Here’s a post which I wrote last June but didn’t post up at the time because the NAS Panel report came out and I had other pressing matters to comment on. My post as then drafted started:
Last week, through Chefen, Jean S and myself, here here here and here , we showed that MBH98 contained questionable statistical methodology for assessing the relative contribution of solar and greenhouse gases to increased warming and outright false claims about the robustness- the falseness of which was easily determined, making it remarkable that they’ve remained undetected so long.
Now one of the things that we know about the Hockey Team is that most things have a purpose. It may not be obvious and it may not be stated. (I don’t think that one can properly understand the MBH98 reconstruction and its rapid inhaling into policy other than in the context that the IPCC wanted to “get rid of the MWP”‘?.) So what was the purpose of MBH98 Figure 7 in the context of the times? What scores were they trying to settle? What was the pre-MBH98 status of attempts to correlate forcings and temperature?
IPCC reports prior to TAR are usually a good place to start. Here are some notes, which (I think) place MBH98 Figure 7 in context and which, in turn, provide some very interesting leads to follow up on, now that Figure 7 has been overturned.
The rest of my notes follow only slightly modified. I’ll try to post up some notes at some point on the work of Reid 1991 and White et al (JGR 1997) on correlations between solar and tropical SST.
IPCC 1992. (p 65)
IPCC 1992 referred to then recent results from Reid (1991) which they acknowledged as yielding “strong visual correspondence between solar cycle changes and temperature”, but pointed out concerns from Kelly and Wigley related to the reconciliation of solar and greenhouse forcing. Also note the last sentence of the first paragraph, which I’ll comment on below.
These [airborne] show an apparent change of irradiance between the late 1960s and the late 1970s of around 0.4%. There is considerable doubt however about the representativeness of these values (measured over the time-scales of a day) and about the absolute accuracy of the instruments used to obtain them (Lean 1991). Although Reid (1991) argues against these problems, it is clearly difficult to identify a long-term trend using extremely noisy daily data from instruments of uncertain accuracy.
Apart from these measurements, there are no useful direct irradiance measurements prior to 1978 so various authors have tried to deduce irradiance forcing indirectly. For example Reid (1991) has suggested that low-frequency irradiance changes in parallel to the envelope of sunspot activity which shows quasi-cyclic behavior with a roughly 80-year period and Friis-Christensen and Lessen 1991 have hypothesized that low-frequency irradiance changes are related to changes in the length of the solar cycle. In both cases, there is a strong visual correspondence between temperature changes over the past 100 years “€œ see Section C4.2.1. These results are intriguing but they have yet to be fully evaluated in terms of implied changes in solar forcing compared to greenhouse forcing (Kelly and Wigley 1990).
I don’t disagree with the bolded comment in the first paragraph, but do observe that this bit of prudence has not been consistently applied by IPCC e.g. no such comment was later made about MBH.
The bolded comment in the 2nd paragraph refers to the issue that a physical interpretation linking solar changes to temperature changes requires a greater sensitivity of temperature to solar forcing than to greenhouse forcing. As an editorial aside, I don’t like seeing Wigley cited so often as an omnibus authority to support IPCC, ranging from statistic significance to glaciers in the Holocene Optimum to solar forcing. Given this over-exposure, one needs to see an authority other than Wigley to substantiate a claim for reliance in public documents. (2007 – I also note that the Friis-Christensen results have been criticized by Damon and Laut (as noted by Lee on an another thread).
