool July, less cool August, slightly above par September.

I still think you're hugely overplaying the chances of coming in under the 30 year mean though.

[Stu_London]

I still think you're hugely overplaying the chances of coming in under the 30 year mean though.

If it does happen (I'd say its chances have risen to about 30%), you have rather set yourself up with the coldies by inferring it could only possibly happen in rare cases in our old style climate

[Anti-Mild]

If it does happen (I'd say its chances have risen to about 30%), you have rather set yourself up with the coldies by inferring it could only possibly happen in rare cases in our old style climate

I usually disagree with SF but on this occasion I agree that it is hugely unlikely that we will have an average September. I would go so far as to suggest that we will not get below 14c.

[VillagePlank]

Hi SF,

Your inset chart reminds me very much of the sigmoid function

Strange that natural patterns so closely resemble a mathematical abstract, perhaps? Especially when it's not represented by things tending to infinity ....

Edited September 18, 2007 by VillagePlank

[Stratos Ferric]

Hi SF,

Your inset chart reminds me very much of the sigmoid function

Strange that natural patterns so closely resemble a mathematical abstract, perhaps? Especially when it's not represented by things tending to infinity ....

Yes, and a good spot, but then again perhaps it's not so strange because that function also describes the cumulative sum of a normal distribution: in large numbers nature tends to scatter around a mean - albeit a dynamic mean when viewed across time. From the perspective of species it maximises the chances of survival and also explains how and why species evolve. However, even in the physical sciences the same distribution often holds true.

[VillagePlank]

Yes, and a good spot, but then again perhaps it's not so strange because that function also describes the cumulative sum of a normal distribution: in large numbers nature tends to scatter around a mean - albeit a dynamic mean when viewed across time. From the perspective of species it maximises the chances of survival and also explains how and why species evolve. However, even in the physical sciences the same distribution often holds true.

Your representation of the data fitting so well with the sigmoid function (which, as you say, is the cumulative sum of the bell curve) should be a worry for AGW protaganists.

This, I hesitantly add (mainly because I know that it is entirely possible for this snippet of information to be completely misused), would seem to suggest that the varying degrees of temperature for Septmeber are well within normal bounds - this is, of course, not to say that they wouldn't and are not just about to exceed normal bounds, though.

Hence, although we see warming, we haven't seen enough to significantly skew the normal distribution?

Edited September 18, 2007 by VillagePlank

[Stratos Ferric]

Your representation of the data fitting so well with the sigmoid function (which, as you say, is the cumulative sum of the bell curve) should be a worry for AGW protaganists.

This, I hesitantly add (mainly because I know that it is entirely possible for this snippet of information to be completely misused), would seem to suggest that the varying degrees of temperature for September are well within normal bounds - this is, of course, not to say that they wouldn't and are not just about to exceed normal bounds, though.

Hence, although we see warming, we haven't seen enough to significantly skew the normal distribution?

I think we might return to this as a later piece of analysis. However, if you look at the CET trend plots (posted in several places recently and in danger of becoming as ubiquitous as OON, late of this parish, most recently in the analysis thread over in climate / environment) the fact that the ten year mean is racing ahead of the thirty year mean requires one of two scenarios to be prevalent - or a blend of both. Either, warm months, when they occur, are more warm than the cold months are cold; or, warm months are happening more frequently than cold months.

The mathematical concern for cold lovers in a warming climate is, therefore, that for warmth to be delivered we need either more warm months, or warmer months; these coupled with fewer and less cold months.

Yes, things can occur outside the normal range, but ordinarily 90% of events lie within 2SDs of the mean (ordinarily would exclude a moving mean as this should imply momentum in a particular direction, be that warming or cooling). I haven't got the time to run the numbers right now but it might provide an interesting analysis for the upcoming winter.

[snowsure]

Looking at Philip Eden's analysis of Sep 07 compared to Sep 06, would you agree that this month is both cooler and also cooling more than Sep 06?

[Stratos Ferric]

Looking at Philip Eden's analysis of Sep 07 compared to Sep 06, would you agree that this month is both cooler and also cooling more than Sep 06?

Well, given that last September was no.1 in the all time list we shouldn't be surprised that this month is cooler, and that much is very clear. I'd also agree that the cooling is more obvious: last September was remarkable, above almost anything else bar the temperature itself, for the consistency from start to finish.

[snowsure]

Well, given that last September was no.1 in the all time list we shouldn't be surprised that this month is cooler, and that much is very clear. I'd also agree that the cooling is more obvious: last September was remarkable, above almost anything else bar the temperature itself, for the consistency from start to finish.

