Buy

Books
Click images for more details

Support

 

Twitter
Recent comments
Recent posts
Currently discussing
Links

A few sites I've stumbled across recently....

Powered by Squarespace

Discussion > (A)GW Lite

I want you to leave, so the grown-ups can resume talking.

That's the tragedy BBD. You speak a lot of sense, you may even be mostly or entirely correct. But your lack of social skills makes everyone hate you and refuse to have anything to do with you.

Wet wipes, anyone? I think you need a mirror.

May 25, 2012 at 10:51 AM | Unregistered CommenterBBD

I'm certainly not aroused by telling you this rather grim truth about your persona.

If you believe that proving that you're right though browbeating is the best and only way to convince people of a point of view then your personality is even more incomplete than your persona here suggests.

Sine wave?

May 25, 2012 at 11:02 AM | Unregistered CommenterTheBigYinJames

" My only fault was not demonstrating what is generally known and understood rigorously the first time around. And that would only matter if I was wrong."

Translate this BBDspeak: "I used a wrong method. But that would matter only if my conclusions were wrong."

How come you have not given a simple, straightforward direct answer to my question yet?

I'll gladly admit - the land trends on the BEST series shows an increased trend between the time periods you chose, *by using the right method*.

That means, your earlier method was wrong. Am I correct?

A simple 'yes' or 'no' would suffice.

May 25, 2012 at 11:58 AM | Registered Commentershub

shub

The answer is of course, no.

You claim that the first method is 'wrong' but it isn't. Of course the long trend vs short trend isn't as robust as comparing the two shorter trends. But it isn't 'wrong'. So you won't get a 'yes'. I'm rather surprised that you haven't managed to work this out yet.

But of course, you will persist in trying to force the conversation away from the real point. Which is that using either BEST or CRUTEM3 (or CRUTEM4) one can demonstrate that the rate of warming doubled between the two periods.

You know and I know that this is the point. Which is why you are banging on still about method. It's as tedious as it is transparent.

Enjoy the weekend.

May 25, 2012 at 12:35 PM | Unregistered CommenterBBD

TBYJ

Sine wave?

No and no.

Sorry. FWIW, I used to think that Akasofu had a case. I also used to be a lukewarmer.

Enjoy the weekend too.

May 25, 2012 at 1:55 PM | Unregistered CommenterBBD

Has he gone yet? High-fives all round.

May 25, 2012 at 1:59 PM | Unregistered CommenterTheBigYinJames

Just for that, you can have this pretty picture to contemplate over your weekend:

BEST land and CRUTEM4v land (common 1981 - 2010 baseline).

It's interesting to see a suggestive sinusoid in the early portion of the BEST curve. The contrast with the C20th is very marked. Note the excellent agreement between BEST and CRUTEM4v. Looks like the bastards got to Muller too. Or maybe that's just tinfoil-top conspiracy theorising.

;-)

May 25, 2012 at 2:15 PM | Unregistered CommenterBBD

I'm sorry I hurt your feelings BBD. It must be hard, having jumped ship once before, to see that you would have been better staying where you were.

May 25, 2012 at 2:29 PM | Unregistered CommenterTheBigYinJames

"You claim that the first method is 'wrong' but it isn't."

I get it. It is not wrong, but merely less 'robust' and 'rigorous'.

I understand climatespeak.

And, you, go around proselytising Hansen and everyone is supposed to bend over.

May 25, 2012 at 2:37 PM | Registered Commentershub

I've re-read the thread and TBYJ has a point. I wasn't patient or persuasive.

Puts me in mind of von Clausewitz:

Everything in war is very simple. But the simplest thing is difficult.

May 25, 2012 at 10:52 PM | Unregistered CommenterBBD

shub

WRT 'climatespeak', being 'wrong' and bending over for Hansen.

From BEST:

Decadal trends:

1900 - 2010 = 0.09C

1950 - 2010 = 0.18C

1975 - 2010 = 0.28C


The rate of warming 1950 - 2010 is 2x the 110 year decadal trend. The rate of warming 1975 - 2010 is 3.1x the 110 year decadal trend and 1.5x the 60 year decadal trend 1950 - 2010.

