Mann 08 The Missing Data
Posted by Jeff Condon on September 15, 2008
On September 5, 08, on the Mann site for the creation of the hockey stick Michael Manns published data used to generate his latest hockey stick paper showed 1357 series. His final paper referenced only 1209 with no mention of why the original series weren’t used. I have obtained a copy of the original data by a link at Climate Audit and decided to plot it here.
First, a graph of all the proxies together.
Then I took an average of all the data rejected before use in the latest hockey stick.
I thought I found the MWP. The Yellow line is 100 year average of the first curve. But I noticed the two smooth blue arcs between 300 and 1000 years in the first graph. These are two series from Fisher so I plotted the trend without them.
Looking at this last graph, the MWP is also quite strong mwp in comparison with the data he actually used plotted on fig 4 for comparison.
Take the end slope of Fig 4 and the end of the graph peaks at about 0.4 while at 600 AD the max peak is at about 0.8. On Fig 3 the max peak at the recent end is aobut2.5 while the MWP range from 0 – 700 several times reaches about 0.7. If we remember that the 1900-2000 year period is scaled to match recent temperature it would result in a strong magnification fo the MWP. I’m not saying that is why this data was scrapped, I’m just pointing out what I noticed.
I wonder what percentage of this would pass the “calibration”.
If anyone wants a copy of the spreadsheet data for this graph add it in your comments and I will send it by email. Commenter emails are visible to me only so no need to write it for everyone.
Raven said
The problem with Mann’s methods is he takes proxies which may be nothing but noise and tries to extract a temperature signal from them. One must be careful not to fall into the same trap when critiquing his work. In this case I suspect the signals are noise but the fewer samples in the early period lead to a spurious positive result. I do not think this is temperature signal – if it was then everyone would be wrong because the MWP was supposed to have occurred from 900-1100.
Jeff Id said
Raven,
I think you’re pretty much on the money. I have taken the time to plot and sort the proxies hundreds of different ways. One thing is for sure, there is a rising trend at the end of the proxies. No matter how I sort them the most negative reasonable averages are so far flat.
Still, flat is a far cry from a hockey stick!
Demesure said
Yeah, spot on. Thanks for this clearest way to show the Mannomatic.
The graphs’ titles are fuzzy anyhow:
- For graph1, you said it’s “a graph of all the proxies together”. It’s inconsistent since the graph is for the 148 rejected proxies (1357-1209).
- For graph2, I presume it’s the average & smoothed average of the 148 rejected proxies ?
- “Complete”, why ?
Jeff Id said
Sorry about the fuzzyness. I just got tired and a bit lazy about the graphs. Pretty boring looking at this stuff.
The tile on graph two is Mann data scrapped complete. Every one of the scrapped proxies averaged together.
There aren’t many proxies in the MWP so I deleted the two big arcs (from fisher’s data) from the set visible in graph 1 and plotted the minus fisher graph 3.
Craig Loehle said
Jeff: are you plotting raw proxies or are these calibrated to temperature?
Stephen McIntyre said
The two “Fisher” graphs are actually borehole temperature estimates from Dahl-Jensen. They are listed as being used in Mann and Jones 2003, but I have been unable to determine how the Mann and Jones 2003 was calculated. It is even more mysterious than MBH right now. Coauthor Jones told me that he did not know how the weights were calculated.
Jeff Id said
The proxies are calibrated to standard deviation through the period specified.
i.e calc standard deviation and normalize series to that.
Demesure said
I notice that proxies of Chuine_2004 based on havest dates of grapes in Burgundi is included in the Mannomatic. It’s hockey schtic shaped but Doug Keenan has found it’s based on a faulty model, not to say the author’s refusal to publish the original data, a manner so typical of the Team.
And Chuine’s proxies are supposed to represent temperature only for April-August but it has not prevented them to be slipped into the Mann spaghetti machine.
Mann’s cuisine with Burgundi wines, what a waste.
Jim Lindgren said
1.Please email me the data.
2. Would you mind retitling each figure and indicating the number of proxies? ie:
n=1209 proxies (selected by Mann)
n=1357 proxies (all in database)
n=148 proxies (excluded by Mann)
n=146 proxies (excluded by Mann, with 2 Fisher proxies also deleted)
As a rule, charts should be comprehensible without additional text.
Jeff Id said
Email isn’t working. I will send the data tomorrow.
Robin Edwards said
Hello, Jeff,
I’ve done loads of work on Mann’s original data – 1400 series of 112 columns and up to 593 rows – using entirely different methods from those normally employed. My clear conclusion is that these data cannot possibly generate a hockey stick when handled in a rational straightforward way.
I would be most interested in the data you are working with. It would be a waste of time for me to go through all your heavy duty operations that result in the final parameter(s) you have displayed here, but I would be delighted to have a copy of your “final” data set(s) (preferably as a text file) so that I can import it easily into my rather maverick computer system – a 1994 30 mHz Acorn machine running under RISC OS – and using my own software. I would /of course/ email you with my results!
Hope to hear from you ere long.
Robin Edwards Bromsgrove UK