EAS Seminar, Georgia Tech

1
EAS Seminar, Georgia Tech

• Stephen McIntyre
• Toronto Ontario
• Atlanta Feb 8, 2008

2
An interesting ride

• Left Front page, Wall Street Journal, Feb 2005
  Middle - House Energy and Commerce Committee,
  Oversight and Investigations Subcommittee Mann,
  Ralph Cicerone (NAS), me, Jay Gulledge, Ed
  Wegman Right Best Science Blog 2007

3
Conclusions

• No meaningful conclusion about modern-medieval
  climate can be derived from the existing corpus
  of 1000 year studies and which is warmer remains
  an open question
• The main problem is not finding the right
  multivariate method, but getting better proxies
  and better local data.

4
Some Disclaimers

• Ive never asserted or implied that AGW policy
  stands or falls on the HS.
• Ive never suggested that perfect certainty is
  needed to make climate decisions. People make
  decisions under uncertainty all the time.
• Ive avoided all policy discussions other than
  archiving. If I had a big policy job, I would be
  guided by official institutions. Ive never
  suggested that any of the issues that Ive been
  involved should affect climate policy decisions.
• Conversely, the big picture doesnt excuse
  using poor methods or poor dislosure.
• Even if the HS is irrelevant to climate policy,
  Ive become interested in the statistical and
  historical issues in 1000-year climate
  reconstructions.

5
The Hockey Stick

• Left John Houghton at IPCC TAR WG1 Press
  conference right - Al Gore, Inconvenient Truth

6
The iconic hockey stick
7
The Warmest Decade and Year

•   The past decade was the world's warmest decade
  of the century. And that century was the warmest
  of the past millennium. Without action, the
  long-term consequences will be devastating.
  David Anderson, Oct. 27, 2001
• The 20th century was the warmest in the Northern
  Hemisphere in the past 1000 years. The 1990s was
  the warmest decade on record and 1998 was the
  warmest year - in Canada and internationally." -
  David Anderson, April 5, 2002
• The 20th century was the warmest in the Northern
  Hemisphere for the past 1000 years and the 1990s
  the warmest decade on record... The science of
  climate change has been subjected to
  international scrutiny, open to all qualified
  experts, peer review, atmospheric modeling and
  process studies Liberal Caucus, Aug. 22, 2002

8
Forgotten the Location

•   Dear Dr. Mann, I have been studying MBH98 and
  99. I located datasets for the 13 series used in
  99 at ftp//eclogite.geo.umass.edu/pub/mann/ONLINE
  -PREPRINTS/Millennium/DATA/PROXIES/  (the
  convenience of the ftp location being excellent)
  and was interested in locating similar
  information on the 112 proxies referred to in
  MBH98, as well as listing (the listing at
  http//www.ngdc.noaa.gov/paleo/ei/data_supp.html
  is for 390 datasets, and I gather/presume that
  many of these listed datasets have been condensed
  into PCs, as mentioned in the paper itself. Thank
  you for your attention.  Yours truly, Stephen
  McIntyre, Toronto, Canada
• Dear Mr. McIntyre,  These data are available on
  an anonymous ftp site we have set up. I've
  forgotten the exact location, but I've asked my
  Colleague Dr. Scott Rutherford if he can provide
  you with that information. best regards, Mike
  Mann
• Steve, The proxies aren't actually all in one
  ftp site (at least not to my knowledge). I can
  get them together if you give me a few days. Do
  you want the raw 300 proxies or the 112 that
  were used in the MBH98 reconstruction? Scott