The rate of warming has increased over the century.

You need to explain exactly how this is wrong.

As we agree, adding the comparison between the decadal trend 1910 - 1940 (0.14C) and 1975 - 2010 (0.28C) confirms the above. But what interests me is your proof that the first part is 'wrong'.

May 26, 2012 at 5:24 PM | Unregistered CommenterBBD

"You need to explain exactly how this is wrong."

You don't take two overlapping periods of time and compare them they way you did. The periods are non-independent and therefore no meaningful conclusions can be drawn.

Time moves forward even though we draw trend-lines starting more and more back in time. So, if at all, a allowable method would be to compare the trend in the early part of the 20th century warming trend, to the overall twentieth century warming trend. This would give us a rate of 1910-1940 - 0.14C/decade and a reduced rate of 0.09/decade trend for the whole century. This squares with the fact that we had a period of cooling in the middle as well.

Paul M demonstrated this to you by using a diagram, right at the beginning. TBY is referring to the same thing above.

If you want to make a statement: "the rate of warming is increasing", you take two different non-overlapping from the recent past (say, the previous three decades) and an earlier period, compute the warming rates, and compare them. That is a valid way of doing it.

Following this method, BEST does show an increased rate in the latter decades of the 20th century compared to the earlier warming period.

But what does it mean? I am not sure, on two counts.

Firstly, I am not sure I can say things about land data in terms of the CO2-driven global warming (it is supposed to show up on land more yada yada). What does the land data mean? I don't know.

Secondly, take a look at this:
ALL land series data - 1977 to 2010

The different series are so wildly different - in magnitude and trend both - that the whole thing looks like a Japanese hand fan. I am sure if you plot the different model projections for the time period, it would truly become a fan. This tells me that land temperature data and dependent calculations (like the ones we do) are sensitive to the series used and its underlying assumptions. If I look at CRU 3, the trend in the last three decades or so, is similar to the trend in the early 20th century. It also speaks to the parlous state of affairs in land temperature reconstruction.

Leave alone the satellites - scientists are still haggling how to reconcile the differences seen and their meaning. The same is evident in the graph.

Thirdly, take land+ocean (i.e., global temperatures). The trend differences you gleefully point, disappear miraculously. Their corresponding ocean counterparts are exactly mirror images such that, overall, (land+ocean) trends are almost similar between the early 20th century warming

I plotted HADCRUT4 here for our benefit.

The last part of the above notwithstanding, the way to compare different rates of increase in a time series that seems to go up and down is certainly not the way you did it.

May 26, 2012 at 7:11 PM | Registered Commentershub

Looking at the HADCRUT4 plot, posted by Shub, I am struck but the periodicity of it, almost as if there is a sine wave superimposed. Has there been any analysis of this characteristic? What period does it have on average and what causes it (el niño, sunspot cycles)?

Given this level of short term variation superimposed upon a longer term trend, how can claims that there has been no warming in the last 15 (say) years be serious?

May 26, 2012 at 9:26 PM | Unregistered CommenterBitBucket

Shub

Thank you for your response. Can we continue the discussion without emotive language etc? I would like to do that. I'm genuinely interested in your methodological objection and to prove it, here is rather a long response...

You don't take two overlapping periods of time and compare them they way you did. The periods are non-independent and therefore no meaningful conclusions can be drawn.

I'm still not clear why no meaningful conclusions are possible. We are investigating the evolution of a phenomenon through time. Surely one approach is to compare changes in the latter half of the time-series to the whole thing? So long as it is done appropriately, eg halve the series then halve the second period and compare each successive trend. If each trend is higher than the previous one, the rate of change is shown to increase over time. You say yourself one needs to look at the data and consider appropriate tools for examining it. I suggest that what I did was appropriate, but if you were to move from 'wrong' to 'use with caution' I would agree.

