Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Mann et al: Optimal surface temperature reconstructions using terrestrial borehole data

3 views
Skip to first unread message

w...@bas.ac.uk

unread,
Jun 30, 2003, 10:31:36 AM6/30/03
to
Warning: read this carefully.

Optimal surface temperature reconstructions using terrestrial borehole data
Mann ME, Rutherford S, Bradley RS, Hughes MK, Keimig FT
JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES 108 (D7): art. no. 4203 APR 3 2003

Abstract:
[1] We derive an optimal Northern Hemisphere mean surface temperature
reconstruction from terrestrial borehole temperature profiles spanning
the past five centuries. The pattern of borehole
ground surface temperature (GST) reconstructions displays prominent
discrepancies with instrumental surface air temperature (SAT) estimates
during the 20th century, suggesting the presence of
a considerable amount of noise and/or bias in any underlying spatial SAT
signal. The vast majority of variance in the borehole dataset is efficiently
retained by its two leading eigenvectors. A sizable
share of the variance in the first eigenvector appears to be associated
with non-SAT related bias in the borehole data. A weak but detectable
SAT signal appears to be described by a combination
of the first two eigenvectors. Exploiting this eigendecomposition,
application of optimal signal estimation methods yields a hemispheric
borehole SAT reconstruction that is largely consistent with
instrumental data available in past centuries, and is indistinguishable
in it smajor features from several published long-term temperature
estimates based on both climate proxy data and model
simulations.

-W.

--
William M Connolley | w...@bas.ac.uk | http://www.antarctica.ac.uk/met/wmc/
Climate Modeller, British Antarctic Survey | Disclaimer: I speak for myself
I'm a .signature virus! copy me into your .signature file & help me spread!

Alastair McDonald

unread,
Jun 30, 2003, 1:05:20 PM6/30/03
to
<w...@bas.ac.uk> wrote in message news:3f00...@news.nwl.ac.uk...
> Warning: read this carefully.

Well I have read it carefully and it seems to say that, although the data from
boreholes does not tie in with those from modern scientific instruments, by
applying sofisticated mathematical techniques it is possible to make it fit
with the results of other proxies (for instance those published by
M.E. Mann, thus proving him right!)

Of course I am just a cynical old b***** who don't even trust the IPCC TAR.

Cheers, Alastair.

w...@bas.ac.uk

unread,
Jun 30, 2003, 3:58:13 PM6/30/03
to
Alastair McDonald <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote:
><w...@bas.ac.uk> wrote in message news:3f00...@news.nwl.ac.uk...

>Well I have read it carefully and it seems to say that, although the data from


>boreholes does not tie in with those from modern scientific instruments, by
>applying sofisticated mathematical techniques it is possible to make it fit
>with the results of other proxies (for instance those published by
>M.E. Mann, thus proving him right!)

The vital bit is:

>> A sizable
>> share of the variance in the first eigenvector appears to be associated
>> with non-SAT related bias in the borehole data.

-W.

Roger Coppock

unread,
Jun 30, 2003, 7:18:37 PM6/30/03
to
w...@bas.ac.uk wrote:
>
> The vital bit is:
>
> >> A sizable
> >> share of the variance in the first eigenvector appears to be associated
> >> with non-SAT related bias in the borehole data.
>
> -W.
>

And you say this factor analytic result is vital because . . .

--

"One who joyfully guards his mind
And fears his own confusion
Can not fall.
He has found his way to peace."

-- Buddha, in the "Pali Dhammapada,"
~5th century BCE


-.-. --.- Roger Coppock (rcop...@adnc.com)


-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 80,000 Newsgroups - 16 Different Servers! =-----

w...@bas.ac.uk

unread,
Jul 1, 2003, 5:27:19 AM7/1/03
to
Roger Coppock <rcop...@adnc.com> wrote:

>w...@bas.ac.uk wrote:
>>
>> >> A sizable
>> >> share of the variance in the first eigenvector appears to be associated
>> >> with non-SAT related bias in the borehole data.

>And you say this factor analytic result is vital because . . .

Oh go on then, I'll say it. This is from memories of a talk at EGS this april.
There is a "problem" in that borehole thermometry shows greater warming than
various other records (includuing Manns). Various people argue that this is
because the soil-sfc and sub-soil-sfc T's are not the same as the sfc-air-t
measured by, eg, stevenson screens. Notably because of lying snow, if I recall
correctly. Mann is saying (I think) that the "true" signal is nonetheless
recognisable in the borehole record and can be extracted, if you're careful.

See the bottom and somewhat up on:

http://www.people.virginia.edu/~mem6u/preprints.html

for preprints (but they are in Evil pdf).

-W.

>> >> Optimal surface temperature reconstructions using terrestrial borehole data
>> >> Mann ME, Rutherford S, Bradley RS, Hughes MK, Keimig FT
>> >> JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES 108 (D7): art. no. 4203 APR 3
>> >2003
>> >>
>> >> Abstract:
>> >> [1] We derive an optimal Northern Hemisphere mean surface temperature
>> >> reconstruction from terrestrial borehole temperature profiles spanning
>> >> the past five centuries. The pattern of borehole
>> >> ground surface temperature (GST) reconstructions displays prominent
>> >> discrepancies with instrumental surface air temperature (SAT) estimates
>> >> during the 20th century, suggesting the presence of
>> >> a considerable amount of noise and/or bias in any underlying spatial SAT
>> >> signal. The vast majority of variance in the borehole dataset is efficiently
>> >> retained by its two leading eigenvectors. A sizable
>> >> share of the variance in the first eigenvector appears to be associated
>> >> with non-SAT related bias in the borehole data. A weak but detectable
>> >> SAT signal appears to be described by a combination
>> >> of the first two eigenvectors. Exploiting this eigendecomposition,
>> >> application of optimal signal estimation methods yields a hemispheric
>> >> borehole SAT reconstruction that is largely consistent with
>> >> instrumental data available in past centuries, and is indistinguishable
>> >> in it smajor features from several published long-term temperature
>> >> estimates based on both climate proxy data and model
>> >> simulations.

--

Alastair McDonald

unread,
Jul 1, 2003, 7:59:14 AM7/1/03
to
Perhaps I should admit that I do not know what "the variance in the
first eigenvector" means! I would like to claim that I have no problems
with second eigenvectors, but I can't :-(
OTOH, ignorance makes it easier to speculate, so here goes.

Mann's 'Hockey Stick' paper seems to have been crucial in persuading the
IPCC that global warming was anthropogenic. As I understood it, the
reasoning was that temperatures are climbing faster than at any time in
the last 1000 years, so the rate of increase cannot be natural. IMHO the
error bars on the these results were so large that such a claim was
unjustifed, though it appeared valid because the data had been averaged
and smoothed. This would also apply to any claim that temperatures today
are higher than at any time in the last 1000 years.

Mann in fact writes in "Lessons for a New Millennium" Michael E. Mann
Science 2000 July 14; 289: 253-254. (in Perspectives) that "(ii) the dramatic
warming of the 20th century can almost certainly not be explained by the
natural forcings, but instead requires the emergent anthropogenic forcings of
the 20th century," which in my mind leaves open whether it is the warming,
or rate of warming to which he is referring.

