159 bottles of beer on the wall....#3.5 of auditing the audit
Josh Halpern
Thought I would take a night off, but playing with pcproxy and the ../pub/MBH98 site last night, I realized where the answer that NP wanted in concrete lay. Now this is, I believe, strictly speaking not a question that C2300 asks, but it is related to question 3, so we will treat it as question 3.5
If you are a newcomer to this series of auditing posts I use several abbreviations which are explained at the bottom under my signature.
On 15 Dec 2003 Nigel Persaud) wrote: SNIP....
>BTW have you figured out the 159 series yet? I guess not or you'd
> have told us. Why don't you finish that before you start something else?
Ooo an assignment from Nigel. He has repeated this point very often and one suspects he hopes there is not an answer or that at least none of us can figure it out so he can keep this piece of trivia going as a brick with which to beat folk. I did provide a big fat hint at the end of the Auditing the audit #3 post for him, but he did not notice.. Had not quite worked all the numbers out, but I saw what Mann was talking about. To find the answer you need pcproxy (hopefully reduced to a spreadsheet for your sanity), and text file backto1820.txt, which were sent to M&M by Mann's UVa group and can be found on C2003, access to the MBH98 web site at holocene, and the NSM Remember that MBH98 (see, for example ftp://eclogite.geo.umass.edu/pub/mann/) the multiproxy network includes 81 individual series (ie series that have not been reduced to sets of principal components, see backto 1820.txt for a listing). In addition there were 25 series from Jones and Bradley that were used to calibrate the network of which 22 are used according to the NSM. Which accounts for 103 files. So if 159 series were needed for the fit, principal component series were used. As we did to answer M&Ms question 3, we compare the pc0x.out files in each of the subdirectories of ../pub/MBH98/.... called /BACKTO_XXXX (see auditing the audit #3 for a detailed discussion) with the listings in pcproxy. From this we can count the number of files that were used (assuming that pcproxy did contain all of the principal component series that were used. This is a useful caution, because the real issue is MBH98 as published in Nature, not pcproxy, which was compiled by a (graduate student?/postdoc?) from Mann's UVa group five years later, not the /holocene, and not the /eclogite ftp archives. Secondly, we will see that pcproxy does contain all of the necessary series, though in a less than obvious way, and I believe with two errors besides the offset errors we discussed in #3) We get: 3 from the Stahle OK tree rings (Col 69-71) from /BACKTO_1700 1 from the Stanle SWM tree rings (Col 72-80) from /1400 1 from the Stanle SWM tree rings (Col 72-80) from /1450 1(2) from the Stanle SWM tree rings (Col 72-80) from /1500 * 4 from the Stanle SWM tree rings (Col 72-80) from /1600 9 from the Stanle SWM tree rings (Col 72-80) from /1700 1 from the Vaganov tree rings (Col 81-83) from /1450 2 from the Vaganov tree rings (Col 81-83) from /1600 3 from the Vaganov tree rings (Col 81-83) from /1760 2 from the ITRDB/NoAm tree rings (Col 84-92) from /1400 4(2) from the ITRDB/NoAm tree rings (Col 84-92) from /1450 ** 7 from the ITRDB/NoAm tree rings (Col 84-92) from /1600 9 from the ITRDB/NoAm tree rings (Col 84-92) from /1700 2 from the ITRDB/SoAm tree rings (Col 93-95) from /1600 3 from the ITRDB/SoAm tree rings (Col 93-95) from /1750 4 from the ITRDB/Aust tree rings (Col 96-99) from /1750 56 (see below) ** for the ITRDB/NoAm data pcproxy lists the first four components at 1500 from the /1450 pc folder but does not include pc03 and pc04 between 1450 and 1499. * for the Stahle SWM data pcproxy lists the first two principal components from the folder /1500 starting at 1500. MBH98 divide the 600 year period into 1820 - present, 1760-1819, 1700-1759, 1600-1699, 1450-1599 and 1400-1449. While various series start in the middle of various periods, one should not be able to use a series that is incomplete over a period. For example if a series started in 1650, it would not contribute to the 1600-1699 period, but would first contribute to the 1700-1759 period. Therefore, I conclude that the second component in the Stahle SWM series, since it starts in 1500 should not be used. Indeed, it is my understanding that the /1500 values should not have been copied over into the pcproxy, but the pc01.out from /1450 should have extended to 1599. OTOH, the first four pcs from the /1450 ITRDB/NoAm data should have been posted to pcproxy, or at least the number of pcs should be the same from 1450 to 1600. And, we see that 159 separate series are needed as Mann said. OTOH, we see that essentially all of these series are contained in pcproxy although you have to use ../pub/MBH98 to figure this out. So Nigel and Steve, over to you. josh halpern *******************************
The issues of C2300 refer to the file pcproxy and the paper MBH98 as published in Nature, we will call that MBH98 and take care to differentiate it from the ftp site /pub/MBH98. We will call the published supplemental materials NSM.