9
MBH98

•   The Neofs-length solution vector g is obtained
  by solving the above overdetermined optimization
  problem by singular value decomposition for each
  proxy record i 1,Nproxy. This yields a matrix
  of coefficients relating the different proxies to
  their closest linear combination of the Neofs PCs
  .This set of coefficients will not provide a
  single consistent solution, but rather represents
  an overdetermined relationship between the
  optimal weights on each on the Neofs PCs and the
  multiproxy network.
• These Neofs eigenvectors were trained against
  the Nproxy indicators, by finding the
  least-squares optimal combination of the Neofs
  PCs represented by each individual proxy
  indicator during the N79 year training interval
  from 1902 to 1980 (the training interval is
  terminated at 1980 because many of the proxy
  series terminate at or shortly after 1980). This
  proxy-by-proxy calibration is well posed (that
  is, a unique optimal solution exists) as long as
  NgtNeofs (a limit never approached in this study)
  and can be expressed as the least-squares
  solution to the overdetermined matrix equation,
  Ug Y,i , where U is the matrix of annual PCs,
  and Y,i is the time series vector for proxy
  record i. The Neofs-length solution vector g is
  obtained by solving the above overdetermined
  optimization problem by singular value
  decomposition for each proxy record i1,Nproxy.
  This yields a matrix of coefficients relating the
  different proxies to their closest linear
  combination of the Neofs PCs .This set of
  coefficients will not provide a single consistent
  solution, but rather represents an overdetermined
  relationship between the optimal weights on each
  on the Neofs PCs and the multiproxy network.

10
Statistical Skill

• IPCC TAR reconstruction which had significant
  skill in independent cross-validation tests.
  Self-consistent estimates were also made of the
  uncertainties
• MBH98 ß or RE correlation (r) and
  squared-correlation (r2) statistics are also
  determined.

11
Robustness to presence/absence of dendro proxies

• Mann et al. 2000 possible low-frequency bias due
  to non-climatic influences on dendroclimatic
  (tree-ring) indicators is not problematic in our
  temperature reconstructionsWhether we use all
  data, exclude tree rings, or base a
  reconstruction only on tree rings, has no
  significant effect on the form of the
  reconstruction for the period in question.
• MBH98 the long-term trend in NH is relatively
  robust to the inclusion of dendroclimatic
  indicators in the network, suggesting that
  potential tree growth trend biases are not
  influential in the multiproxy climate
  reconstructions. (p. 783, emphasis added.)

12
MM2003, 2005a,b,c,d

• MBH principal components algorithm was severely
  biased such that it mined for HS-shaped data.
  This precluded attributing any statistical
  significance to such a reconstruction.
• their reconstruction failed important statistical
  tests in early portions e.g. the verification r2
  statistic shown in one of their figures. These
  adverse results were not reported.
• we reflected on what it means when one statistic
  is 99.9 significant and another statistic is a
  bust.
• their reconstruction was not robust to the
  presence/absence of all dendro proxies. It was
  not even robust to the presence/absence of
  bristlecones..
• Bristlecones are located only in high arid U.S.
  Southwest and had been previously identified by
  specialists as problematic, e.g. Biondi (Hughes)
  et al. (1999) said Bristlecones are not a
  reliable temperature proxy for the last 150
  years. Without bristlecones, no HS.
• Merely using a standard covariance PC algorithm,
  the weights of bristlecones were reduced such
  that no HS was obtained.

13
Red Noise Simulations

• Data decentered against post-1902 mean.
  Preferentially adds weight to hockey stick-shaped
  series in PC1

14
The counterattack

• Mann This claim by MM is just another in  a
  series of disingenuous (off the record plainly
  dishonest) allegations by them about our work.
• I hope you are not fooled by any of the "myths"
  about the hockey stick that are perpetuated by
  contrarians, right-wing think tanks, and fossil
  fuel industry disinformation. UCAR Press
  Release the highly publicized criticisms of the
  MBH graph are unfounded.