Soon et al. have just published a paper claiming that during the Medieval
Optimum, temperatures were higher than now. If borehole data is also
showing temperatures were higher in the past, then it is probably true.
However that does not mean that the warming during the last century
was not mainly caused by anthropogenic greenhouse gasses, just that
Mann's argument is invalid.

HTH,

Cheers, Alastair.

<w...@bas.ac.uk> wrote in message news:3f01...@news.nwl.ac.uk...

Ian St. John

unread,
Jul 1, 2003, 2:05:30 PM7/1/03
to

"Alastair McDonald" <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote
in message news:bdrsoo$4di$1...@news8.svr.pol.co.uk...

> Perhaps I should admit that I do not know what "the variance in the
> first eigenvector" means! I would like to claim that I have no problems
> with second eigenvectors, but I can't :-(
> OTOH, ignorance makes it easier to speculate, so here goes.

It also makes it easier to be totally wrong.

>
> Mann's 'Hockey Stick' paper seems to have been crucial in persuading the
> IPCC that global warming was anthropogenic. As I understood it, the
> reasoning was that temperatures are climbing faster than at any time in
> the last 1000 years, so the rate of increase cannot be natural. IMHO the
> error bars on the these results were so large that such a claim was
> unjustifed, though it appeared valid because the data had been averaged
> and smoothed. This would also apply to any claim that temperatures today
> are higher than at any time in the last 1000 years.

When dealing with noisy data, a filtering operation is perfectly valid. It
is based on the concept that the signal CANNOT change at the rate you are
measuring so the variance is not part of the signal. Expecially true if the
noise is random and includes increases and drop and increase and drop, etc.
Unless god is playing dice with the climate, the noise is not meaningful.
Filtering of noisy data is totally based on scientific principles.

>
> Mann in fact writes in "Lessons for a New Millennium" Michael E. Mann
> Science 2000 July 14; 289: 253-254. (in Perspectives) that "(ii) the
dramatic
> warming of the 20th century can almost certainly not be explained by the
> natural forcings, but instead requires the emergent anthropogenic forcings
of
> the 20th century," which in my mind leaves open whether it is the warming,
> or rate of warming to which he is referring.

The rate of warming ( dramatic warming ). This just need reading
comprehension. To make something go FAST you need a big force! This is easy
to understand. You don't get race cars with go cart engines..

>
> Soon et al. have just published a paper claiming

Claiming is the operative word. And published in an obscure magazine where
the peer review will me minimal. The field of dendrochronogy is sufficiently
specialised that the journal would not have the means to note how poor the
article was.

> that during the Medieval
> Optimum, temperatures were higher than now. If borehole data is also
> showing temperatures were higher in the past, then it is probably true.

The Soon article was not even a study but an 'opinion' based on cherry
picking data from actual science studies. It was little more than an op-ed.
It was bought and paid for by the Fossil fuel companies and showed how
little respect Soon, Baliunas, Idsos, etc have for peer reveiw and
scientific honesty. They are almost bragging about their abiltiy to publish
total crap.

> However that does not mean that the warming during the last century
> was not mainly caused by anthropogenic greenhouse gasses, just that
> Mann's argument is invalid.

It isn't Manns argument. It is a study of the Northern Hemisphere
temperature reconstruction. The IPCC established that the warming was both
unprecedented, and driven by Solar, Aerosols, and GHGs of which GHGs are
currently the primary and almost exclusive factor since the aerosol cooling
counters the solar warming almost exactly.

Thomas Palm

unread,
Jul 1, 2003, 2:43:50 PM7/1/03
to
Alastair McDonald wrote:
> Perhaps I should admit that I do not know what "the variance in the
> first eigenvector" means! I would like to claim that I have no problems
> with second eigenvectors, but I can't :-(
> OTOH, ignorance makes it easier to speculate, so here goes.

Easier certainly, but the speculations are correspondingly less
interesting.

> Mann's 'Hockey Stick' paper seems to have been crucial in persuading the
> IPCC that global warming was anthropogenic.

Not really. Scientists have lots of arguments for why the current warming
is anthropogenic. The curve by Mann and similar studies are more important
to convince the public since the result is easy to understand. This may
be why propagandists are so eager to discredit it in op ed pieces, but I
think the result is that it has just become more known. "Big bang" was
originally used as a derogatory term by Hoyle who didn't believe in it,
but the term stuck. I suspect the same will go for the "hockey stick".
It is a good description.

> As I understood it, the
> reasoning was that temperatures are climbing faster than at any time in
> the last 1000 years, so the rate of increase cannot be natural. IMHO the
> error bars on the these results were so large that such a claim was
> unjustifed, though it appeared valid because the data had been averaged
> and smoothed. This would also apply to any claim that temperatures today
> are higher than at any time in the last 1000 years.

If you go inte the ends of the error bars there is a possibility that
temperatures some time during those 1000 years approached the current,
but it is not likely. Even in this case the current climate change
would still be highly unusual if not unique.



> Mann in fact writes in "Lessons for a New Millennium" Michael E. Mann
> Science 2000 July 14; 289: 253-254. (in Perspectives) that "(ii) the dramatic
> warming of the 20th century can almost certainly not be explained by the
> natural forcings, but instead requires the emergent anthropogenic forcings of
> the 20th century," which in my mind leaves open whether it is the warming,
> or rate of warming to which he is referring.

The rate of warming is the obvious sign. Had our both our emissions
and the rate of warming been an order of magnitude lower it would be
a lot more difficult to separate it from imperfectly understood
natural changes.



> Soon et al. have just published a paper claiming that during the Medieval
> Optimum, temperatures were higher than now.

Actually I don't think the paper says that even if the press release
does. The article is some kind of meta study that just try to analyze
a selection of earlier studies and concludes that at some point some
areas may have been warmer. (I haven't read it, only what has been
written about it on the group.)

> If borehole data is also
> showing temperatures were higher in the past, then it is probably true.

Borehole data doesn't show that, at the contrary they show a strong
warming. It just doesn't look the same as the surface data:
http://www.ngdc.noaa.gov/paleo/globalwarming/huang.html

> However that does not mean that the warming during the last century
> was not mainly caused by anthropogenic greenhouse gasses, just that
> Mann's argument is invalid.

So far no one has given a credible argument refuting Mann's work
or produced anything better. There are alternative studies and
the pros are trying to remove discrepances between them. This is
still a new field.

w...@bas.ac.uk

unread,
Jul 1, 2003, 3:00:39 PM7/1/03
to
Alastair McDonald <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote:
>Perhaps I should admit that I do not know what "the variance in the
>first eigenvector" means!

The various eigenvectors are orthogonal. So the total variance of
the series can be decomposed into the variance associated with the
various eigenvectors. If the first few eignevectors account for most
of the variance, then the series "really" has few degrees of freedom.

>Mann's 'Hockey Stick' paper seems to have been crucial in persuading the

Um. We don't say HS in polite company.

