Are We Still Capable Of Getting A <1c Cet Month?

[J10]

Applying the CET Figures From the Meto website and putting into Excel and creating a chart from those figures for the period (1960 to 2006 inclusive) it gives the formula

y= 0.033x + 3.9919

This give a gradient of 0.033c per year or 3.3c per century, or approximately 1.0c per 30 years, pretty much confirming the above analysis.

So From a base of approximately 4.0c in 1960, we should reach 6.0c in 2020, assuming the current rate of growth continues.

However, the rate of growth is slowing, with a formula of

y= 0.0366x + 3.938

For the period (1960 to 1999 inclusive) a rise of 1.0c in 27 years is suggested

A few final stats.

1960 to 1969 Average CET (Jan, Feb, Mar, Dec) 4.0c

1970 to 1979 Average CET (Jan, Feb, Mar, Dec) 4.7c

1980 to 1989 Average CET (Jan, Feb, Mar, Dec) 4.5c

1990 to 1999 Average CET (Jan, Feb, Mar, Dec) 5.4c

2000 to 2006 Average CET (Jan, Feb, Mar, Dec) 5.3c

Whichever way you look at it Winter Temps have risen over the past 45 years and continue to be much higher than 1960 levels.

[summer blizzard]

This is a very interesting discussion and despite the data which was posted, i do feel that a <1C month is still possible , this is because i believe that the majority of the current warming period is natural, as we have been in a Gliessburg Maxima since the late 1950's, which means that we have only had the eleven year sunspot cycles to rely on for cold periods, this means that to me, the question is whether or not these periods can deliver, if not we may have to wait until after 2011, before the Gliessburg Minima takes effect.

I believe that it is entirely possible that can have a <1C month however i must agree that we do not seem to be able to get all the ingrediants to come together and as such, i do not expect such a month until the Gliessburg Minima is taking effect however i do feel that we can come close and as such, my preliminary CET predictions for the coming winter period are:

November: 3.9C

December: 3.4C

January: 2.6C

February: 3.1C

March: 2.6C

[Snowly does it]

As others have said, the question is perhaps wrong.

It is possible to get a month below 1c, that is not in question really, the mere fact that Ice Days have been recorded this winter past is evidence that the right set up would give this result.

Is it likely? No, it never was likely outside of an ice age.

Will it happen in the next ten years? Were I a bookie I would offer very short odds on it simply on the basis of probability and spread. A sub 1c month would prove nothing and support nothing in isolation, it would just be something nice for the snowies amongst us. Three on the bounce or three in three consecutive years.... now that is more of a talking point and more of an outside chance, that would tend to herald something unexpected.

[Guest Mike W]

I don't personally think we are capable of getting a below 2 CET let alone below 1, even to get a below 3 is going to be very hard by the looks of things. I can't see anythin out their to set up a cooling yet.

Edited August 21, 2006 by Mike W

[Snowly does it]

I don't personally think we are capable of getting a below 2 CET let alone below 1, even to get a below 3 is going to be very hard by the looks of things. I can't see anythin out their to set up a cooling yet.

It doesn;t need to be a cooling though Mike, as a one-off it remains possible, just unlikely in any given winter month. To say we are 'incapable' of it assumes that the climate has changed beyond recognition. Whilst that may be its evental fate, it is not yet there.

[Weather-history]

There have only been 27 days with a daily CET mean that was subzero since 1st January 2000 to present

That compares to 56 such days from 1st January 1990 to August 1996

Edited August 21, 2006 by Mr_Data

[Stratos Ferric]

Does anyone know if we are receiving more than our "normal" share of N winds?

My observations appears to show a plethora of NW-N-NE winds whatever the season. (I am only speaking for one part of Doncaster so you may not agree. That is why I am wanting to know about the "bigger picture." )

Philip would have the data on that one, or you could check his site for the westerliness and southerliness indeces. My instinct is that northerlies have, if anything, declined in recent years, but that is purely subjective assessment after the event. I think NWlys are certainly far less frequent.

[Stratos Ferric]

As others have said, the question is perhaps wrong.

It is possible to get a month below 1c, that is not in question really, the mere fact that Ice Days have been recorded this winter past is evidence that the right set up would give this result.

Is it likely? No, it never was likely outside of an ice age.

Will it happen in the next ten years? Were I a bookie I would offer very short odds on it simply on the basis of probability and spread. A sub 1c month would prove nothing and support nothing in isolation, it would just be something nice for the snowies amongst us. Three on the bounce or three in three consecutive years.... now that is more of a talking point and more of an outside chance, that would tend to herald something unexpected.

