[alancalverd]

It's quite clear where the "skeptical" graph came from, but I'm interested to know what its authors actually plotted. Data that suggests that some winters are warmer than their adjacent summers deserves serious investigation.

The "metaclimate" graph does at least state a geographical area. Amazingly, it includes data from the North Pole in 1880, at least 30 years before anyone actually got there, and probably 90 years before anyone measured temperatures above 80 deg latitude for an entire year.

Skeptic? Moi? No, just wondering how much "climate data" has been falsified, and why it was done in such a transparently amateurish manner..

[Bored chemist]

"Amazingly, it includes data from the North Pole in 1880"
No, it doesn't and, of course, nor does it pretend to.
Here's the graph's title
"Here’s one of his graphs, covering 70 – 90 deg north latitude"

And here's where that data is explicitly stated to be from
"stations north of 70N latitude — instead of defining separate grid boxes for stations north of 80N latitude, I’ll lump them together with the stations north of 70N latitude."
So the graph is the combination of all the data North of 70 degrees.
That data may be a bit sparse, but that's not the same as saying it's impossible.
Such a comment might be thought of has having " been falsified" in "transparently amateurish manner"

No, I don't think you are a sceptic. A sceptic would look at the data to see what it says, rather than launching a strawman attack on what it doesn't say.

[alancalverd]

The underlying problem is that the world is not a smooth, stationary, homogeneous surface. The air above it moves to transport heat, and clouds form and disperse in different ways at different points. This makes it impossible to generate a sufficient sample from which to extract a meaningful mean, if substantial areas are not mapped - you have to survey the entire surface with a constant finesse, and ensure that your data points are not in any way "special" or time variant. Hence historic temperature observations, which in the main are necessarily "special" and time variant, are not useful in a discussion of trends in global heat exchange or climate.

Fortunately we do have recent satellite data, which can give us good random samples of constant finesse over the entire surface . On the downside there have been significant recalibrations of that data (who ever said physics was easy?) but generally we can accept that satellite data from say 1990 onwards is consistent and representative.

I approve of your definition of a sceptic. So I looked at the "skeptical science" graph and it clearly says that some winters were warmer than the adjacent summers, so once again I ask what is the source data for this interesting graph? It's the counterintuitive that makes life interesting.   

[Bored chemist]

Since Summer and Winter are local events, it's easy to see how they might get messed up if you took a global average: do you mean Australia's Summer or England's?
Also, if I wan't to know if the temperature in, for example, London, is changing over time, I clearly don't need to measure the temperature in Antarctica.
All I need are a series of measurements in London.
If those are going up then London is warming.
There are plenty of such measurements going back into history and, if we don't assume that those doing the measuring were incompetent, then we can track the local temperature changes for many places.
In fact, they are generally rising..
As I have said, if you put a forth blaknet on the bed and you get warmer, it's hard to rule out cause and effect.

[alancalverd]

I have no idea what the "skeptical science" graph represented - it was your suggestion, and you still haven't told us! However if it represents a true global average, you wouldn't expect to see any seasonal cycles, and if it represents one or other hemisphere, or even one point on the globe, you'd expect the cycles to be consistent. But it showed significant, inconsistent cycles, which makes it very interesting.

The temperature in London is indeed changing with time, because the surface albedo, the concentration of human activity, and the nature of that activity, are all tending to increase the outdoor temperature. But the question we are trying to answer is about the global effect of increasing CO2, not the local effect of urban heat islands, which even the most ardent warmists agree are anomalous. Interestingly, I think the highest local concentration of CO2 in London was probably in the 1940's and 50's when everything was coal fired - and it was a lot colder than recently!

The "fourth blanket" is an interesting analogy. As you add more blankets, so the incremental effect of each becomes smaller. It's called "close" in clothing design, and "saturation" in infrared spectroscopy. If you add a sheet of paper to a continuously changing heap of blankets, you'd be hard put to pin down its causality from observation. And if,  as in the geo-historic CO2 case, the additional blanket always arrived after the temperature had risen and departed after the temperature had fallen, you might question which was the cause and which the effect.

Causality demands correlation, consistent sequencing, and the elimination of co-causation. Until you have demonstrated all three, you can't use a presumption of causality to predict anything qualitatively. And if you want to make a quantitative prediction, you need a good handle on the effects of nonlinearity and saturation. Unfortunately, all the evidence I have seen to date points strongly to temperature being the cause of CO2, not the other way around, so it's difficult to answer the original question.