15
The Defense

• Our criticisms were wrong because
• They could get an HS without using principal
  components.
• If they retained more PCs (down to the PC4), they
  could still get a HS
• Other people got an HS using different methods.
• In MM (EE 2005) we discussed these permutations
  and combinations. Yes, you can get an HS by
  changing your methods, but each of these other
  methods has its own problems.Wegman was vey
  critical of after-the-fact methodology changes.
• After Wahl and Ammann grudgingly confirmed the
  verification r2 failure, they argued that
  verification r2 statistic was no good as it was
  supposedly prone to rejecting valid
  reconstructions. Mann to the NAS Panel
  calculating a verification r2 statistic would be
  a foolish and incorrect thing to do.
• Wahl and Ammann agreed that you cant get a HS
  without the bristlecones, but argued that a
  bristlecone-free reconstruction fails
  verification RE statistic) and would never have
  been proposed and thus claim that bristlecones
  contain valid and necessary information.

16
Proposal to Ammann and Wahl

• in my view, the climate science community has
  little interest at this point in another exchange
  of controversial articles (and associated weblog
  commentaries) and has far more interest in the
  respective parties working together to provide a
  joint paper, which would set out (1) all points
  on which we agree (2) all points on which we
  disagree and the reasons for disagreement (3)
  suggested procedures by which such disagreements
  can be resolved. Because our emulations are
  essentially identical, I think that there is
  sufficient common ground that the exercise would
  be practical, as well as desirable.
• Accordingly I propose the following
• (1)   we and our coauthors (McKitrick and Wahl)
  attempt to produce a joint paper in which the
  above three listed topics are discussed
• (2)   We allow ourselves until February 28, 2006
  to achieve an agreed text for submission to an
  agreed journal (Climatic Change or BAMS, for
  example, would be fine with us), failing which we
  revert back to the present position
• (3)   as a condition of this ceasefire, both
  parties will put any submissions or actions on
  hold. On your part, you would notify GRL and
  Climatic Change of a hold until Feb. 28, 2005. On
  our part, we would refrain from submitting
  response articles to GRL or Climatic Change or
  elsewhere and refrain from blog commentary on the
  topic.

17
Wegman Report 2006

• The debate over Dr. Manns principal components
  methodology has been going on for nearly three
  years. When we got involved, there was no
  evidence that a single issue was resolved or even
  nearing resolution. Dr. Manns RealClimate.org
  website said that all of the Mr. McIntyre and Dr.
  McKitrick claims had been discredited. UCAR had
  issued a news release saying that all their
  claims were unfounded. Mr. McIntyre replied on
  the ClimateAudit.org website. The climate science
  community seemed unable to either refute
  McIntyres claims or accept them. The situation
  was ripe for a third-party review of the types
  that we and Dr. Norths NRC panel have done.
• While the work of Michael Mann and colleagues
  presents what appears to be compelling evidence
  of global temperature change, the criticisms of
  McIntyre and McKitrick, as well as those of other
  authors mentioned are indeed valid.
• I am baffled by the claim that the incorrect
  method doesnt matter because the answer is
  correct anyway. Method Wrong Answer Correct
  Bad Science.

18
NAS Panel Schizophrenic

• Agreed with our point that the MBH PC method was
  wrong
• Agreed that bristlecones were problematic and
  said that strip bark trees should be avoided
  in temperature reconstructions
• Agreed in general terms with our statistical
  criticisms Reconstructions that have poor
  validation statistics (i.e., low CE) will have
  correspondingly wide uncertainty bounds, and so
  can be seen to be unreliable in an objective
  way.
• On no occasion did they contradict any explicit
  MM statement.

19
NAS Panel

• Concluded that things were murky as you went
  further back.
• Overall result was plausible showing other
  recons.

20
North at the House Committee

• CHAIRMAN BARTON. Dr. North, do you dispute the
  conclusions or the methodology of Dr. Wegmans
  report?
• DR. NORTH. No, we dont. We dont disagree with
  their criticism. In fact, pretty much the same
  thing is said in our report.
• DR. BLOOMFIELD. Our committee reviewed the
  methodology used by Dr. Mann and his coworkers
  and we felt that some of the choices they made
  were inappropriate. We had much the same
  misgivings about his work that was documented at
  much greater length by Dr. Wegman.