SM, your assertion of "No, it never was likely outside of an ice age" is factually well wide of the mark, but I'm sure that's probably not what you meant. The plot below shows that there have been 82 such instances in the 340+ years of the CET series, so one per four winters or so. Three on the bounce is absolutely umprecedented and would be more than worthy of comment, particularly in the even larger teapot.

What this latest plot shows is that the current drought v.v. cold months is NOT unprecedented, however, as I have suggested over the past couple of years, if we don't get something very cold soon then we will be passing through the known bounds within which notions of climatic cold and warm in the UK have been defined during the measured record: out on the warm side as well, by the way, lest anyone is in any doubt. There's also a further caution in comparing previous intervals: in all instances there have either been many months which were close to 1C, and which would certainly be cold by today's standards, or else they occurred during periods when winter was relatively warm but years as a whole were not. The one exception, almost, is the period around the 1730s, but our current run of mild winters has already beaten the sequence we had then.

BFTP's analysis compounds the point. Statistically, however you carve the data, all the evidence says warming.

It doesn;t need to be a cooling though Mike, as a one-off it remains possible, just unlikely in any given winter month. To say we are 'incapable' of it assumes that the climate has changed beyond recognition. Whilst that may be its evental fate, it is not yet there.

In theory perhaps, but as TWS intimates there might have been a substantial enough change in the fundamentals to make even an ideal set-up struggle to deliver. Philip Eden made an interesting point to me last winter, relating to TWS's point about air mass modification, that the extent of winter sea ice around the GIN margin being far further north now than it was in the late 60s is perhaps allowing northerlies to be more modified by a warmer ocean surface. The sea may only be fractionally above freezing, but the density of water relative to air still means that there's huge potential for warming flux; on the face of it it certainly stacks up as a theory. When polar masses are starting anything up to 20C warmer than historically they have, you start to see both the problems for polar ice formation, and also the prospects of intense cold reaching the UK.

I don't think wheree quite yet at the point where it cannot happen, but where certainly stood in the doorway and right at the threshold. Whilst there are indications that recent warming is slowing, we would need to see a positive reverse in the trend if I'm not to believe, as I have asserted in here previously, that we're on a slippery slope towards UK winter (in the popular sense of the word) weather being confined intermittently at best to only higher ground in the north.

[pottyprof]

so what happened in a time before records??

to a point im still unconvinced that things are running away.. it only over the last 30 years that we can grow grapes again with enough success to make wine in the UK.. in the past we lost that ability.. and i feel we will lose it again.. we trust what we know in the short timescale that we know and i think that in doing this we dont see the planet as we should.. ill say again that after a major tsunami and quake that nudged our rotation.. we are now looking at different weather patterns.. there has been no other reason for this abrupt change within a couple of months of the event.. add to that the sun is now quiet after its final bow out from the last cycle..

yep that CET is achievable and i think sooner rather than later..

[VillagePlank]

It is certainly possible to have a sub-zero CET.

This, of course, will more likely be the result of successively cold nighttime minima, rather than lovely white scenes at Christmas.

Interestingly, the CET is an anomalous record. I think it is perhaps better to ask: 'Is there a four week period that has a sub-zero temperature mean?'

Another interesting measure, and one, that is not, as far as I am aware, available, is to have high granularity grid across the UK, and to count the sum of stations that observe subzero months (or preferably groups of four weeks) I think that this would show a more accurate picture of the warming significance in the UK.

But then that's the nature of statistics isn't it? I suppose comparing the same month year on year out on the CET series holds some meaning, but then when you compare differences from the solar mean time to the sidereal time, the differences in days between solar, star, and calendar months can be quite substantial (but not enough to move the mean +/- 0.5C) where, say, the end of the winter needs correcting as much as 24 hours (as we approach the 29th Feb in a leap year)

Of course, we still haven't corrected smoothly for daylength as timezones are a discrete human invention, when in reality, as we all know, 'timezones' are a smooth function of solar movement overhead.

Edited August 22, 2006 by Wilson

[drgl]

It is certainly possible to have a sub-zero CET.

This, of course, will more likely be the result of successively cold nighttime minima, rather than lovely white scenes at Christmas.

Interestingly, the CET is an anomalous record. I think it is perhaps better to ask: 'Is there a four week period that has a sub-zero temperature mean?'

