Two recent papers (Lockwood & Fröhlich, 2008 – ‘LF08′; Scafetta & Willson, 2009 – ‘SW09′) compare the analysis of total solar irradiance (TSI) and the way the TSI measurements are combined to form a long series consisting of data from several satellite missions. The two papers come to completely opposite conclusions regarding the long term trend. So which one (if either) is right, then? And does it really matter?
This issue is a very familiar one when it comes to long-time series from satellite data. Each individual satellite only lasts a few years, and so a 30 year time series needs to be stitched together from a series of satellites. Each of those instruments might have a different calibration, and may have non-climatic drifts associated with instrument degradation, or orbital effects. Thus it can often be the case that there is a degree of ambiguity in putting together the series. This issue is at least part of the difference between the RSS and UAH tropospheric temperature trends, and in the CERES/ERBE analyses discussed recently.
The differences between PMOD and ACRIM have already been discussed by the SkepticalScientist and Tamino, so here is just an update in the light of the two recent papers. The important issue here is the so-called ‘ACRIM-gap’, the time between the ACRIM-I instrument ceased and when the ACRIM-II observations started (mid-1989 to late 1991), and how the data from these two instruments are combined using other overlapping observations. Note that the ‘ACRIM’ name for the Willson et al time-series simply implies that it was put together by some people on the ACRIM science team, not that they use different satellite data.
The focus on these papers is what the ‘ACRIM gap’ implies for TSI levels during the solar minimum at solar cycles 21 and 22. Whereas PMOD suggests that the TSI levels during these minima are similar, ACRIM suggests that the TSI level is higher during the minimum of cycle 22. SW08 even claim that there has been a positive ‘minima trend‘.
LF08 conclude that the PMOD is more realistic, since the change in the TSI levels during the solar minima, suggested by ACRIM, is inconsistent with the known relationship between TSI and galactic cosmic rays (GCR). It is well-known that the GCR flux is generally low when the level of solar activity is high, because the solar magnetic fields are more extensive and these shield the solar system against GCR (charged particles). However the two effects don’t always go in lockstep, so this is suggestive rather than conclusive.
It is also clear from the instrumental data that the TSI tends to increase with the solar activity level – at least over the solar cycle. LF08 argue that if the ACRIM ‘minimum trend’ is correct, this will mean that past reconstruction of TSI based on e.g. sunspots are incorrect, and a lot of studies on the past climate variations would be wrong. This does not mean that the ACRIM data are useless, but that there are uncertainties regarding the relationship TSI-levels, solar activity for different time scales.
I found insufficient detailed description in SW09 of the methodology used in their analysis to be able to judge the real merit of their work. The paper provides a link to auxiliary material that does not work. However, the figures in the paper don’t really convince when I don’t know how they were made.
Furthermore, I found the SW09 a bit confusing, as it gives the impression that the PMOD composite relies on ERBS/ERBE data during the ACRIM-gap (“The PMOD team uses the sparse ERBS/ERBE data base to ‘bridge’ the ACRIM gap, conforming the higher cadence Nimbus 7/ERB to it by making adjustments due to …”). However the information in LF08 says PMOD used HF from Nimbus 7 (ERB).
The PMOD analysis involves an adjustment to correct for a glitch in the ERB data (orientation changes and/or switching off), but SW09 claims – without providing convincing arguments – that this correction cannot be justified.
The ACRIM composite does not account for a jump in the ‘ACRIM-gap’ due to instrumental changes. SW09 show a comparison between different analyses and Krivova et al. (2007) modeled TSI, but later acknowledge that the latter modeled TSI disagrees with measurements on decadal time scales. Furthermore, when the TSI is not adjusted over the ‘ACRIM-gap’, there is the apparent inconsistency between TSI and GCR.
Update: My conclusion is that the LF08 paper is far more convincing than the SW09 in terms of whether the TSI data should be adjusted over the ‘ACRIM-gap’. But the this is probably not the final word on the matter.