21
Zorita on the NAS Panel

• in my opinion the Panel adopted the most
  critical position to MBH nowadays possible. I
  agree with you that it is in many parts
  ambivalent and some parts are inconsistent with
  others. It would have been unrealistic to expect
  a report with a summary stating that MBH98 and
  MBH99 were wrong (and therefore the IPCC TAR had
  serious problems) when the Fourth Report is in
  the making. I was indeed surprised by the
  extensive and deep criticism of the MBH
  methodology in Chapters 9 and 11.

22
Norths Texas AM Seminer

• At a Texas AM seminar, North said that they
  didnt do any research, that they just took a
  look at papers, that they got 12 people around
  the table and just kind of winged it
  http//www.met.tamu.edu/people/faculty/dessler/Nor
  thH264.mp4 minute 55 or so
• We did not dissect each and every study in the
  report to see which trees were used.

23
Why hasnt this been settled?

• the issue isnt important enough that it needs to
  be settled
• Science may be self-correcting (so are
  markets). Market imperfections can continue
  uncorrected for some time and so can science
  imperfections.
• still some intellectual work that needs to
  isolate the issues more surgically than has been
  done to date.

24
Outline

• Reconstructions in a statistical context
• Econometric thoughts on spurious regression
• Some important data sets

25
A. Reconstruction Methods Jones and Mann 2004

• Composite-plus-scale (CPS)
• Proxies are normalized and averaged (perhaps with
  weights e.g., based on area represented or
  modern correlations with colocated instrumental
  records Jones and Mann 2004
• average is then simply scaled against the
  available temporally overlapping instrumental
  record
• Climate Field Reconstruction (CFR)
• Multivariate calibration of the large-scale
  information in the proxy data network against the
  available instrumental data
• most involve the use of empirical eigenvectors
  of the instrumental data, the proxy data, or
  both
• Because the large-scale field is simultaneously
  calibrated against the full information in the
  network, there is no a priori local relationship
  assumed between proxy indicator and climatic
  variable.

26
Multivariate CalibrationCook (Briffa, Jones)
1994

• Considers inverse OLS regression of each
  gridcell in a network X (MBH98 1082) against a
  network of proxies Y (MBH98 415)
• Applying this to the MBH98 network would lead to
  the generation of 1082415449,030 coefficients
  from only 7941532,785 proxy measurements in the
  calibration period.
• Cook et al observed
• experience in reconstructing climate from tree
  rings indicates that such models frequently
  produce reconstructions that cannot be verified
  successfully when compared with climate data not
  used in estimating the regression coefficients.
  This can happen regardless of the statistical
  significance of the overall regression equation.

27
MBH Method

1. Standardize gridcell temperatures by
   area-weighting and standard deviations
2. Calculate temperature PCs
3. Estimate transfer coefficients for proxies to
   PCs in calibration period by over-determined
   optimization
4. Estimate reconstructed PCs in reconstruction by
   over-determined optimization
5. Re-scale reconstructed PCs to observed PCs
6. Expand back to gridcell temperatures
7. Calculate NH average temperature

28
Advantage of MBH Method

• MBH98s method yields an estimation of the
  value of the temperature PCs that is optimal for
  the set of climate indicators as a whole, so that
  the estimations of individual PCs cannot be
  traced back to a particular subset of indicators
  or to an individual climate indicator. This
  reconstruction method offers the advantage that
  possible errors in particular indicators are not
  critical, since the signal is extracted from all
  the indicators simultaneously.