Another interesting measure, and one, that is not, as far as I am aware, available, is to have high granularity grid across the UK, and to count the sum of stations that observe subzero months (or preferably groups of four weeks) I think that this would show a more accurate picture of the warming significance in the UK.

But then that's the nature of statistics isn't it? I suppose comparing the same month year on year out on the CET series holds some meaning, but then when you compare differences from the solar mean time to the sidereal time, the differences in days between solar, star, and calendar months can be quite substantial (but not enough to move the mean +/- 0.5C) where, say, the end of the winter needs correcting as much as 24 hours (as we approach the 29th Feb in a leap year)

Of course, we still haven't corrected smoothly for daylength as timezones are a discrete human invention, when in reality, as we all know, 'timezones' are a smooth function of solar movement overhead.

something i've been saying for months but apparently the UK is too small to have any significance

[Stratos Ferric]

so what happened in a time before records??

ill say again that after a major tsunami and quake that nudged our rotation.. we are now looking at different weather patterns.. there has been no other reason for this abrupt change within a couple of months of the event.. add to that the sun is now quiet after its final bow out from the last cycle..

yep that CET is achievable and i think sooner rather than later..

Have I missed something?

[Weather-history]

There have only been 27 days with a daily CET mean that was subzero since 1st January 2000 to present

That compares to 56 such days from 1st January 1990 to August 1996

And the 96 such days from 1st of January 1980 and August 1986

[Snowly does it]

SM, your assertion of "No, it never was likely outside of an ice age" is factually well wide of the mark, but I'm sure that's probably not what you meant. The plot below shows that there have been 82 such instances in the 340+ years of the CET series, so one per four winters or so. Three on the bounce is absolutely umprecedented and would be more than worthy of comment, particularly in the even larger teapot.

I don't think wheree quite yet at the point where it cannot happen, but where certainly stood in the doorway and right at the threshold. Whilst there are indications that recent warming is slowing, we would need to see a positive reverse in the trend if I'm not to believe, as I have asserted in here previously, that we're on a slippery slope towards UK winter (in the popular sense of the word) weather being confined intermittently at best to only higher ground in the north.

You are right, I worded that badly! I meant merely that whilst they occured historically they were never 'likely' in the sense that their occurance was less than usual, a one in four is not a good bet, but its a punt, thats what I meant!

I think the chances here in a GW world are really down to what happens to the jet longer term and on a winter to winter basis and (perhaps) to a slowing of the THC. Whilst this is a highly debatable subject and not something I would want to bet any money on or indeed back (yet) in any significant debate, IF the measured decline in its potency has any factual basis then we may find ourselves more susceptible to influence from traditionally colder areas - Siberia, Scandinavia, the Arctic on a more regular basis. This would drastically increase the chances of colder months. However what would then happen to a continuing warming world if nothing is done is a mystery to me.

Edited August 22, 2006 by snowmaiden

[Stratos Ferric]

It is certainly possible to have a sub-zero CET.

This, of course, will more likely be the result of successively cold nighttime minima, rather than lovely white scenes at Christmas.

Interestingly, the CET is an anomalous record. I think it is perhaps better to ask: 'Is there a four week period that has a sub-zero temperature mean?'

Another interesting measure, and one, that is not, as far as I am aware, available, is to have high granularity grid across the UK, and to count the sum of stations that observe subzero months (or preferably groups of four weeks) I think that this would show a more accurate picture of the warming significance in the UK.

But then that's the nature of statistics isn't it? I suppose comparing the same month year on year out on the CET series holds some meaning, but then when you compare differences from the solar mean time to the sidereal time, the differences in days between solar, star, and calendar months can be quite substantial (but not enough to move the mean +/- 0.5C) where, say, the end of the winter needs correcting as much as 24 hours (as we approach the 29th Feb in a leap year)

Of course, we still haven't corrected smoothly for daylength as timezones are a discrete human invention, when in reality, as we all know, 'timezones' are a smooth function of solar movement overhead.

Can't agree with any of that, I'm sorry.

First up very low night time minima alone would not be sufficient, though they would certainly be necessary, for a very cold month.

Second, you need to clarify what you mean by CET being an "anomalous" record. Taking your point about a "grid" oin the ensuing paragraph I'm assuming you're suggesting that it's somehow representative. It may or may not be, it depends what you're trying to use it to prove. The question as posed at the head of the thread was "are we still capable of getting a <1C CET month". It strikes me that in this regard, use of CET as the yardstick is absolutely correct.