>IPCC that global warming was anthropogenic. As I understood it, the
>reasoning was that temperatures are climbing faster than at any time in
>the last 1000 years, so the rate of increase cannot be natural. IMHO the
>error bars on the these results were so large that such a claim was
>unjustifed,

I don't understand this. The error bars are such that the 20th *is*
warmer in Manns series than the past, even if you assume the past sits
at the top of its error bars.

>though it appeared valid because the data had been averaged
>and smoothed.

Ditto Ians comments: whats wrong with smoothing?

>Soon et al. have just published a paper claiming that during the Medieval
>Optimum, temperatures were higher than now.

My understanding is that this is not true: Soon et al assert that, in
some (varying) regions, T in the past has been higher than now. That,
even if true, is compatible with global T being highest now. Since I
haven't read the paper (have you?) I'm unkeen to comment. But as ISJ
said, its in an odd journal.

> If borehole data is also
>showing temperatures were higher in the past,

No. Quite the reverse. The puzzle with the boreholes is that they indicate
*more* warming.

> then it is probably true.

Ah... so the boreholes are the other way round, does that mean you
think Soon et al is false?

-W.

Alastair McDonald

unread,
Jul 1, 2003, 6:08:52 PM7/1/03
to
<w...@bas.ac.uk> wrote in message news:3f01...@news.nwl.ac.uk...
> Alastair McDonald <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote:
> >Perhaps I should admit that I do not know what "the variance in the
> >first eigenvector" means!
>
> The various eigenvectors are orthogonal. So the total variance of
> the series can be decomposed into the variance associated with the
> various eigenvectors. If the first few eignevectors account for most
> of the variance, then the series "really" has few degrees of freedom.
>
> >Mann's 'Hockey Stick' paper seems to have been crucial in persuading the
>
> Um. We don't say HS in polite company.

This is from Scientific Amercan, and says rougly the same as me ...
ftp://holocene.evsc.virginia.edu/pub/mann/sa50-reprint.pdf

WHILE JUST A POSTDOC at the University of Massachusetts at
Amherst in the late 1990s, Michael Mann made a splash with his
studies of climate variability over the past 1,000 years. Using tree-ring
and ice-core measurements to estimate global temperatures, he and
his colleagues Raymond Bradley and Malcolm Hughes devised a
graph that was dubbed "the hockey stick" because it showed a sharp
upward swing in temperatures after 1850. The research helped to
prompt a shift in the position of the United Nations International
Panel on Climate Change, which reported last year that humans
were responsible for most of the global warming over the past
century. Mann was also the lead author of the panel's chapter on
climate variability. He has subsequently buttressed his findings with
more studies.Mann's work figured prominently in the Environmental
Protection Agency's statement earlier this year that humans are
gradually heating the planet, and although the Bush administration
quickly repudiated that report, the evidence is becoming harder
and harder to deny.

... including the use of HS!

> >IPCC that global warming was anthropogenic. As I understood it, the
> >reasoning was that temperatures are climbing faster than at any time in
> >the last 1000 years, so the rate of increase cannot be natural. IMHO the
> >error bars on the these results were so large that such a claim was
> >unjustifed,
>
> I don't understand this. The error bars are such that the 20th *is*
> warmer in Manns series than the past, even if you assume the past sits
> at the top of its error bars.

I was looking at this chart;
http://www.sciencemag.org/cgi/content/full/289/5477/253/F1
from this Science perspective on a Crowley paper
http://www.sciencemag.org/cgi/content/full/289/5477/253

> >though it appeared valid because the data had been averaged
> >and smoothed.
>
> Ditto Ians comments: whats wrong with smoothing?

Smoothing, and taking means, removes or discards information. IMHO
it is a dangerous action and should only be done if the original data is
also presented. Even when this is done, clowns like JD (another banned
expression) will only look at the smoothed data and label it HS. Ian
says that it is removing random effects, but in fact the weather and
climate jump about. These variations are real facts not random
errors. God does play dice with the weather. It is chaotic

> >Soon et al. have just published a paper claiming that during the Medieval
> >Optimum, temperatures were higher than now.
>
> My understanding is that this is not true: Soon et al assert that, in
> some (varying) regions, T in the past has been higher than now. That,
> even if true, is compatible with global T being highest now. Since I
> haven't read the paper (have you?) I'm unkeen to comment. But as ISJ
> said, its in an odd journal.

I have tracked down the article and it is 64 pages long. I do not propose
to print it far less read it, but from the parts I have read (abstract and
conclusion) you are right. BTW The journal is edited by Sonja
Boehmer-Christiansen of Hull University. In her autobiographical details
at her web site
http://www.hull.ac.uk/geog/html/boehmer.html
she says "Environment becomes fashionable and I turn sceptic". She
obviously revels in being a rebel! Some of the other article in that issue
were;

Climate Change - A Natural Hazard
Kininmonth, William

Reconstructing Climatic and Environmental Changes of the Past 1000 Years:
A Reappraisal
Soon, Willie - Baliunas, Sallie - Idso, Craig - Idso, Sherwood - Legates,
David R.

"Global Warming": Myth or Reality? The Actual Evolution of the Weather
Dynamics
Leroux, Marcel

Do Facts Matter Anymore?
Michaels, Pat

New Little Ice Age Instead of Global Warming? [**!**]
Landscheidt, Theodor

The "Greenhouse Effect" as a Function of Atmospheric Mass
Jelbring, Hans

> > If borehole data is also
> >showing temperatures were higher in the past,
>
> No. Quite the reverse. The puzzle with the boreholes is that they indicate
> *more* warming.

That's what I said! I meant "borehole data is also showing higher
tempertures in the past".

> Ah... so the boreholes are the other way round, does that mean you
> think Soon et al is false?

I think Soon's conclusions as they are described are false whether
their data agrees or disagrees with the borehole data.

Since I seem to have also covered the points made by Ian and Thomas,
I'll sign off now.

Cheers, Alastair.

w...@bas.ac.uk

unread,
Jul 1, 2003, 6:29:55 PM7/1/03
to
Alastair McDonald <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote:
>> I don't understand this. The error bars are such that the 20th *is*
>> warmer in Manns series than the past, even if you assume the past sits
>> at the top of its error bars.

Thats non-public. Try:

http://www.grida.no/climate/ipcc_tar/wg1/figspm-1.htm

>> >though it appeared valid because the data had been averaged
>> >and smoothed.
>>
>> Ditto Ians comments: whats wrong with smoothing?

>Smoothing, and taking means, removes or discards information. IMHO
>it is a dangerous action and should only be done if the original data is
>also presented.

I disagree. Reducing the complex global temperature record to a single
number per year also discards information but I don't see you complaining
about that. We're smoothing in time to one value per year: is that OK?

>New Little Ice Age Instead of Global Warming? [**!**]
>Landscheidt, Theodor

>The "Greenhouse Effect" as a Function of Atmospheric Mass
>Jelbring, Hans

Yes indeed. High quality stuff.

>> > If borehole data is also
>> >showing temperatures were higher in the past,
>>
>> No. Quite the reverse. The puzzle with the boreholes is that they indicate
>> *more* warming.