Have you got any textbook reference that specifically says that is is incorrect to examine a climate time series in the way I did (and many others have done)? That would help.

I have re-done the graphs you link, but with a common 1981 - 2010 baseline for better visual comparison (I have excluded CRUTEM3 as it is now obsolete. I'm not sure why you picked 1977 as a start point; I've used 1979, which is when the satellite data begin).

All land series 1979 - present, common 1981 - 2010 baseline

BEST and CRUTEM4 are in very close agreement, as are RSS and UAH. There are a few things to bear in mind here. First, BEST and CRUTEM4 use *different* methodologies but arrive at essentially the same result. Second, RSS and UAH use *similar* methodologies and get similar results.

This is important because there is a scientific debate over whether the UAH/RSS satellite reconstruction of TLT has cool bias. Most recently, see Po-Chedley & Fu (2011). There is a useful discussion of earlier work pointing to a cool bias in the RSS/UAH-type of analysis at SkS.

This is not to say that the RSS/UAH reconstructions are biased cool, but the possibility needs to be kept in mind when considering the divergence with the land surface records, especially as there we have two different methodologies (BEST; CRUTEM4) in very close agreement.

***

Thirdly, take land+ocean (i.e., global temperatures). The trend differences you gleefully point, disappear miraculously.

As you would expect. The theory states that land will warm more rapidly than the ocean because the ocean has a huge heat capacity compared to the land surface. And that's what we see. SSTs are also rising, of course. As I've mentioned before, aside from being more immediately responsive to a warming troposphere, the land surface temperature is far better sampled than SST, and is therefore the most useful of the two records.

***

So, to summarise: I remain unclear as to why the original method I used is incorrect. I agree that it should be used with caution. I agree that comparing the 1910 - 1940 trend with the 1975 - present trend is sound, but note that it corroborates the original approach.

May 26, 2012 at 10:14 PM | Unregistered CommenterBBD

Well, much though it grieves me to debate with someone prone to boorish jeering (sorry - you call it "clarification"...)

Your errors in my view are two-fold, and several contributors have pointed them out in different ways:

1) Looking at smoothed curves for global temperature there is a vaguely periodic (sinusoidal is putting it a bit strongly) pattern. It's easily noticed in CRU Information Sheet 1 (http://www.cru.uea.ac.uk/cru/info/warming/) for example, with maxima at 1880, 1940 and 2000, implying a 60 year cycle, and minima around 1910 and 1970. It is a reasonable assumption that this pattern corresponds to a natural cycle, and indeed the Pacific Decadal Oscillation fits the bill. The magnitude of the temperature variation is of the order of +/- 0.2C. The time period 1970 to 2000 or any closely overlapping period will give a temperature rise which includes that range. That is potentially, then, an error. (More specifically, the year 1977 saw the Great Climate Shift, and the year 1998 saw a huge El Niño. If you avoid those events, e.g. looking at lower tropospheric temp for 1980 to 1995, the temp shows no change or even a slight decline. The warming goes largely in big steps, which makes it hard to avoid big variations in average rate depending on the dates chosen.)

2) The earlier warming (say 1700 to the early 20th century) is recovery from the LIA. Sceptical sources are unanimous in rejecting the Mann/Bradley/Hughes recontruction. The MWP was probably warmer in parts, certainly not significantly cooler, than the late 20th century. It is reasonable to assume that at least some of the recent warming is a continuation of the recovery. The pre-1940 curve suggests - very approximately once more - about 0.3C per century, or 0.2C per ~60-year cycle. That would form part of the observed warming, and not acknowledging it, and thus assuming human agency, would be an error. (Longer term, such periods of warming and cooling seem to alternate once or twice in a millennium, so likely enough we may expect a downturn in the medium term.)

Looking at the curve again, it is obvious that the 2000 maximum is higher than a straight line through the 1880 and 1940 highs would suggest - again by very roughly 0.2C. Since some warming was to be expected from the observed rise in CO2, that's not surprising. Assuming the CO2 rise begain in earnest around 1940, that would imply again very roughly a 0.2C AGW in 60 years, which is a very low sensitivity.