Your point about whether or not a tighter grid of data would prove a different warming hypothesis is an interesting one, but one which almost certainly would not hold up in practice. I suspect that one reason why CET holds up as an accepted record is that it matches well trends in other longitudinal series. It would, for example, in a landmass as small as the UK, be unthinkable to suppose that whilst the CET is warming across a period, SE Scotland, say, was cooling. Day to day variations are quite possible as air mass boundaries move slowly enough for the 24hr period to be too short to allow for levelling, but over the longer terms these things tend to smooth out. The simple fact is that air mass parcels tend to be far larger than the UK (the same is not true for, say, mainland Europe or the US). Station by station analysis would, almost certainly, show precisely the same general pattern of warming.

Not quite sure what you're arguing re sidereal time, and the necessary corrections to astrological time required to allow a neat 24hr clock to work. Given that, viewed end to end, the sum total of experienced weather is exactly the same however you carve time up, the overall average for the data set would still be same. Individual packets of time, rebounded, may be slightly warmer or cooler arithmetically, but cooling in one packet would be offset by an equal and opposite adjustment in the next. Adding a day to a winter in the leap year provides at most an adjustment of around 1% of the difference in temperature on the extra day with that of the mean for the period as a whole. Assuming you take March 31st as the extra day (even though it's Feb 29th on the calendar), as this would have the largest expected variation from the period as whole, and assuming that that variation was, say, around 4C, the addition of this one day would yield, in exceptional circumstances, a 0.04 skew on the CET for the winter. Lost in the noise I'd say.

The question about the cold 30 day blocks is interesting, but the same logic applies. In simple terms it is more likely that we could still get a cold 30 day period than one that falls within the neat conventions of the calendar, simply because there are around 30 times as many permutations available. However, the change in likelihood of a <1C month is just the same. A warming trend is a warming trend.

You are right, I worded that badly! I meant merely that whilst they occured historically they were never 'likely' in the sense that their occurance was less than usual, a one in four is not a good bet, but its a punt, thats what I meant!

I think the chances here in a GW world are really down to what happens to the jet longer term and on a winter to winter basis and (perhaps) to a slowing of the THC. Whilst this is a highly debatable subject and not something I would want to bet any money on or indeed back (yet) in any significant debate, IF the measured decline in its potency has any factual basis then we may find ourselves more susceptible to influence from traditionally colder areas - Siberia, Scandinavia, the Arctic on a more regular basis. This would drastically increase the chances of colder months. However what would then happen to a continuing warming world if nothing is done is a mystery to me.

I certainly agree there. There is one glimmer for those who hang desperately to any hope for cold winters, and that is a sudden correction in the macro or meso scale system.

My own view is that whilst possible, this is unlikely; and further that even if it were to occur, it might still be a short term downwards blip. With dramatic warming at the pole we might get to a point where the rate of change is simply too great for any localised cooling to buck the trend for long, irrespective of what happens to the THC. The NAD keeps the UK anomalously warm in winter, but only matters so long as air to our north is much much colder.

That, of course, is just my opinion. The fact is that we're into the unknown and, for now, unknowable.

[Weather-history]

PTFD

Here is the surprising or maybe not result

Warmest Jan: 1921 at 7.3C

Feb: 1945 previous at 7.1 exceeded twice in 1990/98 at 7.3C

Mar: 1938/57 AT 9.2C

Apr: 1943 at 10.5C

May: 1919 previous at 13.5C exceeded in 1992 at 13.6C

June: 1976 at 17C

July: 1983 previous at 19.5C exceeded in 06 at ....

Aug: 1975 previous at 18.7C exceeded in 1995 at 19.2C and 1997 at 18.9C

Sept: 1929 previous at 15.3C exceeded 1999 at 15.6C

Oct: 1960 previous at 13C exceeded 2005 at 13.1C

Nov: 1939 previous at 8.7C exceeded 1994 at 10.1C

Dec: 1934/74 at 8.1C

So since 1987 as the bench mark record warm months were achieved on 9 occasions. Coldest month never been set since 1987.

BFTP

Warmest January was in 1916 with 7.5

Warmest February is 1779 with 7.9

Warmest May is 1833 with 15.1

Warmest June is 1846 with 18.2

Warmest September is 1729 with 16.6

Warmest October is 2001 with 13.3

Warmest November was actually 1818 with 9.5 until 1994

So infact only 4 occasions when records have been broken since 1987

The warmest July, August, October and Novembers on record

[Stratos Ferric]

And the 96 such days from 1st of January 1980 and August 1986

That's a decadal half life! Holy cow, carry on like that and by 2026 we'll be "enjoying" around one freezing day per six winters!