Do you feel that Sep 06 was so remarkable? Look here for the Sep 06 and 05 chart. Now that cooling looks remarkable. Sep 05 was warmer by the mid-point than Sep 06.

However, I feel that I am prehaps concurring with your initial post i.e. We won't get the usual prolonged cooling in the latter parts of Sep 07. The only difference is the lower temperatures in the first half of the month.

The question is now: What is classed as normal for September? If we accept the rolling 10 year mean as normal then this month could be sub-normal.

[West is Best]

The question is now: What is classed as normal for September? If we accept the rolling 10 year mean as normal then this month could be sub-normal.

Indeed! This month will be well under the 10 yr mean of 14.8C.

[Stratos Ferric]

Do you feel that Sep 06 was so remarkable? Look here for the Sep 06 and 05 chart. Now that cooling looks remarkable. Sep 05 was warmer by the mid-point than Sep 06.

However, I feel that I am prehaps concurring with your initial post i.e. We won't get the usual prolonged cooling in the latter parts of Sep 07. The only difference is the lower temperatures in the first half of the month.

The question is now: What is classed as normal for September? If we accept the rolling 10 year mean as normal then this month could be sub-normal.

SS, not quite sure what you're querying or saying there. I agree '05 cooled, though I'm not sure anyone has said otherwise: likewise, as per my original point, '06 was very steady. As to 'what's normal', it doesn't exist as a point, it's a range, and this year is well within that range; last year clearly was not. We're always going to have months and years above and below the running mean, that data on its own is meaningless.

[snowsure]

SS, not quite sure what you're querying or saying there. I agree '05 cooled, though I'm not sure anyone has said otherwise: likewise, as per my original point, '06 was very steady. As to 'what's normal', it doesn't exist as a point, it's a range, and this year is well within that range; last year clearly was not. We're always going to have months and years above and below the running mean, that data on its own is meaningless.

Please can you clarify how the range is calculated (e.g. which years are used, etc)?

Is there an "industry standard" for such a range? Would I be able to talk about "the range" of CET to all learned people without further explaination?

As to whether your perception of my point being meaningless is valid I say this; Last year my net worth was less than this year. No amount of range comparison would dispute this point. It is, for me, better to look at year-on-year variances as this is able to give a real comparison with something concrete. As many people say; There are lies, damned lies and statistics. I prefer facts. Not that I am right or wrong. I am perhaps just less complex than others.

[Stratos Ferric]

Please can you clarify how the range is calculated (e.g. which years are used, etc)?

Is there an "industry standard" for such a range? Would I be able to talk about "the range" of CET to all learned people without further explaination?

As to whether your perception of my point being meaningless is valid I say this; Last year my net worth was less than this year. No amount of range comparison would dispute this point. It is, for me, better to look at year-on-year variances as this is able to give a real comparison with something concrete. As many people say; There are lies, damned lies and statistics. I prefer facts. Not that I am right or wrong. I am perhaps just less complex than others.

Well, I agree that in inexperienced hands statistics are open to abuse, but then so is a gun, or a car. Being able to mess with numbers is not the same as being able to make a robust statistical case.

My post was actually trying to understand preciesly what question you were originally posing: I may be being thick, but on my read it's not very clear what your point is.

Assuming we're talking abut the "range" of cooling, then you could take two treatments, which actually give little difference in net result. One is to look at the recent thirty years, the other to look at the full CET series, and take a mean, and then the standard deviation, of the difference between, say, the seven day mean temperature on the first of the month (centred on the 1st) and that for the 30th (again, centred on the 30th).

So, here's the picture you get.

The one impact that the smaller sample has is that the standard deviation of the sample will be smaller than would probably be the case for, say, 100 entries, but not hugely so. What the plot shows is the start-finish for each of the last 30 Septembers. I've added isolines to the chart (red dots) to show intervals at 1C; I've also added a shaded area to show the mean and +/- 1 standard deviation; in any normal population around 68% of data points should fall within this interval, and 95% should fall within 2 S.D.s - clearly the case here.

By any measure, the cooling in 2005 therefore, though fairly large, was not a big outlier and is within the "normal" range.

For those wondering, the warming year was 1985, which started with a cool burst and ended with early October well into the 20s. If anything, this is just an indication of the hasards of choosing a 7 day interval to calculate the means at either end.

...

As to whether your perception of my point being meaningless is valid I say this; Last year my net worth was less than this year. No amount of range comparison would dispute this point. It is, for me, better to look at year-on-year variances as this is able to give a real comparison with something concrete. As many people say; There are lies, damned lies and statistics. I prefer facts. Not that I am right or wrong. I am perhaps just less complex than others.