IPCC 1994 p 191-2
IPCC 1994 re-visited the issue of solar correlation devoting an entire subsection devoted to the topic. Again they noted a seemingly compelling statistical relationship, but argued that (1) the relationship was dubious on statistical grounds and, as bolded below, (2) that the relationship was inconsistent with then current ideas of climate sensitivity, which held that climate sensitivity to solar forcing was not greater than climate sensitivity to forcing from additional CO2. Anyway here’s IPCC:
4.5.3 Correlations between Climate and Solar Variability
“Suggestive correlative evidence for an enhanced role of the Sun in forcing climate has been presented by many authors including Labitzke and Van Loon 1993 and references, Reid 1991, Friis-Christensen and Lassen 1991 and Tinsley and Heelis 1993, In these studies the correlations between solar indices such as sunspots and solar cycle length and observed characteristics of the atmosphere (e.g. temperature at particular locations, global average SST etc) are examined. Some authors have questions the usefulness of solar-cycle correlation studies noting that undersampling and/or aliasing of other periodic atmospheric phenomena could lead to spurious results (Teitelbaum and Bauer 1990, Salby and Shea 1991, Dunkerton and Baldwin 1992)”⤮
Labitzke and von Loon 1993 noted remarkably high correlations between stratospheric temperatures and solar indices (such as solar emissions at a wavelength of 10.7 cm) as well as an apparent net propagation of such correlations in the form of planetary scale temperature patterns throughout the troposphere with amplitudes exceeding 1 deg C in some cases. Kodera 1993 showed by examining running correlations that undersampling is not the origin of the Labitzke/van Loon oscillations. However it is clear that large changes in temperature noted by Labitzke and Van Loon 1993 and others are inconsistent with observed changes in the radiative forcing associated with the well-documented total solar irradiance fluctuations of the past decade and very probably the last century”⥔hese wavelengths( UV) are all absorbed well above the tropopause. If this forcing resulted in only local stratospheric temperature changes, then there would be little direct impact on surface climate. .
Friis-Christensen and Lassen 1991 found a high correlation between solar cycle length and NH land temperatures. Hoyt and Schatten used solar cycle length as one of their parameters in deducing a quantitative variation in solar output over recent centuries. Hoyt and Schatten 1993 also found a good correlation between solar output and NH surface temperatures over the past century. However, these results imply that a 0.14% increase in solar output (equivalent to a forcing of 0.34 wm-2) causes a surface warming of 0.5 deg C; this is a high climate sensitivity which, if applied to the 4 wm-2 forcing associated with doubling the concentration of CO2, would result in a warming of about 6 deg C. Thus the hypothesis that variability in solar irradiance explains the observed temperature variations over the past century is inconsistent with our current understanding of climate sensitivity and would require a dramatically different forcing-response relationship for solar forcing than for other forcing mechanisms; there is no known physical mechanism and no modeling evidence to support such a difference, *****
Studies using limited records indicate correlations of winds and temperatures with the solar cycle. However their interpretation remains controversial on statistical grounds. No physical mechanism has been proposed that is quantitatively consistent with the relationships implied by the correlations.
Again, as an editorial point, “efficacy”‘? seems to me like a large and interesting question: it does not seem at all axiomatic that equal wm-2 of high-energy short-wavelength solar forcing at surface should have the same impact on surface temnperature as low-energy long wavelength infrared forcing at high altitudes from additional CO2. The supposed equivalence is relied on but not demonstrated; and continues to be relied on a virtual axoim (See HAnsen et al 2005). In addition, Labitzke, 2006, with the benefit of a nearly douled amount of data, denied that the validity of the above criticism:
Several publications criticized the short data record and suggested that the correlations are due to aliasing caused by dividing the data according to the phase of the QBO (e.g., Teitelbaum and Bauer, 1990; Salby and Shea, 1991). But even when 20 more years of data became available, the correlations remained stable, see Table 1 (Labitzke, 2006).
IPCC TAR, 2001
IPCC TAR’s treatment of solar correlations is reduced from 1994, with Reid 1991 dropping off the radar screen altogether. In their handling of solar correlations, as Jean S observed, IPCC got “punk’d”‘? by MBH98, which included both misrepresentations and incorrect statistical handling of solar correlations. An amusing statement in IPCC TAR, relying on MBH98, was that the use of “multiple correlations avoided the possibility of spuriously high correlations”‘? to solar. It’s hard to believe that something reviewed by entire stadiums of scientists could say something like this, but here it is:
A number of authors have correlated solar forcing and volcanic forcing with hemispheric and global mean temperature time-series from instrumental and palaeo-data (Lean et al., 1995; Briffa et al., 1998; Lean and Rind, 1998; Mann et al., 1998) and found statistically significant correlations. Others have compared the simulated response, rather than the forcing, with observations and found qualitative evidence for the influence of natural forcing on climate (e.g., Crowley and Kim, 1996; Overpeck et al., 1997; Wigley et al., 1997; Bertrand et al., 1999) or significant correlations (e.g., Schàƒwiese et al., 1997; Free and Robock, 1999; Grieser and Schàƒwiese, 2001). Such a comparison is preferable as the climate response may differ substantially from the forcing. The results suggest that global scale low-frequency temperature variations are influenced by variations in known natural forcings. However, these results show that the late 20th century surface warming cannot be well represented by natural forcing (solar and volcanic individually or in combination) alone (for example Figures 12.6, 12.7; Lean and Rind, 1998; Free and Robock, 1999; Crowley, 2000; Tett et al., 2000; Thejll and Lassen, 2000).