Warmest January was in 1916 with 7.5

Warmest February is 1779 with 7.9

Warmest May is 1833 with 15.1

Warmest June is 1846 with 18.2

Warmest September is 1729 with 16.6

Warmest October is 2001 with 13.3

Warmest November was actually 1818 with 9.5 until 1994

So infact only 4 occasions when records have been broken since 1987

The warmest July, August, October and Novembers on record

We'd see a much spikier data set if we could go down to individual days, and as Wilson suggest, days by individual site. There will have been cold records set at this level of granularity, but equally a lot more warm records set too. It's also necessary to look at not only "high max" and "low min", but "high min" and "low max".

[snowsure]

We'd see a much spikier data set if we could go down to individual days, and as Wilson suggest, days by individual site. There will have been cold records set at this level of granularity, but equally a lot more warm records set too. It's also necessary to look at not only "high max" and "low min", but "high min" and "low max".

Perhaps the GCOS surface network http://www.wmo.ch/web/gcos/gcoshome.html and select GSN daily data may be what you are looking for?

[Stratos Ferric]

Perhaps the GCOS surface network http://www.wmo.ch/web/gcos/gcoshome.html and select GSN daily data may be what you are looking for?

Good data, and that would certainly provide a more detailed reference. Interestingly I've selected five sites from around the globe at random, and they all demonstrate warming of around 1-2C since the early 60s.

[VillagePlank]

First up very low night time minima alone would not be sufficient, though they would certainly be necessary, for a very cold month.

That's what I meant; without successive nighttime minima it is highly unlikely that we will record a subzero CET

Second, you need to clarify what you mean by CET being an "anomalous" record. Taking your point about a "grid" oin the ensuing paragraph I'm assuming you're suggesting that it's somehow representative. It may or may not be, it depends what you're trying to use it to prove.

Indeed. I outlined a single strategy when of course, there are a lot more. The summation of a number of stations scattered throughout the UK whether aligned to a grid or otherwise, might indeed show the warming trend in a more pronounced way; it might, of course, show the opposite (which I don't think would be the case)

The question as posed at the head of the thread was "are we still capable of getting a <1C CET month". It strikes me that in this regard, use of ET as the yardstick is absolutely correct.

Of course. Slap my wrists for venturing of topic

Your point about whether or not a tighter grid of data would prove a different warming hypothesis is an interesting one, but one which almost certainly would not hold up in practice.

THis fails any sort of test of logic. The CET, which I presume you support, can still, mathematically be described as a grid. A grid does not need to be uniform, nor square, nor linear.

I suspect that one reason why CET holds up as an accepted record is that it matches well trends in other longitudinal series.

Indeed. What does this suggest?

It would, for example, in a landmass as small as the UK, be unthinkable to suppose that whilst the CET is warming across a period, SE Scotland, say, was cooling.

Not true. The CET is already corrected for effects such as UHI and for other factors. If it holds true to reduce effective warming of the CET trend, does it not also hold that the inverse (should it happen) also be true. It may be the case that places that are known frost-hollows have not shown any warming. I don't know. I do not have the data to hand

Day to day variations are quite possible as air mass boundaries move slowly enough for the 24hr period to be too short to allow for levelling, but over the longer terms these things tend to smooth out. The simple fact is that air mass parcels tend to be far larger than the UK (the same is not true for, say, mainland Europe or the US). Station by station analysis would, almost certainly, show precisely the same general pattern of warming.

Watch the weather forecast and compare degC for S/E against N Scotland. This, of course, is a daily pattern. I think you are correct to presume (I take it it's a presumption as no evidence is provided) that the rate of warming for all areas of the UK are similar; but I'd bet my bottom dollar that there are exceptions. Of course, you mitigate exceptions by using trends, means and other such statistical tools to protect your argument.

Not quite sure what you're arguing re sidereal time, and the necessary corrections to astrological time required to allow a neat 24hr clock to work. Given that, viewed end to end, the sum total of experienced weather is exactly the same however you carve time up, the overall average for the data set would still be same.

Missing the point here. Consider a warm 2 weeks to Nov, and cool last two weeks, a very cold beginning to December, and am mild new year. The way you want it is to record an average November, and an average December. In fact the record should show that the four weeks between the middle of Nov to the middle of Dec should have shown a very low CET. There is no statistical way of unpacking that result unless you move the timeframe. The record will show a winter of averageness; which, I presume, is the way you prefer it. In reality, of course, the UK would've experienced a remarkably cool period of time for nearly a month.

