Buy

Books
Click images for more details

Twitter
Support

 

Recent comments
Recent posts
Currently discussing
Links

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

Powered by Squarespace
« Parched earth policy | Main | In the news today - Josh 370 »
Saturday
Apr232016

Gav loses it - Josh 371

When Steve MicIntyre writes "In the past few weeks, I’ve been re-examining the long-standing dispute over the discrepancy between models and observations in the tropical troposphere." you might think you were in for a bit of a technical post - which, of course, it is - but it is also also very funny and well worth reading. It also inspired the cartoon below.

Click image to enlarge

H/t commenter 'See owe to Rich' for the 'hide the gap' phrase.

Cartoons by Josh


PrintView Printer Friendly Version

Reader Comments (155)

A 0.5C difference is the result of a 5-year smooth?

It is not I who is delusional.

Apr 26, 2016 at 10:38 AM | Unregistered CommenterPhil Clarke

Wont someone rid us of this delusional fool?

Mailman

Apr 26, 2016 at 12:31 PM | Unregistered CommenterMailman

No Phil, there isn't a 0.5C difference at the same BAC delta. That's in your head.

At 20 years after BAC in each graph the z-scores are identical. Your 0.5C is a result of comparing apples and oranges.

Apr 26, 2016 at 1:54 PM | Unregistered CommenterSpence_UK

The last post of mine was typed on my phone, so a bit brief. I'll add a bit more to illustrate how wrong Phil Clarke is.

In any comparison, to be apples-to-apples, they must be the same distance from the baseline centre. Phil Clarke innumerately compares data points which are different distances from the baseline centre, resulting in an apples-to-oranges comparison. Quite incompetent, really.

Phil's example includes 2015 from the climate lab book example, an el nino year, some 20 years after the baseline centre of 1995 for that chart. This shows that 20 years after baseline, where the models and observations are aligned to have zero difference, a single year just falls beneath the model mean.

If you now look at Christy's chart, you can see that in 1996, the satellite data falls just fractionally under the model mean - probably within a tenth of a degree, perhaps even closer to the mean than the climate lab book plot at 2015. 1996 is 17 years after the baseline centre of 1979 for Christy's plot.

So in other words, in Christy's plot there is a point 17 years after baseline centre that almost touches the model mean. In the climate lab book plot, there is a point 20 years after the baseline centre that almost touches the model mean. In other words, there is really no difference in this respect between the two.

If Christy hadn't used 5-year smoothing, the 1998 el nino would have almost certainly also got close to the model mean - which would have been 19 years for Christy's plot and 20 years for the climate lab book plot - in other words, both telling the exact same story.

As I've already said, by 2025 the climate lab book plot, if it were to be updated, would very likely show the exact same divergence that Christy's plot already shows by 2009, unless climate scientists are incredibly lucky and some natural warming phenomena (e.g. PDO) goes into overdrive. Odds of this are quite remote though. Then not even Phil's innumeracy will save them.

Apr 26, 2016 at 8:24 PM | Unregistered CommenterSpence_UK

the dumb version for the likes of me is : slope of model trend vs slope of observation trend . baseline no matter . as for where satellite observations taken, this also no matter as temp calibrated to radiosonde data . result,models and phil clarke full of schmidt .
mr mcintyre has his teeth into this. i expect to see further embarrassment heaped on gavin, the models and phil in the not too distant future.
if phil had any confidence in what he states, he would be at climate audit . the fact he is not is very telling.

Apr 26, 2016 at 11:38 PM | Unregistered Commenterbit chilly

Spence_UK & bit chilly

Phil Clarke's responses may not be his own. 'His' attempt to discredit anything or anyone that may damage Hockey Stick credibility does seem part of an organised pattern.

There are the quick pre-prepared responses. These are presumed by their actual authors to be brilliant, and all conquering. Phil Clarke believes them too.

Then things go wrong, when the technically minded raise technical problems that Phil Clarke's experts had not thought of. This is when longer time delays occur, as further instructions are required, possibly co-ordinated from the other side of the world.

All this expertise in computer modelling, and they think their predictable pattern of responses won't be noticed.

Apr 27, 2016 at 12:36 AM | Unregistered Commentergolf charlie

My take on Curry slapping down Schmidt.

Science and sensibility.

Pointman

Apr 27, 2016 at 9:26 AM | Unregistered CommenterPointman

Nobody here is making a better argument than McI and there's more to come on the phoney confidence intervals I gather. As McI noted, Schmidt was silent when Mears did exactly the same thing as Christie (because it is a better way to show the discrepancy). This is only because Mears is also part of the hysterical nonsensus that prefer unvalidated, pessimistic models over actual data and is therefore beyond reproach.

Apr 27, 2016 at 10:32 AM | Unregistered CommenterJamesG

In any comparison, to be apples-to-apples, they must be the same distance from the baseline centre. Phil Clarke innumerately compares data points which are different distances from the baseline centre, resulting in an apples-to-oranges comparison.

LMAO. Really?

They are different distances from baseline centre, but they are the same point in chronological time, ie 2015, but at the same time, different charts show a different model-observations difference, due to the choice of baseline.


Which was my point, actually, which you've just handsomely confirmed.

Apr 27, 2016 at 11:20 AM | Unregistered CommenterPhil Clarke

Hi Phil,

You can shift time series like these up and down to your hearts content and get them to fit in the "error bars". Doesn't tell us anything.