Nigel Persaud
ingenuity. Anyway here are the problems with your re-construction
below:
1) you are relying on pcproxy. This undoubtedly provides evidence on
the matter, but Mann has deleted it so you are not relying on data at
Mann's FTP site.
2) the number 159 is not directly derivable from Mann's FTP site, but
only exists in comments made in 2003 in response to M&M;
3) the 22 Jones-Bradley instrumental series are included in the 81
non-PC series. So you've double-counted them in getting to 159.
4) your proxy totals do not balance to disclosure in Nature: you have
27 proxies in 1450 vs 24 in Nature; 54 in 1600 vs 57 in Nature; 75 in
1700 vs 74 in Nature.
It's more contradictory than you think. Back to you, Josh.
Josh Halpern <j.ha...@incoming.verizon.net> wrote in message news:<dtRDb.11157$G9....@nwrddc01.gnilink.net>...
> --
Joshua Halpern
Jones and Bradley were columns 21 through 31 and the 11 precipitation
records used were 32 through 42, which means that there are only 137
separate records in pcproxy, not 159, unless some of these are
composite pc reconstructions. Apologies.
Dig dig.
I saw this from the exchange between David Ball and Nigel Prasaud.
josh halpern
Josh Halpern <j.ha...@incoming.verizon.net> wrote in message news:<dtRDb.11157$G9....@nwrddc01.gnilink.net>...
> --
Joshua Halpern
Jones and Bradley were columns 21 through 31 and the 11 precipitation
records used were 32 through 42, which means that there are only 137
separate records in pcproxy, not 159, unless some of these are
composite pc reconstructions. Apologies.
Dig dig.
I saw this from the exchange between David Ball and Nigel Prasaud.
josh halpern
Josh Halpern <j.ha...@incoming.verizon.net> wrote in message news:<dtRDb.11157$G9....@nwrddc01.gnilink.net>...
> --
Nigel Persaud
reconstructions.
Steve, so the great David Ball joins Josh Halpern in not being able to
get to 159 series. I guess he'll blame M&M for that.
Nigel
----== Posted via Newsfeed.Com - Unlimited-Uncensored-Secure Usenet News==----
http://www.newsfeed.com The #1 Newsgroup Service in the World! >100,000 Newsgroups
---= 19 East/West-Coast Specialized Servers - Total Privacy via Encryption =---
Steve Schulin
pers...@yahoo-dot-com.no-spam.invalid (Nigel Persaud) wrote:
> BTW, I will require a unique 146 series to do my
> reconstructions.
>
>
> Steve, so the great David Ball joins Josh Halpern in not being able to
> get to 159 series. I guess he'll blame M&M for that.
Nah. He wouldn't do that. My understanding is he could not care less how
Mann et al 1998 analyzed the data. In fact, the more blind his efforts
are, the better.
But I do want to thank you for the info about Jones et al using the
early years of Central England data in the 1998 Holocene paper, and the
earlier appearance of the same truncated and summer-only data in Bradley
and Jones 1993 treatment.
Regards,
Steve Schulin