Thanks for the link, the raw data surely helps! But I’m still not sure if we can compute that 0.5 C without seeing the full 12-proxy reconstruction. Weighting by 2/3 is mentioned, but does it have a theoretical basis even if redness is p=2/3? Maybe that is because they use RE and not RMS. RMS would catch that extra variance due to redness without any extra tricks, right? (if the sample window is large enough). Still a bit confused.

Oh, IMHO, UC, as a Ph.D. student, you should not look too closely to Mann&Lees 96, it may give you too many bad ideas ;)

OK;) Median smoothing of spectrum will remove that peak right away. So, ‘background noise’ can never be very red.

]]>Oh, IMHO, UC, as a Ph.D. student, you should not look too closely to Mann&Lees 96, it may give you too many bad ideas ;)

]]>Wow – their best work to date. ]]>

165: Yeah, but there are other places like RC or NPR which try to act serious

You must have missed this RC piece.

]]>If one uses optimal estimation techniques for a NH temperature reconstruction, assuming that NH temperatue time series is nice (ie ergodic, widesense stationary,gausssian) than the optimal estimate for the time series with no or poorly correlated proxies will always have a hockeystick shape.

That’s where we need to think about splicing. But if you assume the average to be 1890 value or something like that, then yes. Theoretical AR1 wanders around 0, in real life we have additional random constant to estimate.

#164

If p of the calibration residuals really is around 0.67… That’s something. p of global temp during calibration period is 0.64, proxy noise is not red they say, 12 thermometers couldn’t bring the 2-sigma down to 0.5 C, ozone in the stratosphere affects tree rings, but CO2 doesn’t… Well, then we can just forget MBH99.

]]>It just seems so bloggy.

]]>Thks. Good guess, quite close. Still a student, though (PhD candidate who is spending too much time here? :) ). Linear optimal estimation with Gaussian inputs is fun, but things get tricky in real life when we have non-linear systems with non-Gaussian inputs.

Homework for math geeks: using the first eq in #160 show that is a good approximation of random walk power spectrum.

]]>