[JP]

This reminds me of an interesting book I just picked up: The Norm Chronicles by Michael Blastland.  It's about our perceptions of risks vs. the actual statistics.  There's a nice section on climate science which points out that it's virtually impossible to convince die-hards on either side, and this thread backs that up.  His argument is (and he cites a bunch of psychological studies to back this up) that people generally choose a side due to their politics, and since climate science can't be 100% proven either way, they entrench their opinion with whatever ideas they understand, be those scientific or not.  It's interesting that on a science site like this, any post on climate change brings out a lot of people arguing against the scientific consensus, and that virtually all of their arguments don't use anything approaching a scientific methodology.  This strongly argues for better scientific literacy among the voting public (at least among those of us who think policy decisions should be based on sound science.)

Most of the arguments here against climate change are "not even wrong" as Wolfgang Pauli famously put it.  Sure, many are technically true, but they don't have much to do with the scientific method and aren't useful in evaluating climate science one way or the other.  As I've pointed out in a prior thread (and no one bothered to address), the scientific method of climate change involves two major steps:
1) Collecting and analyzing data on climate (primarily temperature).  This means collecting data from modern satellites, historical temperature readings with thermometers, ice cores and so on.  Some of these data sources are more accurate than others, which is where analysis comes in.  We've had a few centuries to perfect our methods of statistical analysis, and so scientists can estimate both mean temperatures (over regions or the whole globe) as well as the uncertainties of those mean estimates.  Arguments against temperature records in this thread ignore those uncertainties.  Sure, the plot of means might or might not be true, but it's important to look at the ranges within which we are 95%, 99%, 99.99999%, etc. that the true mean temperatures lie.  In the end, it's how confident we are of the trend that matters.

2) The second step is coming up with a model that explains and predicts the data.  It's easy to come up with a best fit curve through some section of data and claim that temperature is falling.  It's also just as easy to pick some section of data which shows a massive upward trend and predict we'll all bake to death in a few decades.  What's important is to come up with a model (not just a best fit curve) that makes predictions based on some underlying science (such as the physics/chemistry of CO₂), which matches the data.  What do we mean by matching the data?  Again, you have to take a scientific perspective which has been woefully lacking in this thread.  You can compute what are called confidence intervals which tell you that if this model is correct, the data will lie in this range with X% certainty.  Based on that, you can't tell if a model is correct, but you can tell if it's consistent with the data.  You can even say how certain you are that the model is wrong--so if enough data falls outside of the 95% confidence interval, for example, you can throw out the model.

Since this is a science forum, we should be evaluating criteria on these bases, not on the basis of nitpicking details and ignoring the extensive body of literature analyzing both data and models.  Nitpicking details while accounting for 1) and 2) is very useful however, so I look forward to seeing Alan and Henry post some details about how their points correlate with the confidence intervals on the models or uncertainties in measured data. 

[alancalverd]

Nothing to do with me, JP, but when someone publishes an untitled graph which contains counterintuitive data, I'd like to know what it represents and why it behaves that way. Is that nitpicking or just asking the sort of question that we professional scientists are paid to ask?

Politics? I'm a dyed-in-the-wool atheist leftie tree-hugger. But that derives from studying the data, not the other way around.

A model based on the nonlinear greenhouse effect of water goes a long way to explaining the Vostok ice core data over periods of thousands of years, and recent Mauna Loa data clearly shows the consistent lag of CO2 behind temperature, but the inherently chaotic nature of the planetary atmosphere makes short-term prediction a very risky business.

My preference is always for clean, raw data. Hence Mauna Loa, which represents a "good site" with no obvious CO2 anomalies or heat island effects, and Vostok, which has used the same data collection process for millions of years, are more likely to yield understanding of the process of climate change than any attempt at meta-analysis of incoherent data and proxies.

History has shown that we should be wary of "scientific consensus". Phlogiston, the geocentric universe, the aether, the flat earth, aristotelian gravitation....all held sway as consensus at some time. Early on in our careers, we learn that data is more important. To paraphrase Einstein, when confronted by a debunking consensus paper signed by 100 Nazi professors: "If I had been wrong, one student would have been sufficient." So let's look at the data, please. 

