Nathan wrote in the global warming thread:


>It looks like the R code first applies a 12-month smoothing filter to the data before computing differences. That is, the value for each month is replaced by the average of that month, the preceding 5 months, and the following 6 months. Then the smoothed data are differenced (i.e. differences between successive smooth months are computed). All the rest of the code in the diff12() function, including the rounding, are technical details to convert between year-month date formats.

thanks for looking at that unfortunately quite unreadable comment, I don't know why the XML didn't work.
I had tried to cast my guesses about that Diff12 into a formula. And as you describe the formula, I guess you are also reluctant to cast formulas here in the comments.

Which line in the rcode does the 5 and 6 months averageing? Is it within the diff12 definition? (again in this unreadable comment )


>The script works fine on my Mac, but I don’t see much point in replicating someone’s refutation of a non-interesting paper.


I was mostly confused by the realclimate answer, because as I understood they claim that they could reproduce the curves in the
Humlum et al article.
That is if I look at the curve of co2 concentrations (fig 1 in real climate article), which is looking rather highly symmetric (especially in contrast to the highly wiggly temperature curves) then I would expect that averaging
and taking differences wouldnt disturb this symmetric "look", in particular I wouldn't expect to get a "wiggly" curve
as the Diff12(co2) curve in the Humlum article or the green curve
as that in fig 2. (of the real climate article) (but which (as I wrote in the unreadable comment) may not be the curve of diff12(co2)
but that of the diff12(of one of the temperatures)).

So in short if real climate which critizes the Humlum et al paper gets the same Diff12(co2) curve then something (at least for me)
unexpected happens and I would like to understand whats going on.


Jim wrote:

>I’ve tried to run the R script on my local box but get 404’s (not found) for the data downloads.
It’s good to see if this stuff is reproducible. Is this worth pursuing?

were you connected to the internet? the data is stored nonlocally:

hss2012 <- function(wfl=12,forcing=FALSE,HadCRUT4=FALSE,HadSST3=FALSE,
ftp.co2="ftp://ftp.cmdl.noaa.gov/ccg/co2/trends/co2_mm_mlo.txt",
ftp.hc3="http://www.cru.uea.ac.uk/cru/data/temperature/hadcrut3gl.txt",
ftp.hs2="http://www.cru.uea.ac.uk/cru/data/temperature/hadsst2gl.txt",
ftp.CO2="ftp://ftp.cmdl.noaa.gov/ccg/co2/trends/co2_mm_gl.txt",
ftp.hs3="http://www.metoffice.gov.uk/hadobs/hadsst3/data/TS_all_realisations.zip",
ftp.hc4="http://www.metoffice.gov.uk/hadobs/hadcrut4/data/time_series/hadcrut4_monthly_ns_avg.txt")