Mann et al. (1998, 2000) used a multi-correlation technique and found significant correlations with solar and, less so, with the volcanic forcing over parts of the palaeo-record. The authors concluded that natural forcings have been important on decadal-to-century time-scales, but that the dramatic warming of the 20th century correlates best and very significantly with greenhouse gas forcing. The use of multiple correlations avoids the possibility of spuriously high correlations due to the common trend in the solar and temperature time-series (Laut and Gunderman, 1998). Attempts to estimate the contributions of natural and anthropogenic forcing to 20th century temperature evolution simultaneously are discussed in Section 12.4.
The discussion in Section 12.4 include the following mention of Reid (together with Soon) as follows:
An alternative approach which has been used to reconstruct TSI (Reid, 1997; Soon et al., 1996) is to assume that time variations in global surface temperature are due to a combination of the effects of solar variability and enhanced greenhouse gas concentrations and to find that combination of these two forcings which best combine to simulate surface temperature measurements. However, these authors did not take natural climatic variability into account and a TSI series derived by such methods could not be used as an independent measure of radiative forcing of climate….
However, because of the large uncertainty in the absolute value of TSI and the reconstruction methods our assessment of the “level of scientific understanding”‘? is “very low”‘?.
At this point, we know that the MBH98 claims in respect to solar correlations cannot be relied on (for quite distinct reasons than principal components or bristlecones – see links at top of post). The dismissal of Reid 1997 here does not include any peer-reviewed citation, and, to that extent, is merely an editorial opinion of the section authors – and these are the same authors who made the silly comment about multiple regression.
At this point, I’m far from arguing that anyone has established a connection between solar irradiance changes and temperature changes. However, it is incorrect to say that no statistical correlations between solar irradiance changes and temperature changes (in this case, to tropical SST in particular) have been observed. Whether those correlations are valid is a different issue. IPCC relied to some extent on MBH98 in dismissing these supposed relationships, but, given the defects in this specific area of MBH98 (as well as more general problems), alternative grounds for dismissal have to be sought if one repudiates MBH98. I’m not saying that such alternative grounds are not possible – merely that it is not prudent to rely on MBH98 in respect to taking a position on solar correlations. The other large issue is whether there are physical reasons why the efficacy of solar forcing (high-energy low-entropy at surface) might differ from the efficacy of additional CO2 forcing (low-energy high-entropy at altitude). I am not in a position to make statements one way or the other, but merely observe that I see no reason why this should be an axiom and that establishing differing efficacy would be a necessary aspect of research for anyone seeking to argue an increased role for solar forcing in temperature change.
Labitzke website http://strat-www.met.fu-berlin.de/publications/publications_90s.html
K. Labitzke and H. van Loon 1993Some recent studies of probable connections between solar and atmospheric variability Ann. Geophysicae, 11, 1084-1094, 1993
Reid, George. “Solar variability and climate change on the time scale of decades to centuries,” Eos, 69:18, 567, 1988.
Reid, George. “Solar total irradiance variability and global ocean temperature variations.” Eos, 73:14 supplement, 244, 1992.
Reid, GC 1991. Solar total irradiance variations and the global sea surface temperature record. Journal of Geophysical Research 96: 2835-2844.
Reid, G.C. 1997. Solar forcing of global climate change since the 17th century. Climatic Change 37: 391-405.
Reid, G.C. 1999. Solar variability and its implications for the human environment. Journal of Atmospheric and Solar-Terrestrial Physics 61(1-2): 3-14.
Reid, G.C. 2000. Solar variability and the Earth’s climate: introduction and overview. Space Science Reviews 94(1-2): 1-11.