Diana et al,
Thanks for providing assistance with respect to the available annual data (1960-2008) on CO2 and Temperature for Mauna Loa.
http://news.bbc.co.uk/1/hi/in_depth/sci_tech/2009/copenhagen/8393855.stm
You and others have asked me to develop a forecast with stated uncertainties. The temperature readings/values appear to have a statistically significant lag structure (polynomial distributed lag or autoregressive distributed lag ) on CO2 for up to 5 years. Any forecast of temperature would be dependent upon a forecast for CO2 since there is a contemporaneous component to the model. Please see
http://www.autobox.com/co2.zip
for the complete analysis using standard pre-whitening and transfer function model revision strategies. Note that the mean-adjusted temperatures were used rather than the smoothed values as anomalous readings in either variable were automatically identified via Intervention Detection schemes ( N.B. these were period 17,39,49,14,15,33,26,and 37 where period 1 is 1960 ). A robust regression equation obtained via Transfer Function identification strategies is:
Y(T) = -3.4607
+[X1(T)][(+ .0108- .0009B** 4+ .0004B** 5)] CO2
+[X2(T)][(- .211)] :PULSE 17
+[X3(T)][(+ .233)] :PULSE 39
+[X4(T)][(- .180)] :PULSE 49
+[X5(T)][(+ .145)] :PULSE 14
+[X6(T)][(- .151)] :PULSE 15
+[X7(T)][(- .145)] :PULSE 33
+[X8(T)][(- .137)] :PULSE 26
+[X9(T)][(- .135)] :PULSE 37
+ [A(T)]
Note variables X2-X9 are the identified anomalous periods
The notation B**L denotes a lag of L
The model generated a noise series that met the requirements of the general Linear Model. These results can be confirmed with any MLR package by incorporating the lag of CO2 by 4 and 5 periods and the 8 indicator variables
Since an outlier was detected at period 49 (the year 2008), this suggests that the last observed temperature was statistically significantly lower than expected which might foreshadow an actual downtrend in temperature. We wait expectantly for the 2009 value to confirm this "good news". Note the observed value for 2008 was .33 some .180 statistically significant less than the expected value of .51 ( alpha of .005 ).
In terms of a forecast for the year 2009 , we find that the C02 is expected to be
TIME DATE LOWER 80% UPPER 80% FORECAST
(T) LIMIT LIMIT
50 2009 386.4 387.5 386.9
Yielding a forecast of temperature of
TIME DATE LOWER 80% UPPER 80% FORECAST
(T) LIMIT LIMIT
50 2009 .4373 .6052 .5213
Diana, you asked and you received. Recall the adage attributed to Box, that “Remember that all models are wrong; the practical question is how wrong do they have to be to not be useful”
Regards
Dave R
Dave Reilly
Senior Vice President
Automatic Forecasting Systems
www.autobox.com
215-675-0652 (office)
215-353-7087 (cell)
-----Original Message-----
From: A UK-based worldwide e-mail broadcast system mailing list [mailto:[log in to unmask]] On Behalf Of kornbrot
Sent: Monday, December 28, 2009 11:33 AM
To: [log in to unmask]
Subject: Re: SUMMARY: Climate Change: Where are the statisticians?
An apology for overenthusiastically sharing, what I saw when looking at the
climate data provided on BBC web site.
I am, in fact, well aware of the flaws in such analyses, as so clearly
summarized by David Reilly.
Nevertheless plots of standardised co2 and temp against time, show striking
variability from one subset of 16kyears to the next. I was surprised at the
magnitude of the variability and so changed my view
A a good Bayesian, on the basis of the data, I went from a low confidence
acceptance to low confidence rejection of human Co2 emissions as a
contributory cause to global warming.
What should be worrying to allstat members is that predictions of
temperature rise, with different levels of co2 emissions are unadorned by
any estimates of accuracy of any kind. Athough the climatologists and their
statisticians are undoubtedly able to estimate accuracy of prediction
Please email me off list if you want to see graphs, or have links to other
data sources.
Will not trouble this list further. Again apologies if you are offended.
Best
Diana
On 20/12/2009 18:08, "David Reilly" <[log in to unmask]> wrote:
> I am shocked and stunned at the notion that you feel comfortable with
> changing your beliefs as a result of " Just simple regression, no
> sophisticated ARIMA" . Are you not aware that the significance tests you
> report REQUIRE that the errors from the model have a mean of "near zero" for
> all subsets of time and that the errors have to be proven to not have any
> autoprojective structure and that the parameters for the model are invariant
> over time and that the errors have constant variance. (apologies for a
> run-on sentence ! )
>
> I don't practice heart surgery and non-statisticians shouldn't attempt to
> practice statistics unless they are accompanied by intelligent colleagues or
> intelligent software acting on behalf of the absent intelligent colleagues.