What's even worse is your propensity to accept the CET as a 'good' record without, it appears to me, to be a thorough understanding of it's semantic shortfalls - of which there are a great deal. I think that the CET shows an average picture, and that, by definition, shows no detail, or no clarity of the changes in the climate unless they are long lived and significant (like the current recent warming trend) To argue, in detail about synoptics affecting UK climate and it's subsequent effect on the CET value for a calendar month to me seems a little wrong. It is entirely feasible that the CET could record a highly anomalous sub-zero temperature for one month.

Individual packets of time, rebounded, may be slightly warmer or cooler arithmetically, but cooling in one packet would be offset by an equal and opposite adjustment in the next. Adding a day to a winter in the leap year provides at most an adjustment of around 1% of the difference in temperature on the extra day with that of the mean for the period as a whole. Assuming you take March 31st as the extra day (even though it's Feb 29th on the calendar), as this would have the largest expected variation from the period as whole, and assuming that that variation was, say, around 4C, the addition of this one day would yield, in exceptional circumstances, a 0.04 skew on the CET for the winter. Lost in the noise I'd say.

Indeed I made this point, myself.

The question about the cold 30 day blocks is interesting, but the same logic applies. In simple terms it is more likely that we could still get a cold 30 day period than one that falls within the neat conventions of the calendar, simply because there are around 30 times as many permutations available. However, the change in likelihood of a <1C month is just the same. A warming trend is a warming trend.

You presuppose that I am looking to change to find a colder pattern. I apolgise if I've inferred that (I haven't) I have performed analysis on the CET (in what can only be considered unique ways) that strongly reinforce a warming pattern. Edited August 22, 2006 by Wilson

[BLAST FROM THE PAST]

Warmest January was in 1916 with 7.5

Warmest February is 1779 with 7.9

Warmest May is 1833 with 15.1

Warmest June is 1846 with 18.2

Warmest September is 1729 with 16.6

Warmest October is 2001 with 13.3

Warmest November was actually 1818 with 9.5 until 1994

So infact only 4 occasions when records have been broken since 1987

The warmest July, August, October and Novembers on record

Mr D

Thanks for that I was concentrating only on the 20th cenury for some reason and should have clarified that, so actually the figures are even more surprising then!? By the way where would I get access to the full recorded data?

BFTP

Edited August 22, 2006 by BLAST FROM THE PAST

[Anti-Mild]

Mr D

Thanks for that I was concentrating only on the 20th cenury for some reason and should have clarified that, so actually the figures are even more surprising then!? By the way where would I get access to the full recorded data?

BFTP

Here you go! Watch out though, looking at the data file makes your eyes go funny !

http://www.metoffice.gov.uk/research/hadle...bsdata/cet.html

[Stratos Ferric]

...

Not true. The CET is already corrected for effects such as UHI and for other factors. If it holds true to reduce effective warming of the CET trend, does it not also hold that the inverse (should it happen) also be true. It may be the case that places that are known frost-hollows have not shown any warming. I don't know. I do not have the data to hand

Watch the weather forecast and compare degC for S/E against N Scotland. This, of course, is a daily pattern. I think you are correct to presume (I take it it's a presumption as no evidence is provided) that the rate of warming for all areas of the UK are similar; but I'd bet my bottom dollar that there are exceptions. Of course, you mitigate exceptions by using trends, means and other such statistical tools to protect your argument.

Missing the point here. Consider a warm 2 weeks to Nov, and cool last two weeks, a very cold beginning to December, and am mild new year. The way you want it is to record an average November, and an average December. In fact the record should show that the four weeks between the middle of Nov to the middle of Dec should have shown a very low CET. There is no statistical way of unpacking that result unless you move the timeframe. The record will show a winter of averageness; which, I presume, is the way you prefer it. In reality, of course, the UK would've experienced a remarkably cool period of time for nearly a month.

What's even worse is your propensity to accept the CET as a 'good' record without, it appears to me, to be a thorough understanding of it's semantic shortfalls - of which there are a great deal. I think that the CET shows an average picture, and that, by definition, shows no detail, or no clarity of the changes in the climate unless they are long lived and significant (like the current recent warming trend) To argue, in detail about synoptics affecting UK climate and it's subsequent effect on the CET value for a calendar month to me seems a little wrong. It is entirely feasible that the CET could record a highly anomalous sub-zero temperature for one month.

I suspect we're agreeing violently on most points. However, since you're flashing your blade...

