On Feb 9, 12:42 am, "Robert Chung" <
[email protected]> wrote:
> Tom Kunich wrote:
> > If you simply run a six point polynomial averaging on those you will
> > see that the temperature peaked in 2004
>
> Somehow I don't find it hard to believe that you would cherry pick through a
> bunch of alternatives to find a "six point polynomial average" that shows
> what you want.
If you want to say his filter choice is wrong for estimating or even
predicting a compression of the warming trend, then you should argue
why the estimator (filter) is a poor choice.
I have mine. For me, I think the time scale of such a filter is too
short to give me condidence, and I would not make the flattened
conclusion at this time. However, since there is a basic tradeoff
between time and frequency, the filter that would satisfy my
confidence would also be incapable of spotting an actual compression
in as short of time as one more opened up (it could also pass through
"misleading" noise). Life is about tradeoffs. I don't know what your
reason is. Why?
I might run some FFT's to see if the "energy" can be spotted very
well. That might give some clues as to where the corner frequency
would be well placed, among other things.
> However, that isn't what Lindzen did. Notice that in his
> graph the raw temps were flat from 2003 onward.
I'm not ready to accuse him of intentionally corrupting data -- it is
a more serious accusation than calling someone kooky, stupid, both,
etc. I did get an email back from him. He said he didn't know why
off-hand (the response was very prompt). He did write "Incidentally,
the differences between the compilations from the USCDC, GISS and the
UKMO-CRU are about as large even though they essentially use the same
data -- though with somewhat different quality control." I have a
couple more questions for him. Hopefully he'll oblige.
How did you know he used variance adjusted data? I didn't see that in
the .pdf doc. It just says it is data from the Hadley Center. Did
you ever ask him? Your posts to that blog are more than a 1/2 year
old -- plenty of time to get an answer.
Also, you never answered why/how you happened upon this particular
discrepancy. How did you? I think it is very obscure.
He could make that particular claim without doctoring -- all he'd have
to do is defend the subjective choice of filter. It wouldn't seem like
a very smart thing to do: to doctor data when you don't need to.