As every rational poster here has already spotted, what matters is the difference in slopes. By base lining early in the series (as was also done by Mears), the difference at the end can be seen very clearly. The models run way too hot. Gavin's version seems to have others issues too, such as averaging runs as opposed to models.

Anyway, as several posters have also already noted, why are you posting your arguments here and not at Climate Audit? Gone on Phil, I dare you to actually stand by what you have written and argue it with Steve McIntyre. I can have the popcorn ready in a moment.

Go on Phil - I double dare you.

Apr 27, 2016 at 12:35 PM | Registered Commenterthinkingscientist

No Phil. You are utterly ignorant.

You are comparing apples and oranges. The graphs deliberately zero the difference between models and observations at a particular year. All comparisons must be made relative to that year.

Only an absolute fool would directly compare the same absolute year. The only valid comparison is year relative to baseline.

And when you compare apples to apples, the graphs show the exact same thing.

Your ignorant comparison is the same as someone comparing temperature sets with different baselines. It isn't the use of different baselines which is the problem, it is the fool that doesn't account for it and make a false comparison.

Apr 27, 2016 at 5:39 PM | Unregistered CommenterSpence_UK

Only an absolute fool would directly compare the same absolute year.

Or a member of Congress, or John Christy.

That's enough insults, if you can't be civil, I'm off.

Apr 27, 2016 at 8:27 PM | Unregistered CommenterPhil Clarke

Or a member of Congress, or John Christy.

No, John Christy only compared data sets with *consistent* baselines. As long as the models and observations have the same baseline, you can validly compare them.

You cannot compare across two charts.

Consider this simple example. Let's say one person puts up a chart of the HadCRUT4 temperature data, with a baseline set to 1950-1980.

Then a second person puts up a chart, same data with a baseline of 1980-2010.

These two charts show *the same data*. However, they are displaced vertically to one another.

Now let's say some ignoramus comes along and says because the temperature in the first graph for a specific year is 0.45 deg C, and in the other graph it is 0.2 deg C, it proves the first graph must be wrong, because there is a 0.25 deg C difference between the two numbers.

Of course, that would be a ridiculous thing to do. There is absolutely nothing wrong with these graphs - it is quite valid to plot temperature anomalies on different baselines, and it doesn't change the data nor any VALID conclusions drawn from them whatsoever. What it is not valid to do is directly compare across two graphs with different baselines. But the fault here is not with the baseline. Either baseline is valid. The fault is entirely with the person who made a direct comparison across the two different baselines.

Yet this is exactly what you did here Phil. You compared a graph with a baseline at 1979, with a graph baselined at 1995, and complained that the temperatures reported seemed different. This isn't a complicated or difficult error to understand. It is a blatant comparison of apples and oranges, and claiming that because apples aren't oranges, apples must be wrong.

I don't think there is any merit in Phil going to ClimateAudit to have his ridiculous error pointed out by Steve McIntyre. The error is trivial and obvious to anyone with basic numeracy skills; but Phil is too obstinate to ever see an error in his own analysis, even something as ridiculous as making an invalid apples-to-oranges comparison.

That's enough insults, if you can't be civil, I'm off.

No Phil, you're off because you've been making invalid comparisons across graphs with different baselines (as opposed to the valid, sensible comparisons you could have made) and you're off because there is no logical or reasonable basis with which to defend what you've said. Usually, in science, it is a very honourable thing to admit error; but you have no honour in this respect.

Apr 27, 2016 at 11:00 PM | Unregistered CommenterSpence_UK

spence, you have the patience of a saint .

Apr 27, 2016 at 11:55 PM | Unregistered Commenterbit chilly

Spence_UK & bit chilly

In fairness to Phil Clarke, I don't think he worked this out wrong without a lot of help.

The Hockey Team Defensive Strategy does not improve, even with increased expenditure and funding.

Apr 28, 2016 at 1:31 AM | Unregistered Commentergolf charlie

There isn't enough crackdown on insulting behaviour on this blog. Insults are for trolls! Folk can agree to differ without resorting to playground name-calling.

Apr 28, 2016 at 10:00 AM | Unregistered CommenterJamesG

Golf Charlie and bit chilly - thanks for your kind words, but I'm not convinced patience is the right word!

JamesG - your desire to rid bish hill of insults is admirable, so I quickly checked google for some examples:

Ignoring your innumeracy, 45% of those who didn't abstain voted for it including several hundred thousand English residents who presumably all voted no.

You may disagree but you should do the legwork first - not just believe a saddo headline from the 12 year-olds that run the Sun, Mail and Express and increasingly the Telegraph too.

There is an unholy alliance of stupid actors in this play but the SNP, having been fooled not so much by the greens (as Lapogus posits), but more by the clueless UK climate scientists, are still least at fault in this particular instance.

Astonishingly, these were on the first three pages containing comments by you to this site. It seems if you want to reduce the amount of playground level jibes on this website, your own comments would be a great place to start, before berating others. Just mind you don't trip over that beam in your eye as you do it.

Apr 28, 2016 at 10:20 PM | Unregistered CommenterSpence_UK

It has been a long time since I saw so much effort being put into missing the point, denying the point, arguing a different point, misrepresenting the point ….

Here it is, again.

Worrying about baseline used for the anomalies can seem silly, since trends are insensitive to the baseline. However there are visual consequences to this choice. Given the internal variability of the system, baselines to short periods (a year or two or three) cause larger spreads away from the calibration period. Picking a period that was anomalously warm in the observations pushes those lines down relative to the models exaggerating the difference later in time. 

