73 replies to this topic

[Little Fish]

Doug1o29, on 03 July 2011 - 04:03 PM, said:

Here we have a graph that shows three consistent lines (four counting the black one) and one inconsistent one (in pink).  Which one is the most-likely to be wrong?

choosing the data to fit what you consider "likely" is the political method, not the scientific method.
you have started with your conclusion and chosen the 3 lines which fit your conclusion, the scientific method requires you daw your conclusion from all lines.

Quote

As I understand the problem, only the line on the graph was deleted.

"the line"??
the pink lines were deleted, the data between the two pink lines (dashed line) of that single series was kept.

Quote

The data remained in the model.  Conclusions are not based on graphs; they are based on math/stat.  The pink line is thus, irrelevant to the truth or lack of it.  Attack the model on which the graph is based:  attacking the graph isn't going to get you anything.

???
there was no "model".
the paper was a reconstruction of past temperatures from tree ring proxies (with instrumental measurements bolted on the end).

here is the graph that was published:


here is the paper:
http://www.meteo.psu...99-science.html

here is what the grpah looks like when we put the deleted data back:


It is incredible that anyone would defend data manipulation and deletion.

[Doug1o29]

Little Fish, on 04 July 2011 - 12:10 AM, said:

choosing the data to fit what you consider "likely" is the political method, not the scientific method.
you have started with your conclusion and chosen the 3 lines which fit your conclusion, the scientific method requires you daw your conclusion from all lines.

"the line"??
the pink lines were deleted, the data between the two pink lines (dashed line) of that single series was kept.

???
there was no "model".
the paper was a reconstruction of past temperatures from tree ring proxies (with instrumental measurements bolted on the end).

here is the graph that was published:

It is incredible that anyone would defend data manipulation and deletion.

From the paper you just referenced:

Quote

The confidence levels around the Mann et al. NH series are based on calibration against the instrumental record. However, additional uncertainty may come from the earlier sections of the tree-ring data, because tree-ring chronologies often exhibit a progressive degradation in statistical quality further back in time, a product of their diminishing internal replication (that is, series are often made up of fewer samples). Also, the production of a long tree-ring chronology normally involves some degree of detrending (known as "standardization") to reduce bias in the final chronology resulting from temporal changes in the average age of the samples (young trees have wider and more dense rings than older ones). As a result of standardization, many long tree-ring chronologies may not represent all of the long-term climate variability that influenced tree growth in their region.

The early portion of the chronology was deleted because it didn't meet the statistical standards.  It was based on only 12 series, while 20 is considered the minimum at present (This will probably go up as more series become available.  My current project uses 468 series; four were deleted because they could not be cross-dated.).

Mann's data did not extend past 1980, so there wasn't anything there to delete.  What was deleted was a reconstruction based on other people's data.

At present, the best available shortleaf chronology for Oklahoma was taken in 1982 by David Stahle and has not been updated since.  Seeing as my chronology has over 400 cores that cover the 1980 to 2007 interval and I can cross-date these with the Stahle data, it is not a problem to create a single chronology using both data sets.  But that uses multiple regression which Briffa doesn't like (A lot of biologists don't trust statistics.).

Altogether, I would see no problem in throwing out a reconstruction, replacing it with my own, which would then have to meet all the appropriate tests and pass muster with several statisticians right here in my office before being submitted to peer review.  Neither do I see a problem with throwing out a portion of a chronology that only has 12 replications, especially when the oldest data is derived from young trees (Detrending does not necessarily correct for all age-related variability.).

AND:  Briffa is wrong; juvenile wood is LESS DENSE than mature wood.
Doug

[Br Cornelius]

Quote

It is incredible that anyone would defend data manipulation and deletion.

What seems equally incredible is that you would presume to know better.

My physics lecturer, who did the bulk of the work on climate change, was a bit sniffy about dendrochronology (because physicists work in absolutes), but I think he was equally ignorant of the amount of theory which is involved in the field. We can all be experts when we know just enough to draw the wrong conclusions.

Br Cornelius

[Little Fish]

you don't delete data because of uncertainty. you show the data and state the uncertainty.
there is no explanation for deleting the data 1400-1600 and 1950-2000.

if the tree ring proxy does not measure the instrumental record from 1950-2000, then how can we know that the tree ring data measures past temperatures?

by deleting data that diverges they delete uncertainty.