I'm sure it was, but it's not the same type of data. Your worth year-to-year is interrelated: this year's worth is dependent on last year's. This September's cooling is NOT dependent on last year's, we are talking about independent variables and the statistical treatment of them is necessarily different.

If we want to know whether cooling in a particular year is significant our only frame of reference is other similar events. A more meaningful comparison for your analogy would be 'is the change in my net worth this year significantly different to normal'?

Making a single year-on-year comparison is statistically about as useful as reading one page of War and Peace and expecting to divine the whole plot.

[snowsure]

Thanks for the post SF. Again, you help me to put into context my simple point of view. From your graphs, I see that the 2005 cooling was outside 75% the years regarding closing-opening. As I quite shambolically put it, that is a lot of cooling! What I do not understand is this; If 2005 was within a normal range, and it is outside where about 75% of the data is, then all the last 30 years have been normal. So, by your reckoning, there has been no unexpected change for September in the last 30 years. Why doesn't GW affect September?

Making a single year-on-year comparison is statistically about as useful as reading one page of War and Peace and expecting to divine the whole plot.

What if you read the pivotal page with all the re-counting of what happened? There is one!

I also feel that the industry standard point remains unanswered. Is it subjective?

[Stratos Ferric]

Thanks for the post SF. Again, you help me to put into context my simple point of view. From your graphs, I see that the 2005 cooling was outside 75% the years regarding closing-opening. As I quite shambolically put it, that is a lot of cooling! What I do not understand is this; If 2005 was within a normal range, and it is outside where about 75% of the data is, then all the last 30 years have been normal. So, by your reckoning, there has been no unexpected change for September in the last 30 years. Why doesn't GW affect September?

What if you read the pivotal page with all the re-counting of what happened? There is one!

I also feel that the industry standard point remains unanswered. Is it subjective?

Ah, hang on, we're confusing two things there. What the plots show is the monthly change, in each month: i.e how much does a September cool: this is much more a matter of energy flux with the apparent movement of the sun towards the horizon than climatic variability. There is no reason to suppose that with GW the rate of cooling would change in any particular month; it is just as likely that all that happens is that each month gets a degree or so warmer. For example, in stead of cooling from, say, 14 to 12C on average, we star to see 15 to 13C.

The 'industry standard' isn't in any way subjective, it's a statistical rule. What we choose to call an 'outlier' is up to us in a way, however, there is popular misconception that just because a data point appears to be a long way outside the norm, it is statistically significant (i.e. it isn't just a chance occurrence). That's why it's always best to look for patterns and trends, and hence why the debate about whether this summer marks the start of cooling is way too premature.

Back to 2005: yes, there was a lot of cooling, but it was far from being unprecedented, and was comfortably within the statistical bounds (unlike, say, the actual mean for September 2006). That said, if we recast the calculation of the opening and closing averages, and used a larger sample, then that assessment might change.

What if you read the pivotal page with all the re-counting of what happened? There is one!

Don't understand that: can you clarify? What 'pivotal page'? What 'recounting'?

[snowsure]

Ah, hang on, we're confusing two things there. What the plots show is the monthly change, in each month: i.e how much does a September cool: this is much more a matter of energy flux with the apparent movement of the sun towards the horizon than climatic variability. There is no reason to suppose that with GW the rate of cooling would change in any particular month; it is just as likely that all that happens is that each month gets a degree or so warmer. For example, in stead of cooling from, say, 14 to 12C on average, we star to see 15 to 13C.

But does that not equate to my net-worth point? The starting point of September surely influences that finishing temperature, doesn't it? And doesn't that rely on the August temperature? Or are you saying that climate doesn't have memory?

...

The 'industry standard' isn't in any way subjective, it's a statistical rule. What we choose to call an 'outlier' is up to us in a way, however, there is popular misconception that just because a data point appears to be a long way outside the norm, it is statistically significant (i.e. it isn't just a chance occurrence). That's why it's always best to look for patterns and trends, and hence why the debate about whether this summer marks the start of cooling is way too premature.

But if we continue to wait for a trend to form, the general opinion will be that the climate is, again, normal. Let's compare 1997-2007 with 1987-1997 and 1977-1987. I understand the role of a 10 year rolling average but it somehow seems inadequate in being diagnostic of current weather. A CET of -4 oC for October 2007 would not show up for several months, by which time cooling could be well underway. You use of rolling data seems too cautious. Perhaps I am too much of a knee-jerk reactionary!

Don't understand that: can you clarify? What 'pivotal page'? What 'recounting'?