Schmidt eviscerates Christy

May 9, 2016 at 11:43 AM | Unregistered CommenterPhil Clarke

I see your point, there, PC. Just like, say, picking 1979 as your baseline for Arctic sea ice extent, when just a little research (see page 61 of the .pdf) will show that the extent was anomalously high, that year. 1974 was only slightly higher than this year’s.

May 9, 2016 at 2:37 PM | Registered CommenterRadical Rodent

Oh wow Phil Clarke links to Gavin Schmidt's astonishing nonsense . . . what a surprise!

The first thing that Schmidt does is compare apples to oranges - yep he plots lines from different baselines on the same chart and then compares the lines directly. You can't directly compare a baseline that centres in 1983/4 with a baseline that centres on 1979 - you need to shift the X-axis to align the centres to form a valid comparison. I can't believe Schmidt has made such a dumb error! McIntyre already notes in the comments in ClimateAudit that Schmidt has made another schoolboy statistics error, which he refuses to admit even though several other climate scientists have made it clear Schmidt is in the wrong.

However, we can make an interesting observation from one of the few valid observations from Schmidt's rather flawed chart. We can directly compare the 1979 alignment with the Christy's method of aligning the baseline at 1979 (based on the trend). This highlights the absurd wrongness of Phil Clarke's earlier claim that Christy's method is sensitive to the exact value in 1979 - a wrong claim by Clarke which he still has yet to acknowledge. I guess Clarke takes his lead from Schmidt - never admit an error even when everyone else in the discussion can clearly see it.

The interesting thing to note is that if you select the single year of 1979 as a baseline, the uncertainty of the models is expanded due to the uncertainty in the baseline. This actually makes it more difficult to use to falsify the model predictions, as the model spread is wider.

Just think about that for a minute. It means a couple of things. The first thing is that the uncertainty of a short baseline is intrinsically captured in the analysis. The additional uncertainty is automatically captured in the spread of model runs.

The second thing to note is that a short baseline favours the models. Yep, there is a bigger spread from the single year example that reduces the likelihood the models will be shown to be wrong.

Of course, the other thing that favours the model is reducing the time between the last sample and the baseline. If I set the baseline to 2005, and test the models at 2015, they only have 10 years to diverge, and the model spread is such that there is not enough time for the divergence to happen, and the models are not falsified. The longer the gap becomes, the more likely it is to diverge (as the model error increases with time, and sufficient time is required for the shift in climate to diverge to be outside of the model spread). Christy's test runs 36 years (1979-2015) compared to Schmidt's graph running around 31 years. In five years time, I suspect Schmidt's plot will look very much like Christy's.

May 9, 2016 at 9:47 PM | Unregistered CommenterSpence_UK

Anyway, ignoring Clarke's misunderstanding of these charts, I'll quickly grade Schmidt's article, as I agree with some of the points he makes.

1) Baselines

Grade: D-. Must try harder
Poor grade. Directly comparing different baselines on the same graph. Tut. The graph also shows that a short baseline favours the models, yet wasn't mentioned. Failed to note the critical factor that a later baseline centre favours the models heavily. Not good enough.

2) Inconsistent Smoothing

Grade: B+. Good point
Good grade. Smoothing here is quite unnecessary and naturally produces problems at the end-points. Of course, Christy's smoothing here is nowhere near as bad as the ludicrous "minimum roughness" farce of Mann's end point handling, nor is it as bad as Kerry Emmanuel's awful bin-and-pin end point handling, neither of which seemed to offend Schmidt at the time. But good to see an old dog can learn new tricks!

3) Model Spread

Grade: E-. Not the way to do things at all
The bad grade here is because even from Gavin's own charts you can see the skewness in the model spread being hidden. Hiding such inconvenience results is very poor practice. Gavin plots a 2.5% to 97.5% spread (approximately two sigma assuming a Gaussian distribution). Given that there are 102 model runs, we expect 2-3 points above the top line and 2-3 points below the bottom line. But we see consistently more than this above the top line, and fewer below the bottom line. The distribution does not look Gaussian - it appears to be skewed. Therefore the 95% confidence intervals Gavin is plotting really go down to the first percentile at the bottom, and around the 96th percentile at the top - extending the confidence interval at the bottom. Gavin does this under the pretense of simplifying the graph, but the consequence of hiding an inconvenient aspect of the model spread is very bad practice. Hence the very low grade.

4) Structural Uncertainties

Grade: C-. Average
It is good to include structural uncertainties in a measurement - but it is better again to use a completely independent source of information, which Christy did with the balloon data set. But it would have been better if Christy had included some other satellite data sets to give an impression of the spread.

Overall Grade: D-. Not good work.
As McIntyre and others have noted at ClimateAudit, the fact that the 95% intervals sort of just about touch is not impressive. That Gavin had to cheat quite a bit to achieve this - shortening the period of test from 36 years to 31 years, and placing his bottom confidence interval nearer to 1% than 2.5% - simply reinforces the original view that Christy presented. At scales beyond 30 years, on out-of-sample data, where we have been told models are supposed to come good, they are doing really very badly. Christy's plot shows this, and despite all the tweaking, Schmidt's plot simply confirms it.

May 9, 2016 at 10:15 PM | Unregistered CommenterSpence_UK

In the above comment I noted that I agreed with Schmidt's point on Christy's smoothing and end-point handling; it is worth noting that as far as I am aware, Schmidt was silent on Mann's appalling end-point handling. UC at ClimateAudit did a great little video of Mann's end-point handling, showing how awful it was and how it tended to produce unprecedented output almost wherever you chopped a data set to here:

Youtube link

Compare that to the barely-visible-on-the-graph transgression by Christy. Oh dear!