The CET is corrected ONLY for longitudinal effects not for spatial ones (apart from where there have been changes in contributing stations), this in order to adjust for external background changes across the data series, otherwise, for example, a warming trend might appear when in actual effect it's simply a consequence of the effect of the built environment on the microclimate. Not quite sure what point you're making otherwise in the first para above when you talk about the "inverse": what inverse? Are you suggesting that there are missing corrections?

You're correct. To be sure of my general contention (that the CET is broadly representative of the UK as a whole) we'd probably need to check data for every single UK station. My general hypothesis is, however, that since the UK is a small landmass, and the CET represents an average over a fair proportion of that area, then the lack of climatic variability across the whole (holding asaide, say, the mountain tops in Scotland) suggests that a warming in CET would be reflected elsewhere as well. There is certainly NOTHING to suggest that given the general change in UK synoptics specific locations would have been cooled where the general trend has been to warm. In any case, I'd be disinclined to include frost hollows as in any way represenatative, as by definition they are microclimatic outliers.

Your argument about my use of "trends, means and other such statistical tools to protect [my] argument" is, if you do mind me saying so, rather specious. How else can a trend b e identified other than by using trend analysis. Any trend analysis, by definition, requires calculation of means somewhere in the process. It's not protection therefore, so much as a mathematical requirement.

I don't think I did miss the point re the carving up of them month either, as I said originally...

the sum total of experienced weather is exactly the same however you carve time up, the overall average for the data set would still be same. Individual packets of time, rebounded, may be slightly warmer or cooler arithmetically, but cooling in one packet would be offset by an equal and opposite adjustment in the next. Adding a day to a winter in the leap year provides at most an adjustment of around 1% of the difference in temperature on the extra day with that of the mean for the period as a whole. Assuming you take March 31st as the extra day (even though it's Feb 29th on the calendar), as this would have the largest expected variation from the period as whole, and assuming that that variation was, say, around 4C, the addition of this one day would yield, in exceptional circumstances, a 0.04 skew on the CET for the winter. Lost in the noise I'd say.

The question about the cold 30 day blocks is interesting, but the same logic applies. In simple terms it is more likely that we could still get a cold 30 day period than one that falls within the neat conventions of the calendar, simply because there are around 30 times as many permutations available. However, the change in likelihood of a <1C month is just the same. A warming trend is a warming trend.

It is perfectly easy to "unpack" the data, simply by going back to the daily values. If you carve the year up into each possible packet of 30 consecutive days, then allowing for overlaps into the year before and afterwards, all you end up with is 365 sets of thirty days, where the calendar has 12. What you end up with is around 30 times the likelihood of getting a period averaging <1C, simply because the base sample is thirty times larger. However, if across the whole CET sample the frequency of calendar months with <1C has halved, then statistically speaking, the proportion of all thirty day envelopes with an average <1C will also have halved. Yes, we can argue the toss about fractions here, but for all intents and purposes they are part of the same data set so the trend for any sub sample CANNOT be different to the trend for the data set as a whole.

Not sure what you mean by "sematic shortfalls" - perhaps you'll elaborate. It may have many shortfalls but I'm not aware that any are "semantic". And I'm not clear why you assert that I have a "propensity to accept the CET as a 'good' record": it may have passed me by but nowhere have I stated that; I refer to the CET simply because that was the original question - you can post as many sardonic folded arms smileys as you like, it's not about sticking to thread so much as responding to the question posed. I'm also not aware that I said anything about synoptic effects and CET over a given month. The CET does show an average picture, but averaged spatially NOT temporally. By presenting the data in a time series then longer term variations DO become apparent and the data is necessarily averaged. What is driving those changes is clearly the subject of heated debate, but changes they are. Therefore, and I think I said precisely the same originally, there is always a chance (and in that sense it is feasible) that we will have a month with a CET <1C, my point is, given the continuing trend towards increasing temperature, particularly in winter, and particularly re minima, that this is now highly unlikely. Therefore, feasible, yes; likely, no.

[BLAST FROM THE PAST]

Here you go! Watch out though, looking at the data file makes your eyes go funny !

http://www.metoffice.gov.uk/research/hadle...bsdata/cet.html

Cheers AM

BFTP

[VillagePlank]

The CET is corrected ONLY for longitudinal effects not for spatial ones (apart from where there have been changes in contributing stations), this in order to adjust for external background changes across the data series, otherwise, for example, a warming trend might appear when in actual effect it's simply a consequence of the effect of the built environment on the microclimate. Not quite sure what point you're making otherwise in the first para above when you talk about the "inverse": what inverse? Are you suggesting that there are missing corrections?