[Doug1o29]

Little Fish, on 04 July 2011 - 10:36 AM, said:

you don't delete data because of uncertainty. you show the data and state the uncertainty.
there is no explanation for deleting the data 1400-1600 and 1950-2000.

Briffa said, right in the section I quoted, that the 1400-1600 data was deleted because of insufficient replication and because it was based on young trees.  Insufficient replication does two things:  first, it creates a large variance, which reduces, or even destroys the data's usefuleness.  Second, small sample sizes conform to the Student's-t distribution rather than a normal distribution, so ALL data from them is skewed.  While corrections are possible, it adds another level of complexity, and the results frequently aren't worth the effort.

Young trees produce highly-variable ring thicknesses because the detrending models are incapable of removing all the variation associated with age.

Every dendrochronologist on earth knows these problems and has to deal with them.  All one needs to do is state that these data were not used and give the reason.  There is nothing suspicious about it.  If somebody thinks they should have been used, they can usually obtain the data and re-run the test themselves.  Why don't you do that instead of whining about it?

It's only the amateurs who see a conspiracy in this.

Quote

if the tree ring proxy does not measure the instrumental record from 1950-2000, then how can we know that the tree ring data measures past temperatures?

According to Briffa, it was 1980 to 2000 and it was not data that was deleted, but a reconstructed chronology based on previous studies.

No data after 1980 was deleted BECAUSE THERE WASN'T ANY.  By the time a set of cores can be collected, processed and the paper written up, the data are often four or five years old.  And lots of people keep old core collections around for decades to verify previous work and to use in future studies.  Those collections do not have the more-recent years - cores don't keep growing after they're removed from the tree.

There is a move afoot to set up an archive where core collections can be preserved and made available for study.  Such a thing is badly needed.
Doug

[Little Fish]

Doug1o29, on 04 July 2011 - 02:42 PM, said:

No data after 1980 was deleted BECAUSE THERE WASN'T ANY.

the data has been released in the public domain.
it was deleted from the graph from 1950-1999.
where do you think the pink line after 1950 came from in the chart I showed?

[Doug1o29]

Little Fish, on 04 July 2011 - 05:37 PM, said:

the data has been released in the public domain.
it was deleted from the graph from 1950-1999.
where do you think the pink line after 1950 came from in the chart I showed?

The graph has been processed.  It's a reconstruction.  Real tree ring data jumps all over the place.  If there were 12 series, there'd be twelve separate lines on that graph.  I don't know whether the processing was a regression, a smoothing, a running average or something else, but the curve depicted is far too smooth to be made directly from unprocessed data.

And you just said that the data is in the public domain, so download that data, do your own analysis, post the results here and let's see what it says.  I'm calling your bluff.  What do YOU think the data says?
Doug

Edited by Doug1o29, 04 July 2011 - 07:02 PM.

[Little Fish]

Doug1o29, on 04 July 2011 - 06:55 PM, said:

The graph has been processed.  It's a reconstruction.  Real tree ring data jumps all over the place.  If there were 12 series, there'd be twelve separate lines on that graph.  I don't know whether the processing was a regression, a smoothing, a running average or something else, but the curve depicted is far too smooth to be made directly from unprocessed data.

The issue is not whether the data has been processed or not.

The issue is that the data was deleted from 1950-1999.

did you watch Professor Muller explain this to you?
did you hear him say that the data has been released?
did you hear him say that he would never read or believe anything else written by those authors?

[Br Cornelius]

Little Fish, on 04 July 2011 - 09:26 PM, said:

The issue is not whether the data has been processed or not.

The issue is that the data was deleted from 1950-1999.

did you watch Professor Muller explain this to you?
did you hear him say that the data has been released?
did you hear him say that he would never read or believe anything else written by those authors?

So Dougs explanation of how his field of expertise works means nothing to you.
You'd rather listen to those who speak words you want to hear - now there's a surprise

Br Cornelius

[Professor Buzzkill]

Br Cornelius, on 04 July 2011 - 09:38 PM, said:

So Dougs explanation of how his field of expertise works means nothing to you.
You'd rather listen to those who speak words you want to hear - now there's a surprise

Br Cornelius

Are you any different?