29
NH Composite Linear Algebra
All the operations are linear and on the right of
the reconstructed PCs. Simple expression for
expansion and reflating of reconstructed PCs to
area-weighted NH average
where
- Reconstructed NH temperature index
- Reconstructed TPCs
- Gridcell eigenvalues, eigenvectors (EOFs)
- Gridcell standard deviations
- Gridcell area weights
30
These are the weights (?) for each reconstructed
temperature PC in NH temperature reconstruction
31
All operations in reconstructed Temperature PCs
are linear. Optimizations are Regressions
Calibration
Reconstruction
Rescaling in 1-D Case
1-D Case (AD1000, AD1400)
- Matrix of temperature PCs (individual PC)
- Matrix of proxies (standardized)
- Rescaled estimated PCs Pre-rescaled PC
estimates
- Arbitrary MBH weights correlation to
temperature PC(1)
32
All MBH operations are linear operations on the
right hand side of the proxy matrix Y. Because
matrix multiplication is associative, weights can
be assigned for each individual proxy.
Where, for one-PC case (AD1000, AD1400)
weighted
unweighted
S,V from temperature SVD, s, µ are gridcell
standard deviations, cosine latitude area weights
33
Weights can be assigned for proxies within PC
networks as well
Regional tree ring PCs are linear combinations of
the underlying tree ring network (here X)
The mixed MBH proxy network (regular and tree
ring PCs can be represented as follows (the
dimension of V,k is (say) 79 x2 (for 79 tree
rings and 2 PCs)
34
Contributions by proxy type can be calculated
By allocating weights to individual sites and by
classifying sites by continent and type (ice
core, bristlecone, other trees, coral, etc), the
relative contribution of each class to the MBH
reconstruction can be measured (bristlecone plus
Gaspé in red). (In deg C)
35
Weights can be shown graphically
Prominent weights are for the NOAMER PC1
(bristlecones), Gaspé, Tornetrask and Cooks
Tasmania tree ring. There are 22 weights in this
picture in total.
36
Keynes 1940 on Tinbergen (anticipating RegEM?)
But my mind goes back to the days when Mr Yule
sprang a mine under the contraptions of
optimistic statisticians by his discovery of
spurious correlation. In plain terms, it is
evidence that if what is really the same factor
is appearing in several places under various
disguises, a free choice of regression
coefficients can lead to strange results. It
becomes like those puzzles for children where you
write down your age, multiply, add this and that,
subtract something else and eventually end up the
number of the Beast in Revelation.
37
Different weights yield different
reconstructions. Plausible ex ante methods yield
different weights and results.
Burger and Cubasch 2006
38
MBH and Scaled-Composite Reconstructions

• standardize, average and inflate so variance
  matches target variance

• With correlation weights

• MBH (AD1000, AD1400 steps)

39
CPS with Correlation Weights One-Stage Partial
Least Squares
Bair et al 2004.
40
OLS and PLS

• OLS regression coefficients are a rotation (and
  dilation) of the PLS coefficients in
  coefficient-space

41
From the Frying Pan into the Fire

• In cases where there is little correlation
  between proxies, then the rotation matrix is
  near-orthogonal and PLS increasingly
  approximates OLS

Phillips 1998 Figure 4. n x2 in interval pi to
pi, repeated periodically. The regression using
1000 observations and 125 white noise regressors
42
Wahl and Ammanns no PC case a near-OLS
regression 79 years long against 65-90
predictors. Calibration residuals are
meaningless.
Magenta WA Black - Two synthetic HS series
plus 68 red noise series. Statistical pattern is
identical to MBH under WA variation high RE,
high calibration r2 0 verification r2 negative
CE.. In deg C
43
If you insert synthetic HS series plus low-order
red noise, you get recons that look like the
no-PC recons
                                        Magenta
WA Black - Two synthetic HS series plus 68 red
noise series. Statistical pattern is identical to
MBH under WA variation high RE, high calibration
r2 0 verification r2 negative CE. Similar
examples used in Reply to Huybers. In deg C
44
You can get the same verification results using
tech stocks instead of the Bristlecone PC1Gaspe.
In deg C
45
Ridge Regression is not a magic bullet

• Stone and Brooks 1990 Continuum Regression
  Ridge regression coefficients can be arranged as
  a 1-parameter continuum between OLS and PLS

46
Borga et al 2000 Taxonomy

• Solutions to
  for cases below

47
Two-parameter mixing

• You can do one-parameter mixing of and
  identity matrix to get to CCA (Canonical
  Correspondence Analysis).