Your post said that using individual data was like trying to ascertain the plot of War and Peace from one individual page. I suggest that the correct page may elicit much of the plot to a reader. I do not know if such a page exists within the mighty book!

[Stratos Ferric]

But does that not equate to my net-worth point? The starting point of September surely influences that finishing temperature, doesn't it? And doesn't that rely on the August temperature? Or are you saying that climate doesn't have memory?

But if we continue to wait for a trend to form, the general opinion will be that the climate is, again, normal. Let's compare 1997-2007 with 1987-1997 and 1977-1987. I understand the role of a 10 year rolling average but it somehow seems inadequate in being diagnostic of current weather. A CET of -4 oC for October 2007 would not show up for several months, by which time cooling could be well underway. You use of rolling data seems too cautious. Perhaps I am too much of a knee-jerk reactionary!

Your post said that using individual data was like trying to ascertain the plot of War and Peace from one individual page. I suggest that the correct page may elicit much of the plot to a reader. I do not know if such a page exists within the mighty book!

Some interesting points there.

Dismiss climate for a moment because it's a confusing term: let me reinterpret that slightly to short cycle weather (say 2-3 years, climate is a much longer average): yes, there is some memory, but even so, in the equinoxial months, the change in radiation will be the overriding factor. September will almost always cool (the one month in my plot is an abberation brought about, at least in part, by my choice of start and finish average: if I extended the calc to, say, 14 days, I suspect that the picture would be different - see footnote). Your point is, if summer is warmer will September cool more, or perhaps less, depending on flux / inertia. The answer to that is that it still depends. Temperature is an expression of energy flow, and the temperature we measure is reradiated energy. In principle a warm day cools faster than a cold one, but only if the air above is also cool, and there is nothing else to moderate the cooling. What applies diurnally can still apply across a whole month - hence September 2006 when cloud often conspired to lock overnight loss in, and mild humid air masses further reduced the flux. It would be interesting to test the hypothesis that warmer months preceding lead to larger falls, and on the face of it this MUST be true to an extent, because climate is, in the short term, fairly inelastic (i.e. no matter how hot summer is winter is still cool) - that said, I am also of the opinion that short term 'sub-climate' (by which I mean the 5-10 year mean of the weather) does have the type of memory you're alluding to. You can see in the analysis in the climate change section from a plot I provided there that cold winter months mainly come in clusters, and they do so when the CET is low (surprise surprise). This latter suggests, then, that there is only so much cooling a year can do: to take your net value analogy, you can only spend so much money, so if your worth increases unusually one year, you will finish the year better off than you entered it.

Re trend, I'm afraid we just have to be patient. I was making a photo montage with the kids at the w/e, and we were careful not to stick any photo down until we had precisely positioned one or two layers further up the page. Waiting for a trend is like that; we can't lock down the direction for sure until one or two steps beyond have been taken; and even then, there are wheels within wheels within wheels. It's not quite true that a one off variation in temperature takes a long time to appear in an average, it depends what is falling off at the opposite end. Go see the plots I've produced in the climate section (makes me scream thread I think) showing the 1-10-30-100 year CET. IN 1740 the CET falls off a cliff on the ten year. The point is, that drop occurs not because there was one cold month within context, but a whole series. If your October '4 under' was followed by November '4 over', then all we know is that the weather is oscillating violently, what statisticians refer to as an oscillating mean (i.e. the mean bounces around short-term within long-term bounds - CET did this for a long time until the late 80s).

Re the one page; if such existed then there would be no novel! Thus it is with climate: we cannot divine climate from a single month, any more than I can watch a car pass by on the A1 and know where it's from, or where it's going. All we have is one data point. It might be a noteworthy data point, at which stage we'd form a hypothesis: hence WiB's ponder (with one or two others) as to whether this summer marks the top of current warming. That analysis is in train and might make an appearance by the end of the w/e.

[VillagePlank]

what statisticians refer to as an oscillating mean (i.e. the mean bounces around short-term within long-term bounds - CET did this for a long time until the late 80s).

Yes, where the mean is also known as a strange attractor (see here) The Lorenz Attractor mentioned in this article was directly dervied from trying to model specific aspects of the atmosphere, too.

Edited September 26, 2007 by VillagePlank

[snowsure]

... Temperature is an expression of energy flow, and the temperature we measure is reradiated energy.

Temperature is a measure of the kinetic energy, or total internal energy, of molecules within matter. Nowt to do with energy flow.

Other than that, a most welcome reply. Again, thanks for educating me/us!

Out of interest, if we had 4 consecutive months of below average CET, would that make you question the trend?