>
>
> Dave Reilly
> Senior Vice President
> Automatic Forecasting Systems
> www.autobox.com
> 215-675-0652 (office)
> 215-353-7087 (cell)
>
>
> -----Original Message-----
> From: A UK-based worldwide e-mail broadcast system mailing list
> [mailto:[log in to unmask]] On Behalf Of kornbrot
> Sent: Sunday, December 20, 2009 12:22 PM
> To: [log in to unmask]
> Subject: Re: SUMMARY: Climate Change: Where are the statisticians?
>
> Hi Nigel
> Thanks for this very interesting summary and the links
> Im the airhead who changed her mind. Regret to tell you Ive now changed it
> back again.
> Why?
> Analyzed the 800k data from BBC. Separated into 50 consecutive EONs of 16k
> years each. Then regressed temperature on CO2. Just simple regression, no
> sophisticated ARIMA
> Results at 95% CL: 38 significant positive, 7 ns positive, 4 ns negative, 1
> negative p = .051
> So there is 12/50 = 24% chance that we are living in a 16k yr period where
> there is no association between temperature and CO2. The graphs are a mess
> too, [email me if you would like to see]
>
> So I am a sceptic about limiting carbon emissions, BUT
> The most serious issue, that needs addressing right now is DEFORESTATION
> and human action can influence events
> Its most serious, because its most IRREVERSIBLE. We can clean the smoke
> stacks this year or next. If we over fish this year, we can under fish next.
> If we substitute bio-fuel for food this year, we can abut face next.
> Rebuilding forests takes decades, if not centuries.
>
> Hand it to GreenPeace, its top of their agenda.
> http://www.independent.co.uk/environment/climate-change/historic-chance-to-h
> alt-the-scourge-of-deforestation-1809418.html,
> http://www.greenpeace.org/international/campaigns/climate-change/forests_for
> _climate.
>
> Deforestation doesnt seem to be part of the US-China accord. Commercial
> interests in Brazil & Indonesia?
>
> Best
>
> Diana
>
> On 20/12/2009 16:30, "Nigel Marriott" <[log in to unmask]> wrote:
>
>> > Dear Allstatters,
>> >
>> >
>> >
>> > As promised here is my summary of the responses I received. The original
>> > email I sent is attached at the bottom. In total I only received 6
>> > responses (7 if you include me) to the questions I posed. In addition,
>> > there were 6 other responders who sent comments but who did not answer the
>> > questions. I have also decided to include the 3 responses to Anatoly
>> > Zhigljavsky's query on temperature records that went to the ALLSTAT list
>> > instead of direct to Anatoly himself.
>> >
>> >
>> >
>> > My main reason for posting this query was to find out the level of
>> > involvement & interest of statisticians in the climate debate. Although
> no
>> > meaningful conclusions can be drawn from the responses to my questions, I
> am
>> > going to contrast the total number of responses (12 or 15 if you include
>> > Anatoly) I received with other questions I have posed in the past.
> Looking
>> > through my email folders, I received 6 responses to an SPC query, 11
>> > responses to a classification analysis query and 8 responses to a variant
> of
>> > the Monty Hall problem (the summary was posted on my website which
> received
>> > over 200 visits). The total number of responders to this query was higher
>> > than before but given how topical climate change is at present and its
>> > importance in world affairs, I am disappointed that the response was not
>> > significantly higher than what I received for quite technical queries. If
> I
>> > assume that the ALLSTAT list is a representative sample of statisticians
>> > around the world (perhaps with a bias to English-speakers & academics), I
>> > think the answer to the question in the title in my query ("Where are the
>> > statisticians?") is that they aren't there and they are not involved in
>> > climate science to the extent they should be. Given the importance of the
>> > debate and the inherent statistical nature of the science I find that
>> > extremely worrying.
>> >
>> >
>> >
>> > The 7 responses to my questions can be summarised as follows:
>> >
>> >
>> >
>> > 1. Before reading the above links, how knowledgeable do you think
> you
>> > are about the science and statistics behind the global warming debate?
>> > Please respond on a scale from 1 to 9 where 1 is no knowledge at all, 9 is
>> > you know everything:- RANGE FROM 1 to 7, MEDIAN & MODE WAS 3.
