Comments on: Nic Lewis on Statistical errors in the Forest 2006 climate sensitivity study

By: Non-centring in the Forest 2006 study « Climate Audit

Non-centring in the Forest 2006 study « Climate Audit — Mon, 22 Apr 2013 16:24:05 +0000

[...] basic statistical errors in the F06 code, about which I wrote a detailed article at Climate Audit, here. But those errors could not have affected any unrelated studies. In this post, I want to focus on [...]

By: The Forest 2006 climate sensitivity study and misprocessing of data – an update | Climate Etc.

Thu, 11 Apr 2013 17:12:50 +0000

[...] Forest 2006 likelihood functions. I wrote a detailed article about these errors at Climate Audit, here. But it appears that even in combination with the surface diagnostic 9 month timing mismatch these [...]

By: Hugo M

Hugo M — Thu, 17 Jan 2013 11:35:23 +0000

John, a file of identical size and md5sum (983d3b3ec00c192a9abe32aef82cc1f3) can be found at h ttp://svante.mit.edu/research/IGSM/data/IGSM_1/GRL06_reproduce.tgz . The directory contains more files, possibly also a copy of the lost raw data?

 [TXT]  GRL06_plotfilelist.txt  10-Dec-2012 17:43    12K   
 [   ]  GRL06_rawdata.tgz       10-Dec-2012 17:33   8.4G   
 [   ]  GRL06_reproduce.tgz     01-Aug-2012 11:10   2.0G   
 [   ]  GRL06_reproduce2.tgz    03-Aug-2012 17:33   1.2G   
 [   ]  GRL11_reproduce.tgz     13-Nov-2012 16:53    98M   
 [DIR]  GSOLSV/                 03-Jun-2010 13:57    -    
 [   ]  curry_z4_ctl.tar        08-Aug-2012 15:17    60K

By: John Gilmore

John Gilmore — Thu, 17 Jan 2013 02:22:15 +0000

The 2GB file of code and data at http://bricker.met.psu.edu/~cef13/GRL2006_reproduce.tgz is no
longer accessible. Nothing is accessible at the revised URL from the comments either (http://bricker.met.psu.edu/~cef13/GRL06_reproduce.tgz).

Could someone, who downloaded the file already, make a copy publicly available over the long term? (I am happy to host it long term on my own website if someone wants to get me a copy.)

By: Skiphil

Skiphil — Wed, 19 Dec 2012 23:10:18 +0000

Don’t miss current discussions on further work by Nic Lewis on climate sensitivity and the IPCC estimates. Threads now at Bishop Hill, WUWT, and Climate Etc. Kudos to Nic Lewis and all who have contributed to this progress.

By: Why doesn’t the AR5 SOD’s climate sensitivity range reflect its new aerosol estimates? | Watts Up With That?

Wed, 19 Dec 2012 14:58:44 +0000

[...] in Figure 9.21 of AR4 WG1: Forest 06 and Knutti 02. Forest 06 has several statistical errors (see here) and other problems. Knutti 02 used a weak statistical procedure and an arbitrary combination of [...]

By: Reference

Reference — Sun, 09 Dec 2012 10:46:03 +0000

England and America are two countries separated by a common language.
George Bernard Shaw

By: NicL

NicL — Thu, 22 Nov 2012 17:57:18 +0000

Hu, thanks for your comment. Forest was certainly using the “f_pdf” function, or rather “1 – f_pdf”, to compute a likelihood. And I have found some more statistical “funnies” in the code since writing the above article.

Hugo, I think that in gendetec.pro the f_pdf function is being used, correctly, to get the F-distribution CDF, for a different purpose. I don’t know what the “fcdf” function is there for, but it looks as if it may be something to do with use of PV WAVE, a closley related program to IDL.

By: James Smyth

James Smyth — Tue, 20 Nov 2012 21:59:50 +0000

This is way past the point of pettiness that I would usually care about, but I stand by my original point that whoever originally said “having looked at a bunch of ‘scientific’ or ‘research’ code, you can pretty much tell that it is written on the fly, haha, more like agile dev cause you may have co authors asking you to make changes all along the way and then reviwers can come along and demand changes.” has (or had) no idea what Agile development is about. Maybe by now you’ve educated yourself enough to know that you could have used something like “extreme programming” rather than “agile dev” and gotten it by people who don’t know any better. Agile development is in fact quite adamant that reviewers can’t just some along and demand changes. It is precisely the well-defined, albeit short, cycle which defines Agile.

By: Hugo M

Hugo M — Thu, 15 Nov 2012 07:19:43 +0000

Hu, Forest also uses fcdf in gendetec.pro:

if (keyword_set(ctlind)) then begin if (keyword_set(pv_wave)) then $ Presid(k_t)=1.-fcdf(rssq_k/(ndofd-nscn),ndofd-nscn,ndofn) else $ Presid(k_t)=1.-f_pdf(rssq_k/(ndofd-nscn),ndofd-nscn,ndofn) endif else begin

However, fcdf.pro has an empty function body:
function,fcdf return end

If fcdf is ever called is difficult to say because I hitherto did not succeed in running the program under GDL.