>That's what I said! I meant "borehole data is also showing higher
>tempertures in the past".

No, come on, pause a moment. The boreholes indicate more warming (in
the usual sense). Which means they show it to be *cooler* in the past
than Manns record (or others).

Ian St. John

unread,
Jul 1, 2003, 6:39:21 PM7/1/03
to

"Alastair McDonald" <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote
in message news:bdt0c5$g4m$1...@newsg1.svr.pol.co.uk...

> <w...@bas.ac.uk> wrote in message news:3f01...@news.nwl.ac.uk...
<snip>

> > >though it appeared valid because the data had been averaged
> > >and smoothed.
> >
> > Ditto Ians comments: whats wrong with smoothing?
>
> Smoothing, and taking means, removes or discards information.

No. It removes noise and leaves the signal. Not all data is information.

> IMHO it is a dangerous action and should only be done if the original data
is
> also presented.

Mann, et al 1999 did take that precaution against criticm, if you will
notice.

> Even when this is done, clowns like JD (another banned
> expression) will only look at the smoothed data and label it HS.

Labelling is of little importance. The results of the study are what are
important.

> Ian says that it is removing random effects, but in fact the weather and
> climate jump about.

First, the paper is on climate which has to have a twenty or thirty year
smoothing. Nothing based on year to year variations is meaningful and even a
decadal smoothing is dominated by the variations in climate oscillations.
Alistaire. Do we have to patiently explain the difference between weather
and climate to you again???


> These variations are real facts not random errors.

No. They are mostly based on errors in the dendrochronology due to local
microclimate. You are averaging a lot of individual indicators to get a
noisy signal which is then perfectly valid to smooth to remove noise. The
charactersistic of the noise is random, not systematic. If it were a
variation due to climate oscillations, for example, you would see that in
the data. Instead you see a relatively year to year variance much in excess
of any real climate drift and unsystematic. The filtering seems perfectly
appropriate for the signal selected.

> God does play dice with the weather. It is chaotic

Sure is, on a day to day level, but we are doing climate studies with
dendrochronology. That is already 'averaged' for year to year variations at
a minimum and weather is meaningless in terms of the study.


Alastair McDonald

unread,
Jul 1, 2003, 8:51:06 PM7/1/03
to
"Ian St. John" <ist...@spamcop.net> wrote in message
news:u4oMa.3750$eF3.4...@news20.bellglobal.com...

If you want to argue at that level you may find you are biting off more than
you can chew.

There is more information in tree rings than just the annual temperature,
but even it that was the only information, smoothing that data would
remove any dramatic (sic) changes in temperature..

> > These variations are real facts not random errors.
>
> No. They are mostly based on errors in the dendrochronology due to local
> microclimate. You are averaging a lot of individual indicators to get a
> noisy signal which is then perfectly valid to smooth to remove noise. The
> charactersistic of the noise is random, not systematic. If it were a
> variation due to climate oscillations, for example, you would see that in
> the data. Instead you see a relatively year to year variance much in excess
> of any real climate drift and unsystematic. The filtering seems perfectly
> appropriate for the signal selected.

The annual tree rings are significant. For instance they can be used to
find the year Santorini erupted. See Exodus to Arthur by Prof. Mike
Baillie.

> > God does play dice with the weather. It is chaotic
>
> Sure is, on a day to day level, but we are doing climate studies with
> dendrochronology. That is already 'averaged' for year to year variations at
> a minimum and weather is meaningless in terms of the study.

Oh dear, I think I meant climate not weather.

The age of wood can be found by matching the pattern of its rings. This
can now be done going back over 5000 years. These patterns are
chaotic but they are not random. They depend on the weather/climate
of the year the ring was created.

W. asked me what I had against smoothing. I am telling you. If you
don't agree that is your priveledge. It is a free world. But you
seem to be very close to making the error in believing that tree
rings are random, whereas infact each one tells a story.

Climate is not just the annual average temperature and precipitation.
It is also the months of maximum and minimum tempertures, the
frequency and season of hurricanes, and duration and timing of snow
cover and permafrost. If you average out the hurricanes over
five years, you lose the months when they occur and the fact that
they decrease during El Nino years.

HTH,

Cheers, Alastair.

Alastair McDonald

unread,
Jul 1, 2003, 9:02:20 PM7/1/03
to
<w...@bas.ac.uk> wrote in message news:3f02...@news.nwl.ac.uk...

> Alastair McDonald <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote:
>
> >> > If borehole data is also
> >> >showing temperatures were higher in the past,
> >>
> >> No. Quite the reverse. The puzzle with the boreholes is that they
indicate
> >> *more* warming.
>
> >That's what I said! I meant "borehole data is also showing higher
> >tempertures in the past".
>
> No, come on, pause a moment. The boreholes indicate more warming (in
> the usual sense). Which means they show it to be *cooler* in the past
> than Manns record (or others).

I still don't get it. If you go down the borehole and find perma frost,
then long ago it was colder. If it is hotter deep down, then previously
the climate must have been warmer.

Or am I still missing something?

Cheers, Alastair.


Ian St. John

unread,
Jul 1, 2003, 10:06:01 PM7/1/03
to

"Alastair McDonald" <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote
in message news:bdt9t9$ihp$1...@news6.svr.pol.co.uk...

> "Ian St. John" <ist...@spamcop.net> wrote in message
> news:u4oMa.3750$eF3.4...@news20.bellglobal.com...
<snip>

>
> Oh dear, I think I meant climate not weather.
>
> The age of wood can be found by matching the pattern of its rings. This
> can now be done going back over 5000 years. These patterns are
> chaotic but they are not random. They depend on the weather/climate
> of the year the ring was created.
True. Now, take the next point. Are the rings in a warm year in an oak tree
the same width as that of a poplar? Or a Maple? How about a Sequoia?

First you have to 'overlap' sections of wood to get a continuous record.
Then you must convert the ring width to temperatures. There are a lot of
noise in each sample record. It is NOT due to climate noise. It relates to
the measurement of temperature by dendrochronology.

>
> W. asked me what I had against smoothing. I am telling you. If you
> don't agree that is your priveledge. It is a free world. But you
> seem to be very close to making the error in believing that tree
> rings are random, whereas infact each one tells a story.

If you just look at tree rings they might as well be random You HAVE to
adjust for both species and the local microclimate. Even the same tree
species may not have the SAME tree ring width at the same average
temperature because of differences in local conditions.

>
> Climate is not just the annual average temperature and precipitation.
> It is also the months of maximum and minimum tempertures, the
> frequency and season of hurricanes, and duration and timing of snow
> cover and permafrost. If you average out the hurricanes over
> five years, you lose the months when they occur and the fact that
> they decrease during El Nino years.

What are you babbling about??

>
> HTH,
>
> Cheers, Alastair.
>
>
>


Alastair McDonald

unread,
Jul 2, 2003, 9:41:06 AM7/2/03
to
"Ian St. John" <ist...@spamcop.net> wrote in message
news:f6rMa.3556$bD1.4...@news20.bellglobal.com...
>

Snip...