48
OLS gets cute with coefficients with values all
over the place. In this network, simple is better.
49
How do you test reconstructions the problem of
spurious regression?Is there a valid
statistical relationship between climate field
(the temperature PC1) and bristlecones?
50
Yule 1926 - this is RE-resistant
Mortality per 1000 (points) and proportion of
Church of England marriages per 1000 marriages
(line)
51
Hendrys Theory of Inflation 1980
Hendrys theory of inflation is that a certain
variable (of great interest in this country) is
the real cause of rising prices. .. there is a
good fit, the coefficients are significant,
but autocorrelation remains and the equation
predicts badly. Assuming a first order
autoregressive error process, the fit is
spectacular, the parameters are highly
significant, there is no obvious residual
autocorrelation (on an eyeball test and the
predictive test does not reject the model. C is
simply cumulative rainfall in the UK. It is
meaningless to talk about confirming theories
when spurious results are so easily obtained. 
Doubtless some equations extant in econometric
folklore are little less spurious than those I
have presented.
52
Granger and Newbold 1974
It is very common to see reported in applied
econometric literature time series regression
equations with an apparently high degree of fit,
as measured by the coefficient of multiple
correlation R2 but with an extremely low value
for the Durbin-Watson statistic. We find it very
curious that whereas every textbook on
econometric methodology contains explicit
warnings of the dangers of autocorrelated errors,
this phenomenon crops up so frequently in
well-respected applied work....   It has been
well known for some time now that if one performs
a regression and finds the residual series is
strongly autocorrelated, then there are serious
problems in interpreting the coefficients of the
equation. Despite this, many papers still appear
with equations having such symptoms and these
equations are presented as though they have some
worth. It is possible that earlier warnings have
been stated insufficiently strongly. From our own
studies we would conclude that if a regression
equation relating economic variables is found to
have strongly autocorrelated residuals,
equivalent to a low Durbin-Watson value, the only
conclusion that can be reached is that the
equation is mis-specified, whatever the value of
R2 observed.
53
Canonical reconstructions fail
All multiproxy reconstructions, except MBH99,
fail Durbin-Watson statistic (minimum 1.5).
Passing a DW test is a necessary but not
sufficient test of model validity.
54
Ferson et al 2003
Data mining for predictor variables proxies
interacts with spurious regression bias. The two
effects reinforce each other because more highly
persistent series are more likely to be found
significant in the search for predictor
variables. Our simulations suggest that many of
the regressions in the literature, based on
individual predictor variables, may be
spurious The pattern of evidence in the
instruments in the literature is similar to what
is expected under a spurious mining process with
an underlying persistent expected return. In this
case, we would expect instruments to arise, then
fail to work out of sample.
55
Greene et al 2000

• From this perspective data-mining refers to
  invalid statistical testing as a result of naive
  over-use of a sample. In particular, the use of a
  sample both for learning-inspiration and for
  testing of that which was learned or mined from
  the sample. Any test of a theory or model is
  corrupted if the test is conducted using data
  which overlaps that of any previous empirical
  study used to suggest that theory or model. The
  moral is clear.

56
The same proxies are used over and over again.
Two problems severe data mining renders
statistical testing meaningless lack of
independence between data sets makes multiple
reconstructions vulnerable to data problems.
57
IPCC Box 6.4 Figure 1
The most stylized and repetitively data mined
series are shown in IPCC AR4 Box 6.4 Figure 1.
Manns incorrectly calculated PC1 is even shown.
58
Greene et al 2000 3

• But testing in un-mined data sets is a difficult
  standard to meet only to the extent one is
  impatient. There is a simple and honest way to
  avoid invalid testing. To be specific, suppose in
  1980 one surveys the literature on money demand
  and decides the models could be improved. File
  the proposed improvement away until 2010 and test
  the new model over data with a starting date of
  1981.. Only new data represents a new experiment.
  I do not consider this a pessimistic outlook.
  This is because I thinks much can be learned from
  exploring a sample. Patience and slow methodical
  progress are virtuous.