>> >
>> > 2. Do you work in the field of climate research in any way? NOBODY
>> > WORKED IN CLIMATE RESEARCH
>> >
>> > 3. Prior to reading the 3 links below, can quantify your prior
> belief
>> > in the relative merits of the warmist/sceptic views? Happy for you to
>> > choose your own way of quantifying this but as an example, I would have
> said
>> > my prior belief was 60:40 in favour of the sceptic camp. 4 FAVOURED THE
>> > SCEPTIC VIEW (FROM 60:40 TO 99:1 IN FAVOUR), 3 FAVOURED THE WARMIST VIEW
>> > (FROM 60:40 TO 90:10 IN FAVOUR). JUST TO BE CLEAR, SCEPTIC MEANS
> SCEPTICAL
>> > ABOUT THE EXTENT OF HUMAN INFLUENCE ON CLIMATE, NOT ABOUT WHETHER THE
>> > CLIMATE IS CHANGING IS CHANGING IN THE FIRST PLACE.
>> >
>> > 4. On reading the 3 links below and any others you find, has your
>> > prior belief changed? If so, how would you now quantify your posterior
>> > belief in the light of what you have read? ONLY 1 PERSON CHANGED THEIR
> PRIOR
>> > BELIEF BUT IT WAS A SMALL CHANGE FROM 65:35 SCEPTIC TO 55:45 SCEPTIC. THE
>> > OTHER 6 POSTERIOR BELIEFS WERE UNCHANGED FROM THEIR PRIORS.
>> >
>> > 5. If your posterior belief is little changed from your prior, what
> do
>> > you think is missing from the debate that would help you change your
> belief
>> > either way? MANY COMMENTS RECEIVED WHICH I AM SURE WOULD BE PART OF ANY
>> > STATISTICAL DEBATE. MAIN ONES LISTED BELOW.
>> >
>> > . 3 mentioned lack of availability of the raw data as opposed to
> the
>> > summarised data currently available.
>> >
>> > . 2 wanted to see an examination of the strengths & weaknesses of
>> > the measurements involved including recent temperature records, CO2
>> > emissions, historical temperature reconstructions.
>> >
>> > . 1 wanted a review of the statistical models used and the
>> > assumptions that have been made and whether they were able to separate
>> > common cause variation from special cause variation.
>> >
>> > . 1 wanted greater clarity of whether a temperature series was
>> > derived from global data or local data.
>> >
>> > . 1 had concerns about the quality and validity of the peer review
>> > process.
>> >
>> > . 1 wanted more encouragement for free debate of all views &
>> > expressions.
>> >
>> > 6. Should statisticians be more involved in the debate? If so, how
>> > should this happen? AGAIN MANY COMMENTS, MAIN ONES WERE.
>> >
>> > . Climate science papers should have statistical referees.
>> >
>> > . RSS should encourage media to check quality of climate data with
>> > statisticians.
>> >
>> > . Statisticians should be part of multi-disciplinary climate
>> > research teams.
>> >
>> > . RSS to set up a working party to pull together all climate data
>> > and assess its quality.
>> >
>> > . RSS should be more willing to speak out when bad data &
> modelling
>> > techniques are being used.
>> >
>> > . Being a statistician doesn't automatically make you an expert on
>> > climate science.
>> >
>> >
>> >
>> > If I couple the responses to question 6 with the other comments I
> received,
>> > I was left with the impression that nearly everyone felt statisticians did
>> > need to be more involved in the whole debate, particularly when it came to
>> > passing comment on the data & modelling techniques being used. As one
>> > responder put it, "being pure-minded methodologists, statisticians are not
>> > naturally attracted to policy debates like the one on climate change".
>> > Another stated "I have often been "too busy" to participate in these kind
> of
>> > dialogues. Perhaps it is time that I got "unbusy" and deliver some time
>> > series expertise to the subject at hand." I think it is quite noticeable
>> > that whilst many professional bodies have taken a view on anthropogenic
>> > climate change, the statistical bodies such as the RSS have not done so.
> I
>> > don't think that the RSS should have a view on the science involved but I
> do
>> > think it is in a unique position to pass comment on the data and modelling
>> > methodology. For example, we all see the following kind of statistical
>> > errors being made by both warmist & sceptic camps in the media and
> elsewhere
>> > and it should be quite easy for us to speak out and criticise instances
> such
>> > as these:
>> >
>> >
>> >
>> > . Too short a time span e.g. "the world has cooled for the last 10
>> > years therefore global warming is a myth" .
>> >
>> > . Inferring global effects from local data e.g. "Mount Kilimanjaro
>> > glacier has been receding for decades, we must do something!"
>> >
>> > . Using extremes to infer cause & effect e.g. worst flood on
> record,
>> > clearly due to mankind activities.