> What are you babbling about??

I wondered whether you had the intelligence to follow that argument.
Obviously not!

The hurricanes in the North Atlantic can be correlated with El Nino,
as can other weather events outside the Pacfic Basin. If climate is
treated as average weather over 30 years (or 20 as you prefer) then
this information will be lost (not discovered.)

Each time an average is taken, information is being discarded. It is
not created as many scientists subconciously believe. For instance,
if you take the average of the global temperature (itself an average
I agree) over the last 30 years, you have thrown away the information
that the change in temperature between say year X and X+1 was T.
But you have also thrown away the knowledge of which pair of years
T was the maximum over that period. In other words, an average
contains less information than a series, not more.

HTH,

Cheers, Alastair.


w...@bas.ac.uk

unread,
Jul 2, 2003, 10:20:28 AM7/2/03
to
Alastair McDonald <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote:

>I still don't get it.

Then why not find yourselve some borehole studies to read?

> If you go down the borehole and find perma frost,
>then long ago it was colder. If it is hotter deep down, then previously
>the climate must have been warmer.

If you reconstruct the (hemispheric) T record from boreholes, it looks
like it has got warmer over the last few centuries by more that the sfc T
record shows. Ie, it looks like it was colder in the past. Mann is
trying to reconcile that with the sfc record.

Alastair McDonald

unread,
Jul 2, 2003, 11:25:28 AM7/2/03
to

<w...@bas.ac.uk> wrote in message news:3f02...@news.nwl.ac.uk...
> Alastair McDonald <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote:
>
> >I still don't get it.
>
> Then why not find yourselve some borehole studies to read?
>
> > If you go down the borehole and find perma frost,
> >then long ago it was colder. If it is hotter deep down, then previously
> >the climate must have been warmer.
>
> If you reconstruct the (hemispheric) T record from boreholes, it looks
> like it has got warmer over the last few centuries by more that the sfc T
> record shows. Ie, it looks like it was colder in the past. Mann is
> trying to reconcile that with the sfc record.
>
> -W.

Effectively what he will be measuring is water temperature. Since cold
water is denser than warm water it will sink deeper, and then be heated
by radiogenetic heating from the Earth's crust. He won't be measuring
paleotemperature at all! He ought to give up :-(

HTH,

Cheers, Alastair.


Thomas Palm

unread,
Jul 2, 2003, 11:48:06 AM7/2/03
to

I think William had some excellent advice: "why not find yourselve some
borehole studies to read?". If you pick your sites carefully water
mobility will not be a significant problem. We are talking about
boreholes on land, not ocean temperature profiles!

Obviously there will be heating from below, and this will produce a
bias that has to be removed. Even after doing this and taking all
other known errors into account a large warming is measured.

Ian St. John

unread,
Jul 2, 2003, 2:08:12 PM7/2/03
to

"Alastair McDonald" <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote
in message news:bdun1r$lj7$2...@newsg1.svr.pol.co.uk...

> "Ian St. John" <ist...@spamcop.net> wrote in message
> news:f6rMa.3556$bD1.4...@news20.bellglobal.com...
> >
>
> Snip...
>
> > What are you babbling about??
>
> I wondered whether you had the intelligence to follow that argument.
> Obviously not!
>
> The hurricanes in the North Atlantic can be correlated with El Nino,
> as can other weather events outside the Pacfic Basin. If climate is
> treated as average weather over 30 years (or 20 as you prefer) then
> this information will be lost (not discovered.)

There are always things that are NOT MEASURED because they are not important
to WHAT you are measuring. This has NOTHING to do with losing information.
You still have the same DATA, but are looking as specific INFORMATION. Data
is not information. Do you get that point?

>
> Each time an average is taken, information is being discarded. It is
> not created as many scientists subconciously believe. For instance,
> if you take the average of the global temperature (itself an average
> I agree) over the last 30 years, you have thrown away the information
> that the change in temperature between say year X and X+1 was T.
> But you have also thrown away the knowledge of which pair of years
> T was the maximum over that period. In other words, an average
> contains less information than a series, not more.

No information is lost. If you want to analyse the data for short term
climate oscillations, the cause of noise in dendrochronology readings, the
statistical distribution of noise, spectral power, etc, you can do that.
When you are looking for climate, you extract the climate which does not
need details on the short term weather, because it is the climate. Climate
is not weather. Climate is not the variation from year to year in the
weather. Climate is a LONG TERM AVERAGE of the weather.

>
> HTH,
>
> Cheers, Alastair.
>
>
>
>


James Acker

unread,
Jul 2, 2003, 3:59:34 PM7/2/03
to
Thomas Palm <thoma...@chello.se> wrote:
: Alastair McDonald wrote:

[most deleted for short comment]

:> Soon et al. have just published a paper claiming that during the Medieval


:> Optimum, temperatures were higher than now.

: Actually I don't think the paper says that even if the press release
: does. The article is some kind of meta study that just try to analyze
: a selection of earlier studies and concludes that at some point some
: areas may have been warmer. (I haven't read it, only what has been
: written about it on the group.)

I've read it and have the paper on hand (printed copy, what
a concept!). It certainly does not say that Medieval Warm Period
temperatures were higher than _now_; it literally says that some
variables*, which are related to climate and temperature, may have
exceeded the values achieved during any 50-year period in the 20th
century on a regional comparative basis. They most certainly did
not compare the paleoclimate data to _current_ conditions.

* including things like the thickness of guano deposits for
established seabird rookeries

Jim Acker


*** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
Jim Acker
jac...@gl.umbc.edu
"Since we are assured that an all-wise Creator has observed the
most exact proportions, of number, weight, and measure, in the
make of all things, the most likely way therefore, to get any
insight into the nature of those parts of the creation, which
come within our observation, must in all reason be to number,
weigh, and measure." - Stephen Hales


James Acker

unread,
Jul 2, 2003, 4:23:58 PM7/2/03
to
w...@bas.ac.uk wrote:
: Alastair McDonald <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote:

[most deleted]

:>Soon et al. have just published a paper claiming that during the Medieval


:>Optimum, temperatures were higher than now.

: My understanding is that this is not true: Soon et al assert that, in
: some (varying) regions, T in the past has been higher than now. That,
: even if true, is compatible with global T being highest now. Since I
: haven't read the paper (have you?) I'm unkeen to comment. But as ISJ
: said, its in an odd journal.

Rarely even T. On the last page of the article, they list
the various abbreviations for study types:

Borehole
Cultural
Documentary
Glacier advance or retreat
Geomorphology
Instrumental
Isotopic analysis (ice and sediment cores, trees, corals, stalagmites,
fossils)
Net ice accumulation rate (via dust, chemical counts)
Lake fossils and sediments; river sediments
Melt layers in ice cores
Multiproxy (combinations of the discrete proxies listed here)
Phenological/paleontological fossils
Pollen
Seafloor sediments
Speleothem isotope or luminescence
Tree ring growth
Submerged tree stumps

Just for fun, I'll provide a reference for one of the "documentary"
studies:

Chan, JCL, Shi, JE (2000) Frequency of typhoon landfall over
Guangdong Province, China during the period 1470-1931. Int. J. Clim.
20: 183-190.