Regarding uncertainties and sampling intervals, here are some samples of a set, each with negligible uncertainty

1, 16, 23, 45, 48, 49, 51, 60, 74, -5, -7, -23, -60, -80 

What is the mean? Well it's about 13.7. But these numbers are "samples" of latitude, reported with something approaching the frequency of, say, air temperature reports. Most meteorologists live in the northern bit of the planet, hence the preponderance of numbers around 40 to 50,  and we have a few reliable reporters from the polar regions. But the mean of these samples tells us nothing about the mean latitude of the planet, which is of course 0.  Same problem with terrestrial-based temperature records: however precise they may be, they only tell us about the temperature in places where people live, and even if we correct for heat island effects, the mean is globally meaningless. You can only determine global trends by evenly sampling 100% of the surface (which was not possible before the 1970s) or smoothing the curve of an unequivocal proxy at one fixed point.   

[MoreCarbonOK]

JP says
that people generally choose a side due to their politics,
Henry says
well...in my case it is actually religion, which demands that I speak the truth at all times,
I determined no influence of the CO2
it is rather natural forces, that show decline in temps. from 2002
as proven to you from all available data sets
and that this decline in temps. will continue
as shown to you from my own data set (with data from 1974-2012)

but I am intrigued to know as to why you did not respond to my question to you raised in my earlier post, specifically addressed to you?
(which seems to confirm some kind of bias from you which I have noted before)

rgrds
H

[JP]

but I am intrigued to know as to why you did not respond to my question to you raised in my earlier post, specifically addressed to you?
(which seems to confirm some kind of bias from you which I have noted before)

I'd be surprised if you hadn't noted my bias.  In fact, I flat out told you before that I was done debating you on the topic since you use your posts to promote your own model based on cherry-picked data and a best fit curve.  This is a science forum, so I'm biased towards having science-based discussions.   I also don't like repeating myself like a broken record.  Either you don't understand the scientific method, or you're deliberately trolling us.  In either case, since you've made over 100 posts now without making an effort to make your ideas scientific, I don't see why engaging you would be helpful to either of us or to the forum.

[MoreCarbonOK]

jp says
...to promote your own model based on cherry-picked data and a best fit curve.
henry says
I told you from the beginning that my sample of 47 stations was random,
except for the fact of the choice of stations with complete or nearly complete records...
even choosing more stations won't change the result
if you get a correlation coefficient of 0.997 on the binomial for the drop of maximum temps.....
Either way, even if you believe I am trolling, the question to you was about the error bars,
on the gistemp data set,
which you state were there
but they were not...

[JP]

Nothing to do with me, JP, but when someone publishes an untitled graph which contains counterintuitive data, I'd like to know what it represents and why it behaves that way. Is that nitpicking or just asking the sort of question that we professional scientists are paid to ask?

Yes, but a proper scientist would probably find out what an unlabeled plot represents before saying that it casts doubt on many temperature records.

A model based on the nonlinear greenhouse effect of water goes a long way to explaining the Vostok ice core data over periods of thousands of years, and recent Mauna Loa data clearly shows the consistent lag of CO2 behind temperature, but the inherently chaotic nature of the planetary atmosphere makes short-term prediction a very risky business.

Great!  I've seen many claims like this, so show us a model that's been peer-reviewed and has a proper confidence interval analysis so that we can judge if it's consistent with the data! 

My preference is always for clean, raw data. Hence Mauna Loa, which represents a "good site" with no obvious CO2 anomalies or heat island effects, and Vostok, which has used the same data collection process for millions of years, are more likely to yield understanding of the process of climate change than any attempt at meta-analysis of incoherent data and proxies.

In any single experiment, I agree--all scientists would prefer clean data.  But your argument doesn't hold up, since averaging many lower-quality measurements together can actually produce cleaner statistics (smaller uncertainties) than using a small number of high-quality measurements, especially when we're trying to get a handle on a global mean, rather than local means.  Since climate science is observational, we have to live with what data we have, so it quite probably turns out that large-scale averages produce better estimates than using the cleanest datasets.

History has shown that we should be wary of "scientific consensus". Phlogiston, the geocentric universe, the aether, the flat earth, aristotelian gravitation....all held sway as consensus at some time. Early on in our careers, we learn that data is more important. To paraphrase Einstein, when confronted by a debunking consensus paper signed by 100 Nazi professors: "If I had been wrong, one student would have been sufficient." So let's look at the data, please. 