>> >
>> > . Confusing probability with possibility when making forecasts. I
>> > have seen confidence intervals given which were actually the ranges of
>> > expected outcomes from a number of different scenarios as opposed to being
>> > based on the prediction error from a single scenario.
>> >
>> >
>> >
>> > Many of you forwarded additional links to a variety of topics related to
>> > this query. Rather than make this summary even bigger, I will send these
> in
>> > a separate email to the list.
>> >
>> >
>> >
>> > Finally, can I thank the list for being far better behaved than what I saw
>> > in the links I sent through. Every reply bar two was courteous and
>> > thoughtful which is what I would expect from my fellow statisticians.
>> > Hopefully we can set an example to others on how to conduct a civilised
>> > debate!
>> >
>> >
>> >
>> > Regards
>> >
>> >
>> >
>> > Nigel Marriott
>> >
>> > Chartered Statistician
>> >
>> > <http://www.marriott-stats.com/> www.marriott-stats.com
>> >
>> >
>> >
>> > Ground Floor, 21 Marlborough Buildings, Bath BA1 2LY, United Kingdom
>> >
>> > Tel (office) +44 (0)1225 489033
>> >
>> > Fax +44 (0)870 6221969
>> >
>> >
>> >
>> > Marriott Statistical Consulting Limited, Company No. 5577275, VAT No.
>> > 883304029
>> >
>> > Registered in England, Registered Office - Equity House, 4-6 School
> Road,
>> > Tilehurst, Reading, RG31 5AL
>> >
>> >
>> >
>> > From: Nigel Marriott [mailto:[log in to unmask]]
>> > Sent: 07 December 2009 12:20
>> > To: ALLSTAT ([log in to unmask])
>> > Subject: Climate Change: Where are the statisticians?
>> >
>> >
>> >
>> > Dear Allstatters,
>> >
>> >
>> >
>> > First of all, can I thank Anatoly Zhigljavsky for having the courage to
> open
>> > the debate on global warming on ALLSTAT. Although this is not an open
> forum
>> > and I do not want people to breach the ALLSTAT guidelines, I would be very
>> > interested to hear your responses to my questions. With the Copenhagen
>> > conference taking place, it is a topical subject and I am very happy to
> take
>> > responsibility for preparing a summary of the responses I receive once the
>> > conference itself is over which is next Friday I believe.
>> >
>> >
>> >
>> > Some contributors have already mentioned that Anatoly's argument was based
>> > on only 30 years or so which I agree is not long enough for any definitive
>> > conclusions. Personally, I believe the debate needs to start with whether
>> > today's climate is significantly different from the past. For me, the
> past
>> > has to include the years before industrialisation so as to avoid issues
>> > with confusing correlation with cause & effect which is a real risk if you
>> > only look at the last 100/150 years. To my mind, we need to look at the
>> > last 1,000 or 2,000 years. Unfortunately, since we have no thermometer
>> > temperature records going back before 1700 (the Central England
> Temperature
>> > series is I believe the longest such record), this requires the use of
> proxy
>> > measures such as tree rings, ice cores, coral reefs, etc which have been
>> > calibrated against thermometer records. If you reread this paragraph, you
>> > will see that what I have written is an inherently statistical subject
>> > involving time series analysis, multivariate analysis, modelling, sample
>> > design, etc. If we statisticians are not involved in this work, then the
>> > possibility of errors in both the sceptics & warmers camps are enormous.
>> > Yes I know scientists can be trained in statistical methods, but we all
> know
>> > that even then, they are capable of making statistical errors with
> profound
>> > consequences for us either way.
>> >
>> >
>> >
>> > After reading a number of blogs, articles & websites, my perspective is
> that
>> > the warmers & sceptics do not radically disagree over the temperature
> record
>> > for the last 100 years or so but they violently disagree over the last
> 1,000
>> > years. At the bottom of this email are 3 links which seems to capture the
>> > heart of this particular argument as far as I can see. Frankly, by the
> time
>> > I got to the 3rd link, I was heartily sick of personal vitriol being cast
> by
>> > both sides with each attacking each other's motives instead of the methods
>> > used. In particular, I am finding it hard to evaluate the statistical
> side
>> > of the argument and I am left with the belief that statisticians are
> simply
>> > not involved in the climate change debate to the degree we should be.
>> >
>> >
>> >
>> > So to find out what your views are, I would like to ask you the following
>> > questions. I am very happy to collate responses by the end of the
>> > conference and publish the results. I'm sure some of you will criticise
> my
>> > survey design (I've tried to be Bayesian here!) but I'm only trying to do
> an
>> > indicative survey here.