So if there were more typhoons hitting the south China
coast during ANY given 50-year period in the record than during any
50-year period in the 20th century, that would count as a "yes",
i.e., that period was "more extreme" than any 50-year period in the
20th century.

That's a nice example of how far they were stretching in
this study.

Alastair McDonald

unread,
Jul 2, 2003, 4:03:11 PM7/2/03
to

"Thomas Palm" <thoma...@chello.se> wrote in message
news:3F02FEB6...@chello.se...

> Alastair McDonald wrote:
> >
> > <w...@bas.ac.uk> wrote in message news:3f02...@news.nwl.ac.uk...
> > > Alastair McDonald <alas...@abmcdonald.leavethisout.freeserve.co.uk>
wrote:
> > >
> > > >I still don't get it.
> > >
> > > Then why not find yourselve some borehole studies to read?
> > >
> > > > If you go down the borehole and find perma frost,
> > > >then long ago it was colder. If it is hotter deep down, then previously
> > > >the climate must have been warmer.
> > >
> > > If you reconstruct the (hemispheric) T record from boreholes, it looks
> > > like it has got warmer over the last few centuries by more that the sfc
T
> > > record shows. Ie, it looks like it was colder in the past. Mann is
> > > trying to reconcile that with the sfc record.
> > >
> > > -W.
> >
> > Effectively what he will be measuring is water temperature. Since cold
> > water is denser than warm water it will sink deeper, and then be heated
> > by radiogenetic heating from the Earth's crust. He won't be measuring
> > paleotemperature at all! He ought to give up :-(
>
> I think William had some excellent advice: "why not find yourselve some
> borehole studies to read?". If you pick your sites carefully water
> mobility will not be a significant problem.
> We are talking about
> boreholes on land, not ocean temperature profiles!

So are you talking about phreatic or vadose water then?

Alastair McDonald

unread,
Jul 2, 2003, 4:05:24 PM7/2/03
to

"Ian St. John" <ist...@spamcop.net> wrote in message
news:hcFMa.4224$bD1.5...@news20.bellglobal.com...

So is ENSO a weather or a climate phenomenon?


Ian St. John

unread,
Jul 2, 2003, 4:43:29 PM7/2/03
to

"Alastair McDonald" <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote
in message news:bdvf4i$9mo$2...@newsg1.svr.pol.co.uk...

If you get a normal pattern of a heavy rain followed by three days of
sunshine, darkening clouds, and rain again, what is this? It isn't weather.
Weather is the state NOW.

Closest description would be 'strange attractors' in chaos theory. Patterns
in the weather changes that lead to cyclical repetitions.

ENSO is called a 'climate oscillation' but in truth it does not represent a
change in the climate. It is a pattern in the weather, just a rather long
term one.

The NAO, for example, is defined by the relative air pressure differential
between two points on the planet. Climate is not defined as relative
pressure differentials. The usefullness of the description is ENTIRELY
because it can indicate which 'cyclical repetition' cycle is currently in
place.


John Mashey

unread,
Jul 3, 2003, 3:43:43 AM7/3/03
to
w...@bas.ac.uk wrote in message news:<3f01...@news.nwl.ac.uk>...
> Alastair McDonald <alas...@abmcdonald.leavethisout.freeserve.co.uk> wrote:
> >Perhaps I should admit that I do not know what "the variance in the
> >first eigenvector" means!
>
> The various eigenvectors are orthogonal. So the total variance of
> the series can be decomposed into the variance associated with the
> various eigenvectors. If the first few eignevectors account for most
> of the variance, then the series "really" has few degrees of freedom.
>
> >Mann's 'Hockey Stick' paper seems to have been crucial in persuading the
>
> Um. We don't say HS in polite company.

I think the use of the "Mann chart" [I won't say HS] has been
unfortunate:
Without criticizing anybody, I'd observe the following sequence:

A) The paper by Mann, Bradley, Hughes, 1999, includes this chart,
but with numerous *important* caveats.

B) In the IPCC TAR (Chapter 2, of which Mann is one of the lead
authors),
some of the specific caveats have disappeared, but the chapter gives a
pretty good overall discussion, with plenty of other relevant caveats.
p.134 (of my printed version) shows both the Mann chart and another
that includes other reconstructions. This is important: a savvy
reader quickly understands that this science is *still* under
discussion, as people make great efforts to extract signals from
difficult data sources, and the resulting graphs differ, i.e., science
as usual.
Of course, discussions continue, as in various articles in Science
over last couple years, with various versions of these charts. In
this posting, it seems fair to only use info up to the TAR, not later
papers.

C) In the IPCC TS, the Mann chart appears alone, prominently
displayed,
and most caveats have disappeared. [It's a summary, of course, so
some of this is inherent.] However, the effect of the presentation is
that the Mann chart is *the* record for the last millenium.

D) Finally, in the IPCC SP, the two leading charts (p.3) are the
1860-2000 Global temperature chart and the same Mann chart. A reader
might assume that there is equal confidence in the central lines on
the two charts.
Recall that the SP is intended for policy-makers, i.e., like US
Senators
[of whom many *strongly* prefer words to numbers and graphs. I'd be
astonished if most Senators had read the SP, much less the TS. ]

In general, in the progression from A) to D), caveats get lost and the
appearance of certainty increases compared to original sources.
Again, just an observation, not a criticism.

From long experience, if you can generate a striking chart, it will
stick in people's minds, and the attached fine-print caveats don't.
Likewise, news organizations will often pick up the graph unchanged
and reprint it while losing the caveats [for instance, this has long
been done by computer vendors in showing results to the press :-)]


SO, WHAT'S THE PROBLEM?
As usual, even with a pretty competent secondary source [like the
TAR],
it's a good idea to go back to the original primary paper:

Mann, Bradley, Hughes, "Northern Hemisphere Temperatures During the
Past Millenium: inferences, Uncertainties, and Limitations." AGU GRL
1999.
http://www.people.virginia.edu/~mem6u/preprints.html

The Abstract includes (sentences #2 & #3 of 5): (*CAPS* mine):

"We focus not just on the reconstructions, but the uncertainties
therein, and important caveats. Though expanded uncertainties
*PREVENT DECISIVE CONCLUSIONS FOR THE PERIOD PRIOR TO AD 1400*, our
results suggest that the latter 20th century is anomalous in the
context of at least the past millenium."

Now, if you look at the chart alone there is *nothing* whatsoever to
suggest a lack of "decisive conclusions" before 1400. [IF there was
any MWP, it was likely contained in the 800-1350AD interval, before
1400.]
It is useful to go back to the 1998 paper, which basically describes
the number of *indicators* used:
1820- 112 indicators
1760- 93
1700- 74
1600- 57
1450- 24
1400- 22
============================
1000- 12 (from 1999 paper)
Also, as they go further back, skill of reconstruction gets worse of
course [and they discuss this well in the 1998 paper].