You're shooting yourself in the foot here: yes, data is more important than consensus, but data without analysis is meaningless.  I recall doing an experiment in undergraduate lab where my data, taken at face value, would prove Newton's second law wrong.  Sadly, I didn't win the Nobel Prize for proving Newton wrong, of course, because there were huge uncertainties and errors in my measurement process.  And that's precisely my point--whether you cherry pick data and fit a curve through it or claim that "clean" data is better than the data that is currently being used, you have to back that up with an analysis that favors your version over the consensus version.  Scientific consensus is reached because many scientists have done a intense work analyzing data or models.  Those claiming to prove climate science is wrong on forum threads like this tend to ignore that work and point out that they can draw a better curve through the data or they don't like some feature of a data set or trend-line without actually providing any analysis. 

[JP]

jp says
...to promote your own model based on cherry-picked data and a best fit curve.
henry says
I told you from the beginning that my sample of 47 stations was random,
except for the fact of the choice of stations with complete or nearly complete records...
even choosing more stations won't change the result
if you get a correlation coefficient of 0.997 on the binomial for the drop of maximum temps.....

Yes, and you picked a small subset of the climate record that happens to have a straight line through it.  So what?  I can pick dozens.  Your method is unscientific because all you've done is make a model.  Now you have to extrapolate it and show that it fits more data.  It's fine if you don't understand the scientific method well enough to do that.  We all have to start somewhere, but at least stop telling other users on a science forum that you've done a proper scientific analysis that proves that climate scientists are wrong. 

Either way, even if you believe I am trolling, the question to you was about the error bars,
on the gistemp data set,
which you state were there
but they were not...

They're the green vertical bars on the data plot.  The web page also cites a peer-reviewed article about the methods used to calculate error bars.

[MoreCarbonOK]

JP says
Yes, and you picked a small subset of the climate record that happens to have a straight line through it.  So what?  I can pick dozens.
Henry asks
A binomial fit is a straight line?
Sorry, must I explain the difference to you between linear (straight) fits and polynomial fits?

[MoreCarbonOK]

JP says
They're the green vertical bars on the data plot.  The web page also cites a peer-reviewed article about the methods used to calculate error bars.
henry says
Sorry I missed that, but it seems to me the difference between red (ave of 4 major data sets, including gistemp)  and green (gistemp) on the graph I presented to you, WHICH IS the reality of WHAT WE HAVE, seems a lot more than the average [0.2] error indicated?

[JP]

Ah, yes, I'd forgotten you'd used a polynomial fit.  Even better!  I can find a well-fit polynomial to even more cherry picked chunks of climate data than a linear model since I have more degrees of freedom!  It just makes your argument that much weaker.

[Bored chemist]

The "fourth blanket" is an interesting analogy.
As you add more blankets, so the incremental effect of each becomes smaller.

Indeed,
Can you explain why you think it's negative (or, at least, not positive)?
BTW, I think I may have mentioned spectroscopic saturation earlier in the thread
(it's not really the same thing as the usual "diminishing returns" due to a reduced temperature difference)


Henry, perhaps it would help clarify matters if you were to tell us explicitly what your model is in the form
Temperature = (some mathematical function of) year.
Thanks

[MoreCarbonOK]

henry@bc&jp

I did in fact not use the binomial fit with r²=0.997  because I saw it would cause an ice age...soon.
I applied a sine wave!
do you BC, mean, that you want to know the exact mathematical formula for the sine wave
http://blogs.24.com/henryp/2012/10/02/best-sine-wave-fit-for-the-drop-in-global-maximum-temperatures/
that I did propose for the drop in global maxima for my data, evident from the results on the bottom of the first table
http://blogs.24.com/henryp/2013/02/21/henrys-pool-tables-on-global-warmingcooling/
?

[Bored chemist]

Probably just me being gormless but I'd find it much simpler if you gave the amplitude, frequency (or period) and an indication of the phase.
I think the period is 88 years,
The phase is defined by crossing zero at 17 years,
and the amplitude is 0.037
Am I reading it correctly?

[MoreCarbonOK]

Henry@bc
you got it figured right
1995 was zero as far as maxima was concerned.
Remember my data only goes up to 2012.
2012-17 equals 1995

[damocles]

From AlanCalverD (reply #140)

It's quite clear where the "skeptical" graph came from, but I'm interested to know what its authors actually plotted. Data that suggests that some winters are warmer than their adjacent summers deserves serious investigation.

The caption of the graph reveals all: what is actually being plotted is a "twelve month running average temperature". This is a convenient (and familiar to meteorologists) way of factoring out seasonal effects. What is plotted for each month is an average over the previous 12 months. Since this will always contain one of each month, it shows no seasonal variation, but is capable of resolving longer term variation (whether random or forced) to a monthly level.