May 9, 2016 at 10:26 PM | Unregistered CommenterSpence_UK

LOL. What an unbelivable amount of McIntyresque handwaving.

My fundamental point stands, and has been repeated and confrmed. Christy's graph(s) are misleading , amongst other things because he chose his baseline to maximise the apparent over-estimating of tropical temperature by the models at a point in time. As Gavin correctly surmises, each of Christy's methodological choices 'separately (and even more so together) has the effect of making the visual discrepancy between the models and observational products larger, misleading the reader as to the magnitude of the discrepancy and, therefore, it’s potential cause(s).'

As I wrote , Christy's graph has observations about 0.5C below models in 2015. Spenceboy denies even this obviously true datum and tries to waffle it away with some guff about annual smoothing. He is apparently incapable of giving a yes or no answer to the question ' you believe the comparison between observations and models is insensitive to the choice of baseline?'

Christy's graph is a clear invitation to compare the model-observation discrepancy at the present day. According to Frank the Spencer 'Only an absolute fool would directly compare the same absolute year. The only valid comparison is year relative to baseline.'

Thus, Spence declares John Christy an absolute fool. How true.


;-)

May 10, 2016 at 8:50 AM | Unregistered CommenterPhil Clarke

LOL. What an unbelivable amount of McIntyresque handwaving.

My fundamental point stands, and has been repeated and confrmed. Christy's graph(s) are misleading , amongst other things because he chose his baseline to maximise the apparent over-estimating of tropical temperature by the models at a point in time. As Gavin correctly surmises, each of Christy's methodological choices 'separately (and even more so together) has the effect of making the visual discrepancy between the models and observational products larger, misleading the reader as to the magnitude of the discrepancy and, therefore, it’s potential cause(s).'

As I wrote , Christy's graph has observations about 0.5C below models in 2015. Spenceboy denies even this obviously true datum and tries to waffle it away with some guff about annual smoothing. He is apparently incapable of giving a yes or no answer to the question ' you believe the comparison between observations and models is insensitive to the choice of baseline?'

Christy's graph is a clear invitation to compare the model-observation discrepancy at the present day. According to Frank the Spencer 'Only an absolute fool would directly compare the same absolute year. The only valid comparison is year relative to baseline.'

Thus, Spence declares John Christy an absolute fool. How true.


;-)

May 10, 2016 at 8:51 AM | Unregistered CommenterPhil Clarke

McIntyre already notes in the comments in ClimateAudit that Schmidt has made another schoolboy statistics error, which he refuses to admit even though several other climate scientists have made it clear Schmidt is in the wrong.

Odd, there are two posts at CA on this and I've been through all the comments, nothing on this, and not much at all post-dating the RC post.

Reference/quote please.

And how exactly should the x-axis be adjusted in the baseline graph? What difference does it make?

May 10, 2016 at 11:38 AM | Unregistered CommenterPhil Clarke

Oh.
My.
God.

Seriously Clarke, how do you manage to not understand things that a ten year old engaging in the climate debate can grasp?

Let's start with the easy ones first.

Odd, there are two posts at CA on this and I've been through all the comments, nothing on this, and not much at all post-dating the RC post.

Well, this took me all of two tenths of a second to locate: https://climateaudit.org/2016/05/05/schmidts-histogram-diagram-doesnt-refute-christy/#comment-768877

Please note the issue of Gavin's scientific incompetence, confirmed by Santer, Wigley, McIntyre, Lewis is repeated several times both in the top post and in multiple comments. I really don't understand how you couldn't find it, other than (1) you are trolling or (2) you really don't understand anything being written on this topic. Since I prefer not to assume malice, I am guessing the latter.

According to Frank the Spencer 'Only an absolute fool would directly compare the same absolute year. The only valid comparison is year relative to baseline.'

Thus, Spence declares John Christy an absolute fool. How true.

This is another example of not understanding the simplest thing in this debate. Christy has three lines on his chart:

Models, referenced to 1979 from a 37-year trend
Satellites, referenced to 1979 from a 37-year trend
Balloons, referenced to 1979 from a 37-year trend

You notice my bolding above. All three sets of data are referenced to the same baseline. The same baseline! Imagine that. It allows you to validly compare the results! Awesome. And that's exactly what Christy does in his chart. In other words, he makes a valid comparison as I've explained it.

Now let's look at Climate Lab Book. All of the lines in this chart are referenced to a 20-year mean. That means it is valid to compare results between the lines of this chart. Ed Hawkins does it right as well! Good stuff.

Then you come along. You compare lines on Ed's chart directly with values on Christy's chart. BZZZZ! Wrong answer! These series have different baselines and cannot be directly compared with one another. So far we have:

John Christy - right
Ed Hawkins - right
Phil Clarke - wrong

Now let's go to Gavin's chart. He compares confidence intervals based on:

Models centred on 1979 from one year of data
Models centred on 1979 from a 37-year trend
Models centred on 1981 from a 5-year average
Models centred on 1983/4 from a 10-year average

BZZZZ! Wrong again! These are centred on different years and it is wrong to directly compare them! So to summarise we have:

John Christy - right
Ed Hawkins - right
Phil Clarke - wrong
Gavin Schmidt - wrong

See? This is really simple. As I say, even a 10-year old can follow it. Don't compare across different baselines. Hawkins, McIntyre and Christy and I have been careful to follow this convention which is correct. You and Schmidt have failed to do so. You are wrong.