>> >
>> >
>> >
>> > 7. Before reading the above links, how knowledgeable do you think
> you
>> > are about the science and statistics behind the global warming debate?
>> > Please respond on a scale from 1 to 9 where 1 is no knowledge at all, 9 is
>> > you know everything.
>> >
>> > 8. Do you work in the field of climate research in any way?
> Basically
>> > a yes/no question, not interested in your life's CV!
>> >
>> > 9. Prior to reading the 3 links below, can quantify your prior
> belief
>> > in the relative merits of the warmist/sceptic views? Happy for you to
>> > choose your own way of quantifying this but as an example, I would have
> said
>> > my prior belief was 60:40 in favour of the sceptic camp.
>> >
>> > 10. On reading the 3 links below and any others you find, has your prior
>> > belief changed? If so, how would you now quantify your posterior belief
> in
>> > the light of what you have read?
>> >
>> > 11. If your posterior belief is little changed from your prior, what do
>> > you think is missing from the debate that would help you change your
> belief
>> > either way? Open question here.
>> >
>> > 12. Should statisticians be more involved in the debate? If so, how
>> > should this happen? This is an open question and I will do my best to
>> > summarise.
>> >
>> >
>> >
>> > Here are the 3 links I have found (among others) but no doubt you may wish
>> > to click through the other links listed.
>> >
>> >
>> >
>> > . The people at the centre of the ClimateGate storm include
> authors
>> > of a famous paper published in 1999 which has since become known as the
>> > hockey stick. It is a central piece of the IPCC report. You can find a
> link
>> > to an updated version published in 2008 this near the bottom of this BBC
>> > article
>> > http://news.bbc.co.uk/1/hi/in_depth/sci_tech/2009/copenhagen/8393855.stm .
>> > The authors are Mann, Zhang, Hughes et al but as you can see, links to a
>> > number of other sources are given as well.
>> >
>> > . The original hockey stick paper was criticised by Steve McIntyre
>> > as explained in this essay written by Chris Monckton
>> >
> http://scienceandpublicpolicy.org/images/stories/papers/monckton/monckton_wh
>> > at_hockey_stick.pdf
>> >
>> > . I have not found any paper directly written in response to
>> > McIntyre's criticisms but this blog on the real climate website seems to
> be
>> > the best so far
>> >
> http://www.realclimate.org/index.php/archives/2004/12/myths-vs-fact-regardin
>> > g-the-hockey-stick/
>> >
>> >
>> >
>> > Kind Regards
>> >
>> >
>> >
>> > Nigel Marriott
>> >
>> > Statistical Consultant
>> >
>> > <http://www.marriott-stats.com/> www.marriott-stats.com
>> >
>> >
>> >
>> > Ground Floor, 21 Marlborough Buildings, Bath BA1 2LY, United Kingdom
>> >
>> > Tel (office) +44 (0)1225 489033
>> >
>> > Fax +44 (0)870 6221969
>> >
>> >
>> >
>> > Marriott Statistical Consulting Limited, Company No. 5577275, VAT No.
>> > 883304029
>> >
>> > Registered in England, Registered Office - Equity House, 4-6 School
> Road,
>> > Tilehurst, Reading, RG31 5AL
>> >
>> >
>> >
>> >
>> > You may leave the list at any time by sending the command
>> >
>> > SIGNOFF allstat
>> >
>> > to [log in to unmask], leaving the subject line blank.
>
>
>
> Professor Diana Kornbrot
> email: [log in to unmask]
> web: http://web.me.com/kornbrot/KornbrotHome.html
> Work
> School of Psychology
> University of Hertfordshire
> College Lane, Hatfield, Hertfordshire AL10 9AB, UK
> voice: +44 (0) 170 728 4626
> fax: +44 (0) 170 728 5073
> Home
> 19 Elmhurst Avenue
> London N2 0LT, UK
> voice: +44 (0) 208 883 3657
> mobile: +44 (0) 796 890 2102
> fax: +44 (0) 870 706 4997
>
>
>
>
>
>
>
> You may leave the list at any time by sending the command
>
> SIGNOFF allstat
>
> to [log in to unmask], leaving the subject line blank.
>
Professor Diana Kornbrot
email: [log in to unmask]
web: http://web.me.com/kornbrot/KornbrotHome.html
Work
School of Psychology
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
voice: +44 (0) 170 728 4626
fax: +44 (0) 170 728 5073
Home
19 Elmhurst Avenue
London N2 0LT, UK
voice: +44 (0) 208 883 3657
mobile: +44 (0) 796 890 2102
fax: +44 (0) 870 706 4997
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|