[Br Cornelius]

Professor Buzzkill, on 04 July 2011 - 09:55 PM, said:

Are you any different?

I like to check the credentials of those who are feeding me a line. Certain things get my alarm bells ringing.
Niffty little YouTube Vids is one of them.

My faith in Little Fishes Judgement in science took a huge blow when he spent over ten pages defending the struck off Quack - Doctor Wakefield. On that one he showed an amazing lack of understanding of what basic medical scientific research methods entail and the responsibility of a Doctor to protect the public good until a compelling case can be made for disclosure of controversial findings. In that discussion he also went up against an expert in the field of medical research convinced that he knew better.

Br Cornelius

Edited by Br Cornelius, 04 July 2011 - 10:09 PM.

[Little Fish]

Br Cornelius, on 04 July 2011 - 09:38 PM, said:

So Dougs explanation of how his field of expertise works means nothing to you.
You'd rather listen to those who speak words you want to hear - now there's a surprise

Br Cornelius

it's a simple issue, the data was deleted from 1950-1999.
Doug avoided the issue by windbagging about processing ambiguities, yet you think he gave an explanation. what does that say about your understanding of this issue?

[Br Cornelius]

Little Fish, on 04 July 2011 - 10:30 PM, said:

it's a simple issue, the data was deleted from 1950-1999.
Doug avoided the issue by windbagging about processing ambiguities, yet you think he gave an explanation. what does that say about your understanding of this issue?

It shows I listen to experts.
Not all data is equal and the secret of good science is spotting good data from bad. If young tree's represent bad data then don't use them. If there is a known problem with a data series then don't use the data series or compensate in a way recognised as acceptable within the field. That is what Doug said - not windbagging.

Here is a nice little summery of some of the issues with the Michael Mann graph - but your complaint is not one of them;

Quote

It looks like you are confusing the problems with the Sheep Mountain series and the Yamil series. Michael Smith was referring to the Yamil series when he complained about the presence of a recent temperature rise and the lack of a temperature rise in the local temperature. Phil Jones explained the reason for this in an interview with the BBC: http://news.bbc.co.u...i/science/natur…
80% down the interview under the question:
'In the e-mails you refer to a "trick" which your critics say suggests you conspired to trick the public? You also mentioned "hiding the decline" (in temperatures). Why did you say these things?'
Phil Jones responded:
'This "divergence" is well known in the tree-ring literature and "trick" did not refer to any intention to deceive - but rather "a convenient way of achieving something", in this case joining the earlier valid part of the tree-ring record with the recent, more reliable instrumental record.'

In other words, they hid the fact that the tree ring data did not match global temperature readings, and were dumped in favor of the global temperature readings for the 2001 IPCC TAR.

At least unitl 2007 when they "came clean":
"an issue which was later directly discussed in the 2007 IPCC AR4 Report." - Phil Jones in above link

As for the Sheep Mountain trees, they did show a hockey stick, but local temperature readings did not. For that reason, those trees were not considered to be reliable for temperature readings by the paleoclimatologists who created the data set. http://europa.agu.or...i=/journals/gb/…
These were heavily weighted in M Mann's 2001 hockey stick. This cherry picking example was the result of a fortran program that M Mann used that created hockey sticks more than 99 times out of 100 when given red noise (random data sets) by selectively picking and weighting the data that conformed to a hockey stick shape. However, when McIntyre ran M Mann's cherry picking program without the Sheep Mountain trees, the hockey stick became a broom handle. The cherry picking charge is the result of a graph in M Mann's data as sent to McIntyre labeled "censored" that exactly matched the broom handle graph created by removing the Sheep Mountain trees. http://www.geo.utexa.../courses/387h/P…

The elimination of the beginning of Gaspé cedar ring width series (1400 CE to 1450 CE)
http://www.springerl...om/content/g0n4…
from M Mann's data set deleted the hockey stick by showing the Little Ice Age. Note, that the authors of the series did not consider the data to be reliable until 1600, because paleoclimatologists do not consider series with less then 4 trees significant. Previous to 1600 CE, the data is not contiguous, but the first ring dates to 1404 CE. M Mann used the data to run back to 1400 CE by applying extrapolation techniques to cover the missing data. The data removed to delete the hockey stick shape consisted of 0 trees (1400 - 1404 CE), 1 tree (1404 - 1427 CE) and 2 trees (1427 - 1450 CE). http://multi-science...press.com/conte…

The complaints of cherry picking addressed in your deepclimate.org link were irrelevant straw man attacks, and never addressed the issues involved with tree ring data, or the concerns that skeptics have with the hockey stick.