A set of rules so simple even you could follow them, Phil.

May 10, 2016 at 9:31 PM | Unregistered CommenterSpence_UK

Phil is a nice example of of motivated reasoning. If he was being paid to deny gravity, his strength of faith would be sufficient to lead him to demonstrate the lack of gravity by throwing hi hit anything.mself at the ground, as and then claim he hadn't

May 10, 2016 at 10:42 PM | Unregistered Commenterhunter

The laughs keep on coming!

Spence told us, in relation to Gavin's May 7th post, and specifically the baseline comparison graph that

McIntyre already notes in the comments in ClimateAudit that Schmidt has made another schoolboy statistics error

Asked to back this up, with y'know, facts we get a comment from Nic Lewis on a tangentially related topic. There is no comment from McIntyre. Liar.

And that was the 'easy one'. Ooops.

Next , the trademark condescension is ladled out on the topic of aligning baselines, apparently Christy's graph is Ok because his curves are anchored at the same point, 1979. Spencey, this is what I've been saying.

Anyone can see that the model-obs delta in the ChristyChart at 2015 is around 0.5C. According to Spence Christy used:-

Models, referenced to 1979 from a 37-year trend
Satellites, referenced to 1979 from a 37-year trend
Balloons, referenced to 1979 from a 37-year trend
You notice my bolding above. All three sets of data are referenced to the same baseline.

So 2015 must be the at same difference from baseline centre (BAC to use Spence's invented terminology), right?. But what's this? This is the same Spence who so-politely told me that

No Phil, there isn't a 0.5C difference at the same BAC delta. That's in your head.

Oh, but there is, tragically, there it is in black and white. The Dunning Kruger is strong in this one, I am just shocked, shocked I tell you that a scientist with a BA (Hons) in Mathematics from Oxford University, and a PhD in Applied Mathematics from UCL can plot a graph better than the pseudonymous and demonstrably dishonest blog poster who hilariously feels qualified to grade the former's work.

Better than the telly.

May 10, 2016 at 10:45 PM | Unregistered CommenterPhil Clarke

Spence told us, in relation to Gavin's May 7th post, and specifically the baseline comparison graph that

Ah. The intellectual dishonesty we have come to expect from Phil Clarke.

My point was simply that Gavin has prior examples of statistical incompetence, on the analysis of comparisons between models and data. Which is exactly what I linked you to. And which anyone with half a brain could find. I guess your quarter brain was not sufficient.

Asked to back this up, with y'know, facts we get a comment from Nic Lewis on a tangentially related topic. There is no comment from McIntyre.

McIntyre responds to the Lewis comment I linked to, agreeing to it, on May 7, 2016 at 2:05 PM. He describes the point in great detail in more than one place on that page.

You know what's funniest? You are too incompetent even to read the replies of the points I link to, get everything completely wrong, and then conclude because you get everything wrong, that somehow *I* am a liar!

You are utterly pathetic Clarke. Given that McIntyre replies, confirming the statistical error at the comment above, exactly as I said he did, and that the error is specifically about setting confidence intervals in model-observation comparisons, you now have two options.

You can actually man up, be less of an intellectually dishonest weasel, and apologise for calling me a liar when the evidence is clear for all to see.

Or you can carry on being an intellectual dishonest weasel.

My guess is one the latter, but I'm ready to be surprised.

No Phil, there isn't a 0.5C difference at the same BAC delta. That's in your head.

At the same BAC delta with respect to the Climate Lab Book graph, which is exactly what I explained to you, you ridiculous fool.

Because the same BAC delta doesn't exist on the Climate Lab Book graph.

The 0.5 degree difference on Christy's graph is simple. It's because the models are wrong. But the comparison is correct to do.

The same difference will appear on the climate lab book graph in at about the same BAC delta. It's easy to calculate.

The BAC delta for 2015 on Christy's graph is 2015-1979, or 36 years.

The same point would be reached at the climate lab book graph, centred on the year 1995, in the year 2031. When we reach 2031, we will be able to make the same comparison on the climate lab book graph. My bet is that we will see a similar gross disagreement between models and observations as we see on the Christy graph.

This is really not complicated stuff. It's exactly what I've said all along. You are just too innumerate to understand what I'm saying. As always, you can lead a climate alarmist to data, but you can't make him think.

May 10, 2016 at 11:11 PM | Unregistered CommenterSpence_UK

Intellectual dishonesty. Yep. Full context:

The first thing that Schmidt does is compare apples to oranges - yep he plots lines from different baselines on the same chart and then compares the lines directly. You can't directly compare a baseline that centres in 1983/4 with a baseline that centres on 1979 - you need to shift the X-axis to align the centres to form a valid comparison. I can't believe Schmidt has made such a dumb error! McIntyre already notes in the comments in ClimateAudit that Schmidt has made another schoolboy statistics error, which he refuses to admit even though several other climate scientists have made it clear Schmidt is in the wrong.

I can understand the desire to rewrite history, but clearly you were stating that there was a comment at CA from Mcintyre about Gavin's 'statistics error'. Which, unless you can produce the comment, was a lie, you liar.

At the same BAC delta with respect to the Climate Lab Book graph, which is exactly what I explained to you, you ridiculous fool.

Nope. It was the Christy graph. You apparently are idealogically incapable of admitting Christy's fundamental dishonesty, laid bare by a better scientist.