So straight lines don't do the job.

This approach is based on crude extrapolation, without any necessary commitment to particular mechanisms, but it fits in with the general feeling among sceptics that the Earth's CO2 and temperature history shows that the planet is buffered against catastrophic climate change.

The above reasoning is very vague, and I am sure you are equipped to dispute various issues; but it provides a starting point for more detailed discussion. Notice that there are three suggested elements to the warming, and of the roughly 0.8C increase from 1970 to 2000, perhaps only a quarter is AGW. (And, as you can see in this thread, some sceptics doubt even that; but not I. Also I have assumed that - apart from MBH 98 and subsequent papers attempting to bolster it - the published temperature series are not far from the truth; again, many disagree.) "Lukewarmist" implies something more than that, I think. It's AGW-minimalist.

May 27, 2012 at 2:31 AM | Unregistered Commenterlogicophilosophicus

Has he gone yet? High-fives all round.

May 25, 2012 at 1:59 PM | TheBigYinJames>>>>

Hello James!

Looks like the thread has mutated a little bit.

I'll come back another day when I've finished enjoying the global warming in my garden.

May 27, 2012 at 7:25 AM | Unregistered CommenterRKS

Logico

I assume you were responding to BitBucket, but since it's gone quiet, here's my tuppence-worth:

- PDO. Oscillations oscillate :-) +0.2C... -0.2C... +0.2C... -0.2C. The *net* effect is zero. There will be no effect on long term trend for GAT.

The PDO and GAT diverge significantly since ~1980. I don't think the 0.2C-from-PDO argument stacks up.

- Lower tropospheric temperature 1980 - 1995. You cannot infer anything from short periods. It's much more informative to look at a minimum of 30 years:

UAH, RSS, 1979 - present; annual means, common 1981 - 2010 baseline

- Another way of looking at the 1979 - present period would be to remove the effects of ENSO, solar variation (Schwabe cycle) and volcanism. Foster & Rahmstorf (2011) does just this, and the result is neatly encapsulated by this graphic.

- 'Recovery' from the LIA:

It is reasonable to assume that at least some of the recent warming is a continuation of the recovery.

Actually, it isn't. As someone wiser than me once put it, the climate is not a bouncing ball. Nor is there an ideal state to which it is 'recovering' after the LIA. Some combination of negative forcing (possibly TSI and volcanism) depressed GAT and we got the LIA. Then forcing increased and GAT began its discontinuous rise from the end of the C19th to the present. It does this *in response* to external forcing (here including a substantial transfer of carbon from geological sinks to the atmosphere by human activity).

There's no physical basis for arguing that ~0.3C of C20th warming is 'recovery' from the LIA.

- Low climate sensitivity. One dear to my heart, this ;-)

If climate sensitivity is low, it will not respond much to changes in forcing. Variability would be flattened out. The MWP and LIA would either have been much less pronounced or non-existent. The termination of glacials by orbital forcing (Milankovitch) simply could not happen. In short, known climate and paleoclimate variability is solid empirical evidence for a moderately sensitive climate (I suggest the current most likely value of ~3C for 2xCO2 at equilibrium with a transient climate response of ~2C).

It's ironic that many sceptics argue for a low CS and *in the same breath* that natural variability is responsible for much of recent warming. You see the problem, I'm sure. I'm afraid it is impossible to reconcile known climate behaviour with the lukewarmer position, which requires a relatively insensitive climate system (note - I used to be a lukewarmer too).