The 12 indicators pre-1400 are:
SERIES Loc Year Type
ITRDB (PC #1) N. Amer 1000 Ring width ** +
ITRDB (PC #2) N. Amer 1000 Ring width
ITRDB (PC #3) N. Amer 1000 Ring width
Fennoscandia 68N 23E 500 Ring density +
Polar Urals 67N 65E 914 Ring density +
Tasmania 43S 148E 900 Ring width *
N. Patagonia 38S 68W 869 Ring width *
Morocco 33N 5W 984 Ring width
France 44N 7E 988 Ring width
Greenland core 77N 60W 553 ice core *
Quelccaya(2) 14S 71W 488 ice core * +
Quelccaya (2) 14S 71W 488 ice accum * +

* = Southern Hemisphere
+ have enough red noise
** "one such indicator - PC #1 of the ITRDB data - is found to be
essential"

"Only 5 of the indicators (including the ITRDB PC #1, Polar Urals,
Fennoscandia, and both Quelccaya series) are observed to have at least
median red noise at zero frequency for the pre-calibration (AD
1000-1901) period. it is furthermore found that only one of these
series - PC #1 of the ITRDB data-exhibits a significant correlation
with the dominant temperature pattern ofteh 1902-1980 calibration
period. Positive calibration/variance scores for the NH series cannot
be obtained if this indicator is removed from the network of 12..."

It's well worth reading the 1999 and 1998 papers, as there is plenty
of good-looking work to try to extract meaningful signals from
less-than-great data;
there is of course lots of math to convert/adjust the various data
items into temperatures.

From the above, I infer (and correct me if this is wrong, I'm
certainly no dendrochronologist):

*****
Basically, the 1000-1400 *NH* numbers seem to be primarily derived
from *one* series of North American tree ring data ...
*****
and that's OK, they're doing the best they could ... but I don't like
the chart.
It might be prefectly correct, but it mixes relatively high-confidence
results with ones that are not so high, without making that clear.

I would be much, much happier if:
a) The original chart had interval bands, or something, that clearly
indicated the differing natures of the data sets used for different
intervals. It has a strong visual appearance that is totally at
variance with the caveats.

b) Data caveats hadn't kept disappearing along the way.

c) The comments showed a little less certainty about weakness (LIA) or
non-existence (MWP). The reasoning seemed to be:
[one N. American tree ring series, from 1000-] => non-globalness of
MWP.
I'm not convinced the MWP was global, but I'm not convinced that it's
been proved a minor effect only seen around N. Atlantic either.
[newer papers?]
From the arguments in Science, this is not quite yet a done deal.

d) IPCC would have just said: "We don't yet really know enough about
800-1400; the MWP may or may not have affected anything beyond the
North Atlantic."

=================================
Just so you don't get the wrong idea:

I consider it virtually certain:
- that there is plenty of good science in IPCC TAR
- that most of the IPCC SP is correct, and that I agree with most of
it
- that an important part of the current warming is anthropogenic
- that we've got lots of actions that we need to take,
with some being 100% clear, and some being less so, and with the
practical
politics being sometimes murky [i.e., Kyoto (?)]
- that there are physical mechanisms we don't understand well enough
[i.e., some of the cloud stuff, and maybe the SAT-vs-satellite
stuff,
if that's not just calculation problems]
- that our data-gathering ought to be improved, a lot.
- that datasets vary in reliability and usefulness...
- that reconstructions vary... and get recomputed sometimes ...
- that our computer models aren't done yet...
- that much of the science is settled, but there is plenty that isn't.
- that it is really important to look at everything skeptically,
because
it's complicated, and there's a lot of indirect analysis.

I'd really love to know the real story on the LIA and MWP; from the
various histories, it sure seems like: a) the LIA had effects beyond
Europe
b) at least in Europe-Greenland, there was a pretty big contrast
between the MWP and LIA periods, not just a little change. Maybe the
rest of the world didn't have such a clear change from one regime to
another.
But in any case, it would be good to know. (From my point of view,
the extremes include people who really want an MWP to those who really
don't want it to have existed. I don't care, I'd just like to know
the truth.]

If it turns out that big chunks of the world did have an MWP, and it
was as warm as now ... this doesn't mean to me "OK, don't need to do
anything". It says: a millenium ago (or maybe 1100 years ago), maybe
the Earth was at the same temperature as now naturally ... but (I
think) there were on the order of 300M people, whereas we're at 6B now
and going higher. If humans are going to last a while, we're
certainly going to have to:
- learn how to adapt to inevitable climate swings that happen without
us,
in either direction [after all, given a choice between more global
warming
and an end to the interglacial, I'd probably prefer the former.]
- avoid exacerbating the swings too much ourselves
- do so while *raising* our level of technology
[After all, only a higher-tech civilization has much chance
of stopping the next big asteroid hit, which from past history is
*inevitable*, not just virtually certain.]

========
(I grew up on a small farm that had been in the family 120 years,
started by Swiss who were (presumably) escaping the LIA to the US.
Anybody who grows up this way is required to learn something about
ecology, conservation, carrying-capacity, planning for weather
variations, etc ... by the time they're 10.
You learn not to do stupid things like:
"We're in a drought, and the pastures are over-grazed, so lets plant
water-hungry crops and buy some more cattle."

w...@bas.ac.uk

unread,
Jul 3, 2003, 4:45:01 AM7/3/03
to
John Mashey <old_sys...@yahoo.com> wrote:

>...


>Mann, Bradley, Hughes, "Northern Hemisphere Temperatures During the
>Past Millenium: inferences, Uncertainties, and Limitations." AGU GRL
>1999.
>http://www.people.virginia.edu/~mem6u/preprints.html

>The Abstract includes (sentences #2 & #3 of 5): (*CAPS* mine):

>"We focus not just on the reconstructions, but the uncertainties
>therein, and important caveats. Though expanded uncertainties
>*PREVENT DECISIVE CONCLUSIONS FOR THE PERIOD PRIOR TO AD 1400*, our
>results suggest that the latter 20th century is anomalous in the
>context of at least the past millenium."

Thanks for this post, I hadn't realised the text that you include above.

>1000- 12 (from 1999 paper)

>...

>...


>Basically, the 1000-1400 *NH* numbers seem to be primarily derived
>from *one* series of North American tree ring data ...

I'm no dendrochronologist either but I think this is not quite
correct: as it says: one if essential; I presume that the implication
is that the others are useful (or they wouldn't be there).

>a) The original chart had interval bands, or something, that clearly
>indicated the differing natures of the data sets used for different
>intervals. It has a strong visual appearance that is totally at
>variance with the caveats.

A good idea. It could have had bands; or perhaps an overplotted line
with "number of data sources used at this date". At least that
wouldn't disappear.

>b) Data caveats hadn't kept disappearing along the way.

I think this is an unfair criticisim: summaries are summaries. They
have to get shorter. As long as they point back to the original (and
they do) then if you don't read the original caveats, thats your
problem, not theirs.

>(I grew up on a small farm that had been in the family 120 years,
>started by Swiss who were (presumably) escaping the LIA to the US.