My last word.

May 10, 2016 at 11:35 PM | Unregistered CommenterPhil Clarke

I can understand the desire to rewrite history, but clearly you were stating that there was a comment at CA from Mcintyre about Gavin's 'statistics error'.

There are multiple comments from McIntyre about Gavin's statistics error, including one in reply to the comment I linked you to, at May 7, 2016 at 2:05 PM.

Which, unless you can produce the comment is a lie, you liar

For the third time, I've already produced the comment. Follow the link I gave. McIntyre's reply is at May 7, 2016 at 2:05 PM. There is no lie.

Nope. It was the Christy graph.

This is what you wrote. Your comment, here, at Apr 25, 2016 at 11:32 PM

Yeah, and in one the observations are more than 0.5C below the models, in the other they are splat in the middle. Because: baselines.

The point *you* made, which I was rebutting, was that there was a 0.5C discrepancy between Christy's graph and the Climate Lab Book graph. Note you directly compare values in 2015 between the two graphs, even though it is fundamentally wrong to make such a comparison. The discrepancy is because you are comparing apples and oranges. The numbers are *correct* in Christy's graph and *correct* in Climate Lab Book - but you cannot make the comparison that you made. I predict the climate lab book graph will show the models are wrong in 2031, in just the same way that Christy's graph shows the models are wrong today, as I've already explained.

Nope. It was the Christy graph. You apparently are idealogically incapable of admitting Christy's fundamental dishonesty, laid bare by a better scientist.

I've already shown that you were directly comparing the Christy graph to the climate lab book graph by quoting your exact comment. It is the comparison I objected to. Talking of Schmidt, this is a "better scientist" who, as Nic Lewis points out, does not know the difference between a population standard deviation and a sample standard deviation and when to use the two. The statement "better scientist", in this context, is meaningless. Here's the thing: a PhD in applied maths doesn't mean you know the first thing about statistics. But herein lies the problem: not only does Gavin Schmidt struggle with the difference between biased and unbiased estimators, since you don't understand the difference either, you have no way of knowing that Gavin is wrong.

My last word.

Of course, I've shown every single thing you claim is nonsense and the only thing left you've got is to call me a liar even though I've provided you with the exact comment that shows I'm not. But you are incapable of admitting error, so you just come out with some blowhard bluster and run away. Because your claims are without any merit whatsoever.

Bye bye. You won't be missed.

May 10, 2016 at 11:59 PM | Unregistered CommenterSpence_UK

There are multiple comments from McIntyre about Gavin's statistics error, including one in reply to the comment I linked you to, at May 7, 2016 at 2:05 PM.

But Spence dishonestly claimed it was about Gavin's baseline graph. It was not, Spence lied, it is what liars do.

The point *you* made, which I was rebutting, was that there was a 0.5C discrepancy between Christy's graph and the Climate Lab Book graph.

How can you tell when Spence is lying? It is when he is typing. He gave two references, the first

For convenience, the two graphs I was comparing:
Christy can be found at Judy Curry's place:
https://curryja.files.wordpress.com/2016/04/slide1.png

The response…

Yeah, and in one the observations are more than 0.5C below the models

That would be https://curryja.files.wordpress.com/2016/04/slide1.png

In which the discrepancy is indeed >0.5C.

It really does not matter if Spence is dishonest or incompetent, the net result is the same.

Ho ho.

May 11, 2016 at 12:27 AM | Unregistered CommenterPhil Clarke

Wow Phil, I knew you were innumerate - that's a given - I didn't realise you were illiterate too. Is English your second language?

But Spence dishonestly claimed it was about Gavin's baseline graph.

No. I didn't claim the second statistical error was about Gavin's baseline graph. I just noted that Gavin made another error.

Here's your lesson in the English language, Phil. What I said was this:

McIntyre already notes in the comments in ClimateAudit that Schmidt has made another schoolboy statistics error, which he refuses to admit even though several other climate scientists have made it clear Schmidt is in the wrong.

This simply notes that Gavin has made another schoolboy statistics error. I do not, at any stage, claim this additional error is on the same graph. Had I added the qualifier in the same graph you would have had a point. But those words aren't there. You simply incorrectly assumed I was talking about the same graph, and then projected this incorrect assumption that you made on to me.

You see Phil, us adults can keep more than one related concept in our minds at the same time. Gain's additional error is related - it is discussed at the same climateaudit posts, and it is about model-observation comparisons, but in relation to a different graph. Related, but different.

Your claim is a fabrication cut from whole cloth that exists only in your imagination.

How can you tell when Spence is lying? It is when he is typing.

Phil, it is clear to all that it was the incorrect comparison you made between graphs I objected to, not the lines on a single graph (which are quite correct and as they should, show a gross inconsistency between observations and models). I explained this in great detail, and I even quoted your exact words where you are comparing the two graphs. Stop trying to pretend it is only about one graph. It just makes you look even more stupid.

May 11, 2016 at 9:11 AM | Unregistered CommenterSpence_UK

Phil Clarke, have you been copying someone else's homework again, and submitting it as your own work?

May 11, 2016 at 10:06 AM | Unregistered Commentergolf charlie

This simply notes that Gavin has made another schoolboy statistics error. I do not, at any stage, claim this additional error is on the same graph. Had I added the qualifier in the same graph you would have had a point. But those words aren't there. You simply incorrectly assumed I was talking about the same graph, and then projected this incorrect assumption that you made on to me.

No Spence, you have selectively quoted yourself.