So, in summary: it seems probable that most of the warming since the mid-C20th is mainly caused by the the increasing atmospheric fraction of CO2. Yes, natural forcing regularly over-prints the anthropogenic signal; no, it does not affect longer-term trends. Nor does it make any difference to the underlying phenomenon: a gradual emergence of the anthropogenic signal as dominant. No, you can't have an insensitive climate system that exhibits the kind of variability we see - from the C20th to deglaciation under Milankovitch forcing. No, you can't be a lukewarmer unless you can get around this problem :-)

May 27, 2012 at 6:03 PM | Unregistered CommenterBBD

This is all very interesting, but sadly beyond me. What is not beyond me is this: with a system oscillating with ~10 year periodicity, even if there is no longer term trend, if you pick a relatively short part of the curve, say 15 years, and draw a line from the starting point to the end, then depending upon which starting year you chose you can make the line point either up or down. So drawing any conclusion from a short period is completely invalid. All those contributing to this thread seem like clever people so this must be blindingly obvious.

So both sides please, leave it out! If you want to discuss the trends, do so only over a period that shows your argument at its worst. If your case still holds water, then you can perhaps claim some justification, otherwise you and your argument cannot be taken seriously. Discussions would have more weight if both sides were to accept and enforce this rule.

May 27, 2012 at 7:26 PM | Unregistered CommenterBitBucket

BBD,
How can you compare two periods if one of them is included in the other?

You are falling for a logical fallacy. I am not aware/knowledgeable of logical notation that can be used to demonstrate this formally, but it can surely be done.

I am looking at the same curve as you are. What I see is: "As you include longer intervals going back in time, the rate of temperature rise *decreases*".

II
Why did you leave out CRUTEM3?!! It was good for all of us until just a while back!

That the rate of temperature increase over the land differs from series to series by such significant magnitude tells me that the slope is acutely dependent on the methodology of each instrumental series (BEST vs HADCRU vs satellite). Knowing this, how can you base theoretical conclusions on the rate of temperature increase?

May 27, 2012 at 8:20 PM | Registered Commentershub

BitBucket

I'm not sure we can say that there is a 10 year climate oscillation. I definitely agree that fitting trends to short periods is to be avoided as it can mislead.

So both sides please, leave it out!

I don't think I used any period shorter than 30 years, which is the standard 'climatology'.

If you want to discuss the trends, do so only over a period that shows your argument at its worst. If your case still holds water, then you can perhaps claim some justification, otherwise you and your argument cannot be taken seriously.

Of course. What is your view of the debate above? To recap: I suggest that one approach is to compare changes in the latter half of the time-series to the whole. So long as it is done cautiously, by halving the series then halving the second period and comparing each successive trend. If each trend is higher than the previous one, the rate of change is shown to increase over time.

Shub says:

So, if at all, a allowable method would be to compare the trend in the early part of the 20th century warming trend, to the overall twentieth century warming trend. This would give us a rate of 1910-1940 - 0.14C/decade and a reduced rate of 0.09/decade trend for the whole century.

Which approach do you think would yield the most information about the nature of change over the series?

May 27, 2012 at 9:18 PM | Unregistered CommenterBBD

Shub

How can you compare two periods if one of them is included in the other?

Why can't we do this? You appear to argue that we cannot investigate a phenomenon by comparing its recent behaviour with its past behaviour. Is that what you are saying?

May 27, 2012 at 10:48 PM | Unregistered CommenterBBD

Shub

I am looking at the same curve as you are. What I see is: "As you include longer intervals going back in time, the rate of temperature rise *decreases*".

If our curve was relatively flat, we'd see little change as we compared the latter half of the time-series to the whole by halving the series then halving the second period and comparing each successive trend.

Sorry about CRUTEM3. I felt bad for it. A stalwart for so long, then just cast aside like a fag butt. But such is life.

May 27, 2012 at 11:15 PM | Unregistered CommenterBBD

You are falling for a logical fallacy. I am not aware/knowledgeable of logical notation that can be used to demonstrate this formally, but it can surely be done.

If our positions were reversed, I think you might have something to say about this :-)

I'll behave, and leave it at that.

May 27, 2012 at 11:40 PM | Unregistered CommenterBBD

BBD.