Aha! Certain proof that the LIA wasn't global ;-)

Thomas Palm

unread,
Jul 3, 2003, 8:46:49 AM7/3/03
to
w...@bas.ac.uk wrote:

>
> John Mashey <old_sys...@yahoo.com> wrote:
> >b) Data caveats hadn't kept disappearing along the way.
>
> I think this is an unfair criticisim: summaries are summaries. They
> have to get shorter. As long as they point back to the original (and
> they do) then if you don't read the original caveats, thats your
> problem, not theirs.

Summaries are supposed to give a fair view of the subject being
summarized. You are not supposed to have to read the entire
report just to understand what is being said in the summary, or
the summary becomes worthless.

You will have to drop caveats and other details in a summary, but
if you reach the point where the result is misleading rather than
giving useful information then you either have to add some more
text or drop it altogether. You can't assume that eveyone will
read the full report, or blame them for not doing so.

Now, I don't think IPCC did that error. Looking at figure 1 in the
summary for policymakers the large error bars for earlier data are
quite prominent and the text point out the uncertainties and that
they increase the further back you go. The exact reasons for the
increased uncertainty does not belong in a summary.

It's a good rule never to trust the end points of this type of
diagrams very much anyway. If the first part was accurate the
researchers wouldn't have been able to stop themselves from trying
to extend it even further back.

> >(I grew up on a small farm that had been in the family 120 years,
> >started by Swiss who were (presumably) escaping the LIA to the US.
>
> Aha! Certain proof that the LIA wasn't global ;-)

:-)

John Mashey

unread,
Jul 3, 2003, 4:11:32 PM7/3/03
to
Thomas Palm <thoma...@chello.se> wrote in message news:<3F0425B9...@chello.se>...

Sorry, I thought I'd been clear that I *understand* the difficulty of


summarization when I wrote:
"and most caveats have disappeared. [It's a summary, of course, so
some of this is inherent.]"

So, let me try again, noting that:
1) I long ago read "How to Lie with Statistics" and "How to Lie with
Maps", and I keep several [so I can lend them} copies of Edward
Tufte's fine books, of which some parts are effectively "how to lie
with graphs", although of course most of the books are about "how to
tell the truth with graphs, well."

Those are the book titles. I am *not* in any way claiming that IPCC
was trying to lie. In fact, for most of the TAR->TS->SP summarization
process, I thought they did a pretty good job, and I've spent many
years creating/analyzing technology and technical data and then
presenting it to audiences ranging from very technical to
non-technical [i.e., explain "cache" and "floating point" to your
company's administrative assistants], so I understand how hard it is,
and I appreciate the large amount of fine work.

2) I spent some years of my life trying to improve the state of
computer performance measurement, which in the 1980s often still used
truly awful data (Dhrystones, Whetstones, "mips"), and because of the
commerical implications, was often subject to the "how to lie with
XXX" behavior. Even when I wrote performance documents that carefully
showed multiple realistic datapoints, lots of people would want to
know just one number to summarize it all. Computer performance
measurement and prediction is *way* easier than climatology, but it
was still a mess. It was especially wasteful that there were computer
vendors who spent time and money optimizing systems for some of the
dumber benchmarks ... doing things that in fact gave no help to real
programs.
Out of frustration, a few of us started the SPEC benchmarking group,
which, even with its own limitations, promoted a much better
multi-point methodology using lots of real codes, and causing computer
vendors to spend more time making improvements that actually helped
people. However, it was a real education in learning how to
*communicate* results to non-experts, and how easily mis-communication
can happen.

[Note: for climate, it seems imperative to me that we measure things
well, and that we prioritize the various "fixes" and make sure they
work, given that the stakes are rather higher than whether one
computer is 5% faster than another.]

ANYWAY: I propose the following experiment:
a) Read the original paper [a big chunk of which is devoted to
"inferences, uncertainties, and limitations", just like the title
says.

b) Simulate a "policymaker" reading the SP, i.e., you probably don't
read Science, you don't look at GISS data, you don't read this
newsgroup, and you're probably not used to error bars, and you are not
going to track back to the original paper.

THEN: ask yourself if you feel good about the comparison of a) and b).

All I'm saying is that, given the chart on p.3 of the SP:
1) It is all too easy to draw conclusions from it that are not
warranted by the original source, especially by the audience for which
the SP is intended.
Do you think the SP audience is used to error bars? I deem it
unlikely.

2) When I look at the SP *chart*, what I see is:
a) An interval from 1850 onwards where we have direct measurements,
and this is clearly distinguished.
b) An interval from 1600-1850 where we have data with small error
bar.
c) An interval from 1000-1600 where we have data with a larger
error bar.
and there is not the slightest visual clue of the differences in data,
for which one needs to go back to both the 1999 and 1998 papers.

I care a *lot* about charts that don't mislead people. I can think of
several ways to have done this chart in ways that would be less
misleading. For example, the line from 1000-1400 could have been
dotted, from 1400-1600 dashed, and from 1600- solid. I'll admit I
don't yet understand why the error bars for 1000-1400 are the same
size as 1400-1600 [or at least, they look that way.]
[Maybe somebody can read the two papers and explain that?]

3) Of course, although outside the IPCC chain, but quite typical in
the way data gets degraded, consider "Climate Change Impacts on the
United States", 2000. P 12 has the chart, but the error bars have
disappeared *completely*.

4) Given all of the above, of course I never trust endpoints :-), but
then, I consider most data guilty into proven innocent :-)

5) [for wmc] I am unimpressed by "caveat emptor" arguments about
disappearing caveats :-) that is: of course a lot of caveats
disappear, but I am very uncomfortable with the idea that it is OK to
use a chart that is likely to be interpreted in a particular way, but
that one is "covered" by having footnotes several layers deeper in the
chain of references. I understand this is hard to do, and I
understand quite well that scientific literature and info itnended for
the public have different requirements.

6) But why am I irked at this? Since I mostly take the view that there
is global warming, and that it is unusual, I'm not happy when the net
result of this whole process looks like a premature
drawing-of-strong-conclusions and communication of such, when the
original paper was careful.

I would have been much happier if that chart made it clear that there
was a substantial qualitative difference between the pre-1400 and
post-1400 data.
As it stands, it certainly gives denialists an easier target.

Also, from past experience, given that science is done by humans,
premature conclusions sometimes slow the science down, as positions
sometimes harden prematurely as well, and datasets get defended to the
death on all sides. :-)

I'm always happier when people make it crystal clear what they know,
and what they don't, and the difference between analyses based on
multiple data sources, and an early analysis pushing back into the
unknown on top of one data source. The original article *was* pretty
clear ... except for the chart.

7) [for wmc]: re LIA and Swiss ancestors: this is unclear :-)
In this particular case, they were moving (1840s) from Buren (between
Basel and Bern), ~ 47.3 degrees North, to Pittsburgh, PA, 40.5 degrees
North, which is the same latitude as Tarranto (S end of Italy).
Although Basel has good records (back to 1754), Pittsburgh only goes
back to 1870, and eyeballing the charts says there was a typical 2-3
degreesC difference.

0 new messages