The first thing that Schmidt does is compare apples to oranges - yep he plots lines from different baselines on the same chart and then compares the lines directly. You can't directly compare a baseline that centres in 1983/4 with a baseline that centres on 1979 - you need to shift the X-axis to align the centres to form a valid comparison. I can't believe Schmidt has made such a dumb error! McIntyre already notes in the comments in ClimateAudit that Schmidt has made another schoolboy statistics error, which he refuses to admit even though several other climate scientists have made it clear Schmidt is in the wrong

There is only one reasonable interpretation of that paragraph - that McIntyre has 'already' agreed with Spence and commented on 'the error'. It would take something of a mindreader to understand that when Spence says 'McIntyre' he actually means Nic Lewis drivelling on about a topic totally unrelated to baselining -and an even more savant clairvoyant to realise that when he says McIntyre comments on the error he actually is referring to McIntyre's response to Lewis, which is in reality a totally-content free backslapping circle-jerk of McIntyre congratulating Lewis on how smart they both are ….

In a more charitable mood, one could attribute this bollocks to sloppy writing, or thinking but when the bollocks always flows in the same direction ….

I explained this in great detail, and I even quoted your exact words where you are comparing the two graphs. Stop trying to pretend it is only about one graph.

My exact words were

and in one the observations are more than 0.5C below the models, in the other they are splat in the middle. Because: baselines.

Which elicited the response

No Phil, there isn't a 0.5C difference at the same BAC delta. That's in your head.

I wrote there was a large differential in 'one'. I did not write 'in two', 'across two graphs', 'compared to' or any other permutation. I referred unambiguously to one graph, in which there is indeed a 0.5C model observation discrepency and which is misleading, to put it mildly.

Spence's posts are demonstrably riddled with bullshit. A pattern emerges, handwaving, and insignificant nitpicks (the model distribution is not strictly Gaussian - so what? It’s a good enough approximation) unable to stomach or argue with the facts he constructs elaborate alternative points and argues these instead, he fabricates his own BS rules (it is not legitimate to plot graphs from different baselines on the same x-axis) and of course anyone who doesn't play by his rules is an innumerate, ridiculous fool.

There are far better uses of time than wild goose chases for non-existent posts and engaging with a condescending bully who just makes stuff up. I leave him to mathturbate over Montford and McIntyre, they deserve each other.

May 11, 2016 at 10:14 AM | Unregistered CommenterPhil Clarke

Spence_UK
Most folk understand the difference between clever sarcasm and personal abuse even if they are not big enough to admit when they have descended into the latter. eg, innumerate is fine, absolute fool is not! If I have ever personally abused anyone, and I challenge you to find it, then I'd hope such a comment would be removed as well.

May 11, 2016 at 10:35 AM | Unregistered CommenterJamesG

ATTP on Schmidt's 'error'

And Schmidt on Schmidt's 'error'

d) If your CI values purport to be based on variance of the mean, they appear to be out by a factor of about two. [This is not a problem arising from the difference between Gaussian and t-distribution.]

[Response: I checked this. The uncertainties in the Table S1 are the 90% spread in the ensemble, not the standard error of the mean. The difference is the sqrt(n) term (=2.24). – gavin]

May 11, 2016 at 12:42 PM | Unregistered CommenterPhil Clarke


The first thing that Schmidt does is compare apples to oranges - yep he plots lines from different baselines on the same chart and then compares the lines directly.

This is intentional. He's trying to illustrate how the choice of baseline influences both the trend and the spread. This is the key point! What he's showing is that it can depend on how you choose to baseline. At the end of the day, if you want to compare observations and models, you will have to make a choice about how to baseline your data. How you do so will clearly influence the resulting comparison.

May 12, 2016 at 9:23 AM | Unregistered Commenter...and Then There's Physics

Yes, ATTP that is what I thought, however the poster who regards me as innumerate, illiterate and believes himself in a position to grade Dr Schmidt's work says you cannot do it legitimately.

I asked why a few days ago, surely we will be enlightened soon.

May 12, 2016 at 10:03 AM | Unregistered CommenterPhil Clarke


however the poster who regards me as innumerate, illiterate and believes himself in a position to grade Dr Schmidt's work says you cannot do it legitimately.

Indeed, the expertise some pseudonymous commenters on social media is really quite impressive.

May 12, 2016 at 10:19 AM | Unregistered Commenter...and Then There's Physics

Anyone else noticed that

Phil Clarke = Politically Correct

May 12, 2016 at 10:35 AM | Unregistered CommenterGras Albert

ATTP

Indeed, the expertise some pseudonymous commenters on social media is really quite impressive
Indeed Kenny, almost as impressive as the statistical naivety of some leading climate scientists ...

May 12, 2016 at 11:08 AM | Unregistered CommenterGras Albert

Gras Albert,

*Leading*??

Surely the only acceptable course of action for climate scientists is to *follow*, thereby avoiding any risk whatsoever of heresy/excommunication, etc.

May 12, 2016 at 9:08 PM | Registered Commenterflaxdoctor

Most folk understand the difference between clever sarcasm and personal abuse

Oh, this is priceless. JamesG, I knew you'd come back and come up with a ludicrous slippery distinction as to why your insults were just fine but others were not. Yawn. I've already quoted childish insults from your posts up above, although I'll leave you as a "saddo 12 year old" since apparently that's clever sarcasm, not personal abuse. ROFLMAO. Clever sarcasm just ain't what it used to be.