"PDO. Oscillations oscillate :-) +0.2C... -0.2C... +0.2C... -0.2C. The *net* effect is zero. There will be no effect on long term trend for GAT." That's right; but with a 60-year oscillation, you can see that choosing a 30-year normal (or a 90-year period) which happened to start near the minimum would give a spurious ~0.4C increase (or decrease if the start point was near a maximum). Only multiples of around 60 years would systematically avoid some effect. (Similar arguments would apply to any other oscillation, of course.)

"The PDO and GAT diverge significantly since ~1980..." There is indeed a divergence of ~0.2C putatively from AGW in the simple scenario I described.

"Lower tropospheric temperature 1980 - 1995. You cannot infer anything from short periods." My point was to highlight the danger of starting just before or ending just after that period, because of major climate events in the late 70's and late 90's. I drew no conclusion about warming from the 1980-1995 trend, since I implied that sub-60-year periods were unreliable.

"It's much more informative to look at a minimum of 30 years... [e.g.] 1979 - present..." My post suggested a ~30-year period could involve a temporary rise of up to ~0.4C. Comparing this 33-year period with another in phase (on the basis of a 60-year cycle), i.e. 1919-1952, would of course be fair. The rises in temp across the two periods would differ by more than 0.01C due to moderate AGW as mentioned above, but otherwise would be fairly similar.

"Actually, it isn't [reasonable to assume continuing recovery from the LIA]. As someone wiser than me once put it, the climate is not a bouncing ball. Nor is there an ideal state to which it is 'recovering' after the LIA. Some combination of negative forcing (possibly TSI and volcanism) depressed GAT and we got the LIA. Then forcing increased and GAT began its discontinuous rise from the end of the C19th to the present. It does this *in response* to external forcing (here including a substantial transfer of carbon from geological sinks to the atmosphere by human activity). There's no physical basis for arguing that ~0.3C of C20th warming is 'recovery' from the LIA." Why necessarily "external" forcing? I'm not qualified to get into a "missing heat" discussion, nor did I intend to get into mechanisms; but it is clear that some temperature-related ocean circulation systems are periodic (e.g. PDO of course) and it is possible that there are longer period oscillations; and, of course (again beyond my ability to comment) there is the debated sun-cloud process. From my simple point of view, assuming the depths of the LIA date to the 17th century, the "recovery" ran for over 200 years; even if this was due to external forcing, why would that forcing conveniently switch off in he early 20th century to pass he baton to AGW. Syun-Ichi Akasofu thinks the same, of course. CAGW sceptics don't suggest that there is an ideal state (unlike True Believers).

"If climate sensitivity is low, it will not respond much to changes in forcing. Variability would be flattened out. The MWP and LIA would either have been much less pronounced or non-existent. The termination of glacials by orbital forcing (Milankovitch) simply could not happen." One answer is that MWP, LIA etc might indeed have been much *more* pronounced without the negative feedback = low sensitivity. That's an experiment we can't perform. (Personally I have never believed in the Milankovitch theory, which seems to rely on an inadequate but convenient cause. I also think there's considerable recent doubt about the fit of orbital cycles to the glacial series.) Another answer, given already in my previous post, is that with positive feedbacks the earth would quite probably have drifted long ago to a completely frozen state or, conversely, a torrid desert.

May 28, 2012 at 12:32 AM | Unregistered Commenterlogicophilosophicus

"You appear to argue that we cannot investigate a phenomenon by comparing its recent behaviour with its past behaviour."

How can you investigate a phenomenon by comparing its recent behaviour with its past behavior, if the past behavior includes the recent?

I don't know, it is quite a simple thing. I don't know why this has to trouble you so much.

Moreover, you still haven't gotten the point about CRUTEM3. The earth has just one temperature history. How do we have three different versions, each going in a different direction? You are not willing to look at the why, yet you want to throw away whole data series just because the Hadley Centre does it.

The differences between the series simply means the much vaunted trend doubling you advertised, is likely just an artifact of method. Even over land, there is no anthropogenic signal. Certainly there is none in the global series.

May 28, 2012 at 12:43 AM | Registered Commentershub