Suggestive correlations between the brightness of Neptune, solar variability, and Earth's temperature?

There is a fairly weird paper entitled Suggestive correlations between the brightness of Neptune, solar variability, and Earth's temperature by Hammel and Lockwood. Why is it weird? Various reasons, which I'll try to explain here as best I can, but it really needs someone who knows more about it. These are more notes in case anyone out there feels interested to look.

First off, none of their correlations are significant, a fact which does rather disappoint them: Although correlations between Neptune's brightness and Earth's temperature anomaly--and between Neptune and two models of solar variability--are visually compelling, at this time they are not statistically significant due to the limited degrees of freedom of the various time series. but they don't let that put them off: Nevertheless, the striking similarity of the temporal patterns of variation should not be ignored simply because of low formal statistical significance. Notice the word "formal" they have put in there... they are trying to say that we shouldn't let a formality like lack of significance get in the way of a nice idea. Thats the abstract; within the article text they rather cutely say Low formal statistical significance does not mean the correlations we find are in fact spurious, only that we cannot demonstrate otherwise. What would science be like if all referees let people get away with stuff like that? Later they just about manage to struggle up to 20% sig, but that appears to be based on some rather odd stuff: "We assume four degrees of freedom for the decades long Neptune data set, based on the number of zero crossings in the residual plots in Figure 4." I don't understand why thats a valid method of counting DOF (this is after they have removed a 2nd-order poly from the various datasets). Also they have selected a lag of 17 years between the datasets in order to maximise the correlations, and the sig tests they are using don't account for this.

Second point is the solar irradiance data they use, which shows a rising trend from 1960; and a rising trend from 1980; and looks like nothing I recognise, though they source it to Foukal 2002.

Their reasons for 11-y filtering of solar; and for 11-y filtering of Earth but not Neptune temps; are not terribly well explained.

[Update: I continue to poke at this. So, the paper refs The nature of Neptune's increasing brightness: evidence for a seasonal response by Sromovsky et al. which happens to be online. That paper says the combined disk-averaged variation from 1972 to 2002 is consistent with a simple seasonal model having a hemispheric response delay relative to solar forcing of 30 years.H&L say oh yes, but its not consistent with pre-1970 data. So look again at H&L and indeed there is a break: their Neptune data is two series, with a break between 1965-ish and 1970. I can't quite tell what the difference is, but one obvious question is why S et al didn't use the earlier data? Is it consistent with the later? S et al. note the earlier measurement but say the nominal B filter (Lockwood and Thompson 2002) spans a wavelength range from 390 to 500 nm (half-maximum points), which is about five to six times the width of the b filter and is much more heavily weighted toward shorter wavelengths, with a peak response near 412 nm... and it will take someone who knows these things (hello Stein? Jon?) to work it out.

The other interesting point is that S et al. note the strong latitudinal variation in the brightening - they ignore this in constructing their theory though .

An another maybe interesting paper is, L&J 2006 which says The apparent relationship between Neptune's brightness variation and the 11-year solar cycle seen in cycles 21 22 (1972 1996) has apparently now faded away. - a lesson on troubling yourself about stat sig perhaps, since L now seems to be removing the 11y cycle entirely. Who knows what next years theory might be? -W]

[Update: Tamino looks at the use of the Foukal TSI record in this study and concludes that it cannot validly be used -W]

[Uupdate: some RSS problems may now be fixed... odd non-std CR's -W]