59
Bring the Proxies Up to Date

• Michael Mann paleoclimatologists are
  attempting to update many important proxy records
  to the present, this is a costly, and
  labor-intensive activity, often requiring
  expensive field campaigns that involve traveling
  with heavy equipment to difficult-to-reach
  locations (such as high-elevation or remote polar
  sites). For historical reasons, many of the
  important records were obtained in the 1970s and
  1980s and have yet to be updated.  

60
The Divergence Problem
The graphs below show results from the only
large-population (387 sites) survey
Schweingruber sites chosen ex ante as temperature
sensitive.
Left - Briffa et al 2001 reconstruction (left)
from 387 temperature-sensitive sites right
from Briffa et al 1998 heavy solid MXD (used
in Briffa et al 2001) dashed RW thin solid
temperature.
61
Briffas Cargo Cult
Briffa et al In the absence of a substantiated
explanation for the decline, we make the
assumption that it is likely to be a response to
some kind of recent anthropogenic forcing. On the
basis of this assumption, the pre-twentieth
century part of the reconstructions can be
considered to be free from similar events and
thus accurately represent past temperature
variability.
62
Briffa et al 2001 was an IPCC TAR reconstruction,
but spaghetti graph does not show it reverting to
early 19th century levels.

• IPCC truncated the Briffa et al 2001
  reconstruction (green) in 1960. Thus no visible
  divergence. Also truncated in AR4 (pale blue).

63
Divergence Problem NAS and IPCC
NAS Panel Cook et al. (2004), who subdivided
long tree ring records for the Northern
Hemisphere into latitudinal bands, and found
that divergence is unique to areas north of
55N, IPCC AR4 divergence is apparently
restricted to some northern, high-latitude
regions, but it is certainly not ubiquitous even
therethe possibility of investigating these
issues further a limit on the potential to
reconstruct possible warm periods in earlier
times is restricted by the lack of recent tree
ring data at most of the sites from which tree
ring data discussed in this chapter were
acquired. (p. 473)
64
Updating Almagre the Starbucks Hypothesis
Almagre CO (about 35 miles west of Colorado
Springs CO) is a bristlecone pine site with a
Graybill chronology going back to AD1000. We took
64 cores (36 trees), of which 38 cores (20 trees)
at or near Graybill site. Highest tree ring
millennium chronology in the world!
http//picasaweb.google.com/Almagre.Bristlecones.2
007/
65
Exact Graybill Trees
We located 16 tagged trees of which 8 have been
sampled. We reconciled the tags to the ITRDB
archive (before co-operation ceased). Only 3 of
the 8 sampled trees had been archived. It appears
that Graybill sampled 42 trees, of which only 21
are archived. Reasons for selective archiving
unknown at present.
66
Very Low Recent Growth in Many Trees
In mm/100.
67
Strip Bark Trees
NAS panel said that strip bark trees should be
avoided. However, this information not recorded
in bristlecone and foxtail (or other) archives.
Graybill said that he sought out strip bark. (In
mm/100)
68
Bizarre Strip Bark Forms
Brunstein, C. USGS.
69
Updated Almagre Chronology
Decline in recent ring widths which are
obviously not at levels teleconnecting with
high NH temperatures. 1840-50s are a very loud
phenomenon in chronology. In dimensionless
chronology units, basis 1.
70
Strip Bark At Graumlich Sites
From the plot below, I asked Andrea Lloyd whether
this tree, with the characteristic discrepancy
between cores, was strip bark. She consulted her
field notes the answer was yes. (In mm/100)
71
White Mts very arid
72
Ababneh Sheep Mt Update
Left - Graybill 1987 and Ababneh 2006 (PhD
thesis) for Sheep Mountain (Chronology units
minus 1). Right Both re-scaled on 1902-1980
showing that Graybill dilates relative to Ababneh
after 1840s. (s.d. units) Inset Weights of Mann
and Jones 2003 PC1, showing Sheep Mt dominance.
73
More Low Latitude Divergence
A statistician, considering this as an out of
sample test of the hypothesis that there is a
linear relationship between ring width and
temperature, would conclude that this refutes the
hypothesis. In the literature, this is referred
to as the Divergence Problem. (In sd units)
74
Miller et al 2006
Deadwood tree stems scattered above treeline on
tephra-covered slopes of Whitewing Mtn (3051 m)
and San Joaquin Ridge (3122 m) show evidence of
being killed in an eruption from adjacent Glass
Creek Vent, Inyo Craters. Using tree-ring
methods, we dated deadwood to 815-1350 CE, and
infer from death dates that the eruption occurred
in late summer 1350 CE. Using contemporary
distributions of the species, we modeled
paleoclimate during the time of sympatry the
MWP to be significantly warmer (3.2 deg C
annual minimum temperature) and slightly drier
(-24 mm annual precipitation) than present,
75
Tornetrask Grudd 2008 is drastic revision of
earlier chronologies. Questioned standardization
methods of prior studies.
Grudd Fig. 12 The thick blue curve is the new
Tornetrask MXD lowfrequency reconstruction of
AprilAugust temperatures, with a 95 confidence
interval (grey shading) adopted from Fig. 5. The
thin red curve is from Briffa et al. (1992) the
hatched curve is from Grudd et al. (2002) and
based on TRW.
76
Polar Urals
Left red- Briffa et al (Nature 1995) reported
that 1032 was the coldest year of the
millennium. 11th century was based on only a
couple of poorly dated cores. New tranche of
data (green -1998 in Esper et al 2002) showed
very warm 11th century but Briffa never
published this. Instead he did his own analysis
on Yamal site (right red ) about 100 miles away.
Like Sheep Mt, dilation of post-mid 19th century
results in Briffa version. Briffas Yamal version
used in all but one subsequent study without any
attempt to reconcile to Polar Urals update. (In
sd units)
77
Shiyatov and Naurzbaev