May 12, 2016 at 10:18 PM | Unregistered CommenterSpence_UK

This is intentional. He's trying to illustrate how the choice of baseline influences both the trend and the spread. This is the key point! What he's showing is that it can depend on how you choose to baseline.

And the way he did it was inappropriate, and blurred several issues unnecessarily.

I have no problem with the idea of showing the impact of different baselines if he had used the same baseline centre. For example, you can show the following set of baselines:

1985 only
1984-1986
1983-1987
1982-1988
1981-1989
1980-1990
37 year trend projected to 1985

If you do all of these things, the baselines are all centred about 1985 and you can compare apples-to-apples.

But that isn't what he did.

Let's take an extreme example of what he did. If I plotted an eleven year baseline at 1980-1990, and a second eleven year baseline at 1990-2000, they would have similarish spread, but they would be offset relative to one another. But there is nothing *wrong* with either baseline. They are the same length. The offset is a function of the different baseline centres, which are ten years apart.

This is exactly the problem with Gavin's flawed chart. He wants to show the impact different baselines have but then blurs everything by using different baseline centres, so we don't know how much of the apparent vertical shift is due to the baseline centre, and how much of it is due to the choice of baseline.

As I already pointed out though, a narrow baseline manifests itself as a broader spread in the model outputs, which in turn reduces the chances of model falsification. So a narrow baseline is actually favourable to the models. This is visible in Gavin's chart, although only by comparing the single year 1979 vs. trend to 1979 (as the former has a greater spread) - the rest of the waters are muddied by the baseline centre shift.

May 12, 2016 at 10:29 PM | Unregistered CommenterSpence_UK

I wrote there was a large differential in 'one'. I did not write 'in two', 'across two graphs', 'compared to' or any other permutation. I referred unambiguously to one graph, in which there is indeed a 0.5C model observation discrepency and which is misleading, to put it mildly.

Phil, I don't understand whether you are delusional or not. The point I made is simple to understand, but you seem determined to misinterpret it into something nonsensical.

You said there was a differential in one and not in the other. You were comparing across two graphs.

I agree there is a differential in one. I explained why. It is because the models and observations diverge over a period of 36 years from the baseline centre. This is because the models are wrong. Simples.

The second graph does not show model-observation divergence and 36 years after baseline centre, because such a thing is not present on the graph. The year 2015 is only 20 years after baseline centre on the climate lab book graph. Hence there is no difference at the same BAC delta, because such data does not exist on the second graph. There is no reasonable way to interpret this comment other than comparing across the two graphs.

Put simply: which other baseline centre delta do you think I was referring to? Essentially, this question alone shows that your interpretation of my comment is utterly unreasonable and makes exactly no sense. Which just makes it all the more entertaining when you claim I'm a liar, since this notion lives only in your head.

May 12, 2016 at 10:37 PM | Unregistered CommenterSpence_UK

I asked why a few days ago, surely we will be enlightened soon.

I've already explained several times to you Phil, it's like plotting GISTEMP on a 1951-1980 baseline, then plotting the exact same graph on a 1981-2010 baseline, and getting confused as to why the data point for 2015 is at a different point on the Y-axis on each graph. But despite explaining this remarkably simple concept to you repeatedly, you still don't understand. I can't fix that: you have to fix it yourself, and if you're unwilling to learn the basics, there is not much I can help you with.

You know, it's funny, but years back when I first started getting involved in the climate debate it was usually sceptics incorrectly comparing across baselines and activists pointing out their error. Odd how that now seems to have reversed itself.

May 12, 2016 at 10:43 PM | Unregistered CommenterSpence_UK

Indeed, the expertise some pseudonymous commenters on social media is really quite impressive.

... says the snarky guy who spent years as a pseudonymous commenter on social media. Huh.

But no matter. I have old school science training, which means I assess arguments not on who says them, but on their merit. Notably, neither you nor Phil have managed to lay a technical finger on any of my grading so far. Phil has massively misrepresented my points (the usual straw man - quite a common trait amongst climate activists these days) and you don't seem to have yet properly understood the point I was making about baselines.

I wish there was a higher standard of climate activist who actually *listened* to sceptic arguments and responded directly to them, instead of creating oddball straw men out of whole cloth and responded to those instead. Don't worry though, I won't be holding my breath.

May 12, 2016 at 10:48 PM | Unregistered CommenterSpence_UK

Spence,
Have you considered the possibility that the start point of Gavin's baseline periods is being set by the start of the satellite data and by what Christy chose to do?


I have old school science training, which means I assess arguments not on who says them, but on their merit.

If someone has to tell me what they do, they probably aren't making it obvious that it is what they actually do. Also, appeals to ones own authority are remarkably tedious, especially as you are still a pseudonymous self-professed expert.

May 12, 2016 at 11:42 PM | Unregistered Commenter...and Then There's Physics

It is because the models and observations diverge over a period of 36 years from the baseline centre.

You can attempt rewrite the thread as much as you like, however my point was that the discrepency in Christy's graph (it's more like 0.6C than the 0.5C but I'm eyeballing) was exaggerated by his choice of achoring everything to 1979. baseline. Numerous people have confirmed that this discrepency is real and not 'in my head', as you asserted - amongst many arrogant, offensive and unnecessary jibes.

To say that Christy put his finger on the scale, well, those were pretty heavy fingers

http://rabett.blogspot.co.uk/2016/05/jelly-beans-gavin-steve-attp-lucia.html

May 13, 2016 at 9:08 AM | Unregistered CommenterPhil Clarke

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>