With regard to an inverse being true (apart from being a logical fallacy on my part - as I'm sure you are aware inverses are often not true, have you ever tried to unboil an egg and grow a chicken?) Are corrections being made to ensure that any cooling factors are mitigated? For instance flattening a big mound reduces surface area of insolation which can happen in road building exercises. Reduced surface area means less stored heating (I think) in the same way orographic forcing is, in part, down to a greater surface area in the same area of land. I've looked for papers on land surface area and local weather, and I can't find any. This is why I was vague, and I cannot follow the argument through. I would still argue that land surface area is a significant contributer to weather. I was under the impression that the model view of the weather does not account for density of surface area. I could be (and probably am) wrong.

I am absolutely sure there are missing corrections. Whether they are significant both in a chi-squared analysis, or indeed, in an absolute sense is, I think, the point. Are you really going to correct temperature for a station when the known effect is 0.000000000000001C? I think it's unlikely. There is always the case to presume, I think, that there could indeed be corrections whether warmer, or cooler, that might be significant. I guess the question is: what is the degree of significance?

To be sure of my general contention (that the CET is broadly representative of the UK as a whole) we'd probably need to check data for every single UK station. My general hypothesis is, however, that since the UK is a small landmass, and the CET represents an average over a fair proportion of that area, then the lack of climatic variability across the whole (holding asaide, say, the mountain tops in Scotland) suggests that a warming in CET would be reflected elsewhere as well.

I think, on my understanding that the CET is the mean of only a very few stations, that the mean is too coarse. This is, of course, a matter of opinion. Given the size of the UK I think that your contention is a reasonable one, and probably, for the disinterested, or casual reader, it is certainly enough. Perhaps that's the reason the CET series exists. I do not think, however, that your contention is enough for those with a more vivid interest in UK climate and it's subsequent analysis. I'll concede that this, then, becomes a matter of use, and nothing else.

There is certainly NOTHING to suggest that given the general change in UK synoptics specific locations would have been cooled where the general trend has been to warm. In any case, I'd be disinclined to include frost hollows as in any way represenatative, as by definition they are microclimatic outliers.

There is nothing to suggest this because nobody has looked? I think that the atmosphere is sufficiently complex enough to have an overall warming trend, but to have more local effects such as cooling. This is my opinion (and belief) and I have no evidence to back this; although I'm looking. I will argue, though, that there are plenty of proposals that show a mean global warming, and local cooling. The Antarctic is cooling in a world that is warming for instance. Sure, you might want to argue about scale . . . .

Your argument about my use of "trends, means and other such statistical tools to protect [my] argument" is, if you do mind me saying so, rather specious. How else can a trend b e identified other than by using trend analysis. Any trend analysis, by definition, requires calculation of means somewhere in the process. It's not protection therefore, so much as a mathematical requirement.

Yes, apolgies. My point, though, however ineloquently put, is that the CET is only of use for long term and significant analysis. Synoptic arguments (which are both temporal and spacial) to conclude a future CET seems at the very least spurious, to me, on this basis. The CET, by definition, is a mean. My point was that synoptic arguments are irrelevant; the CET does not, in and of itself, contain any synoptic detail; it does contain long term synoptic variability, though. It may (and is for me) be of interest to record notable event's CET which do not occur, conveniently, on a calendar monthly basis.

If you carve the year up into each possible packet of 30 consecutive days, then allowing for overlaps into the year before and afterwards, all you end up with is 365 sets of thirty days, where the calendar has 12. What you end up with is around 30 times the likelihood of getting a period averaging <1C, simply because the base sample is thirty times larger.

No argument there.

However, if across the whole CET sample the frequency of calendar months with <1C has halved, then statistically speaking, the proportion of all thirty day envelopes with an average <1C will also have halved.

The chances are, as you say, that the whole sample will be self-similar at all scales. This view, however, does result from the flattening of nosie; increasing the sample frequency introduces more 'noise' and I contend that on a number of occasions that the noise contributes in some significant fashion. I'd still like to do the exercise, though. Has anyone got the daily CET for the last 100 years that I can borrow? I'll publish the results here

Yes, we can argue the toss about fractions here, but for all intents and purposes they are part of the same data set so the trend for any sub sample CANNOT be different to the trend for the data set as a whole.

I think we'll have to agree to disagree here. It is possible to make many different 'views' from the same dataset. Edited August 22, 2006 by Wilson