• Shiyatov 1995 From the middle of the 8th to the
  end of the 13th, there was intense regeneration
  of larch and the timberline rose up to 340 a.s.l.
  The 12th and 13th centuries were most favorable
  for larch growth. At this time the altitudinal
  position of the timberline was the highest, stand
  density the biggest, longevity of trees the
  longest, size of trees the largest, increment in
  diameter and height the most intensive as
  compared with other periods under review
• Naurzbaev et al 2004 Trees that lived at the
  upper (elevational) tree limit during the
  so-called Medieval Warm Epoch (from A.D. 900 to
  1200) show annual and summer temperature warmer
  by 1.5 and 2.3 deg C, respectively, approximately
  one standard deviation of modern temperature.
  Note that these trees grew 150-200 m higher
  (1-1.2 deg C cooler) than those at low elevation
  but the same latitude, implying that this may be
  an underestimate of the actual temperature
  difference.

78
dO18 in Ice Cores no global pattern

• Guliya (3rd) is supposedly evidence of global
  warming, while Mount Logan attributed to
  regional circulation changes
• NAS Panel no Antarctic sites show medieval
  warming ahem, what about Law Dome?

All centered on mean.
79
Dunde Ice Core versions are inconsistent and
sample data unarchived
How can interannual climate be calculated when
one ice core yields a spaghetti graph? The
purple series strongly influences the Yang
composite, one of the extreme IPCC Box 6.4
series.
80
Question is there a rational way of choosing (a)
or (b)?

• Left NRC panel. Right Variations of standard
  reconstructions using Polar Urals update instead
  of Yamal and Sargasso Sea SST instead of G
  Bulloides wind speed proxy and Yakutia instead of
  problematic bristlecones/foxtails

81
Econometric References 
 
http//www.climateaudit.org/?page_id2709
82
END