The trees do respond to temperature, but they also respond to other things as well:
- Atmospheric CO2
- Precipitation
- Sunlight
- root crowding
- fire
- disease
- soil fungi
- volcanic ash
- accidents
They also tend to respond to temperature extremes more than average temperatures.

Tree ring data is considered to be second class by many paleoclimatologists, and many prefer other means to analyze temperature. Here is a paper that incorporates all known studies at the time EXCEPT tree ring data:
(Full) http://multi-science...press.com/conte…
(Summary) http://www.co2scienc.../articles/V11/N…
Source(s):
Fraud Investigation: http://voices.washin...ost.com/virgini…
Hockey Stick Publisher: http://www.nature.co...ure/journal/v46…
Fraud Whitewash: http://www.research....du/orp/Findings…
National Academy of Sci: http://www.guardian..../environment/20…
NAS hockey stick: http://www.nap.edu/c...g.php?record_id…
IPCC review process: http://multi-science...press.com/conte…
Medieval Warm Period: http://www.co2scienc.../data/mwp/quant…

The other papers should show why the Mann graph is considered acceptable, if not the best data set.


Br Cornelius

Edited by Br Cornelius, 04 July 2011 - 10:45 PM.

[Little Fish]

Br Cornelius, on 04 July 2011 - 10:36 PM, said:

Here is a nice little summery of some of the issues with the Michael Mann graph

but your complaint is not one of them

Because I'm talking about the Briffa-Osborne paper not the Michael Mann graph.

and none of your links work.

[Little Fish]

"When Lindzen was informed during the interview that the first three allegations had already been dismissed at the inquiry stage, his response, as quoted in the Committee’s report, was: “It’s thoroughly amazing. I mean these are issues that he explicitly stated in the emails. I’m wondering what is going on?”

Dr. Lindzen’s bewilderment is understandable.  Concerning the Committee’s conclusion regarding the first allegation (suppressing or falsifying data) — characterizing the “trick” to “hide the decline” as legitimate application of a conventional statistical methodology, ignored or misconstrued salient facts. While Mann’s own research methodology and results have indeed been challenged as fatally flawed, the actual trick should be examined within a broader context.

First, there is a widespread misconception that the reference to a decline refers to concealing an observed fall in global temperatures since a peak in 1998, the warmest year for some time. Instead, it really has to do with graphic trickery suggesting that man-made CO2 emissions over the past 40 years have produced a nearly vertical temperature escalation.

A 1,000-year-long graph was cobbled together using various proxy data derived from ice cores, tree rings and written records of growing season dates up until 1961, where it then applied surface ground station temperature data. Why change in 1961? Well that’s when tree ring proxy data calculations by CRU’s Keith Briffa began going the other way in a steady decline. After presenting these unwelcome results to Mann and others, he was put under pressure to recalculate them. Briffa did, and the decline became even greater.

This presented what Mann referred to as a “conundrum.” Emails reveal that the late 20th century decline indicated by Briffa would be perceived by IPCC as “diluting the message”, was a “problem”, and posed a “potential distraction/detraction.” Mann went on to say that the warming skeptics would have a “field day” if Briffa’s declining temperature reconstruction was shown, and that he would “hate to be the one” to give them “fodder.” So one aspect of Mike’s “trick” was reportedly to show all of the proxy and surface measurement chartings in different colors on a single graph, but simply cut off Briffa’s in a spaghetti clutter of lines at the 1961 date."

http://blogs.forbes....ewash-part-one/

Quote

Not all data is equal and the secret of good science is spotting good data from bad

read the above then, what is it that you "spot"??

Edited by Little Fish, 04 July 2011 - 11:50 PM.