In the last post, I examined two composites of satellite measurements of total solar irradiance (TSI), and showed that whether one uses the PMOD composite or the ACRIM composite, there is solid evidence that TSI has not changed sufficiently to be responsible for modern global warming.
I also stated in comments to that post, that I have reason to believe the PMOD composite is closer to the truth. This is, of course, a completely separate issue; whether one uses PMOD, ACRIM, or the third contender, the IRMB composite, the change in TSI is insufficient to account for modern global warming. But since I expressed greater confidence in PMOD than ACRIM, I’ll share my reasons for this opinion.
Satellite measurements of TSI don’t come from a single instrument, but many. They begin in 1978 with the HF radiometer aboard the Nimbus 7 satellite, and continue with a number of different missions, with data plotted here:
It’s abundantly clear that the different instruments don’t agree with each other. Each has its own slightly different scale, so they must be adjusted in order to agree with each other. This is generally done by comparing one instrument’s results to another’s during their period of overlap, and determining the offset which will bring them into alignment.
But there are more differences than just a zero-point offset. The response of space-borne radiometers changes over time. There can be a sizeable drift during the early part of a space mission, and a slower but detectable drift throughout the mission; in particular, the response changes due to exposure of the instrument to sunlight. That’s why several of the satellites carried multiple (redundant) radiometers. The radiometers which were only rarely exposed to solar radiation supply a “reference” to indicate how the main instrument’s sensitivity changes over time. Unfortunately, not all missions had multiple radiometers. Furthermore, “glitches” sometimes occur, causing a shift in the scale for some decidedly non-solar reason.
One of the greatest sources of difficulty in matching up the various satellite data sets is the “ACRIM gap.” The lauch of ACRIM II was delayed, due to the shuttle Challenger explosion, so there is a roughly 2-year gap between the end of the ACRIM I data set and the beginning of ACRIM II, from early/mid-1989 to late 1991. However, two different data sets are available which overlap both missions: HF (Nimbus 7) and ERBS. Unfortunately, these two show different trends during the ACRIM gap. Depending on which one accepts as more reliable, one gets decidedly different results when joining ACRIM I to ACRIM II.
Hence piecing the various satellite data series together to form a single continuous data set is a rather complex process. Three principal efforts have been carried out, leading to three rival composites of TSI: the PMOD composite, the ACRIM composite, and the IRMB composite:
I will chiefly consider the comparison between PMOD and ACRIM, not considering the IRMB composite. In the last post, I subtracted a best-fit sinusoid from PMOD and ACRIM to generate residuals, which gives us an idea of how TSI is changing above and beyond the solar cycle:
The two biggest differences between the two are, first, a much greater TSI in ACRIM during the earliest part of the time span (from 1978 to 1980), and second, a sharp rise in ACRIM relative to PMOD from 1989 to 1992. This rise occurs during the ACRIM gap, and is due to the different choices made by the two science teams about how to bridge the gap. The ACRIM composite relies principally on HF (Nimbus 7) data, while the PMOD composite relies more on ERBS data.
Willson & Mordvinov (2003, Geophysical Research Letters, vol. 30, pg. 1199) claim that the increase seen in HF/Nimbus 7 data during the ACRIM gap is more plausible than the decrease seen during the same time period in ERBS data:
The ACRIM gap occurred during a time of increasing and maximum magnetic activity during solar cycle 22. The positive correlation between solar magnetic activity and TSI is compatible with the positive slope of the NIMBUS7/ERB results during this period. The negative slope of the ERBS results is incompatible. The most likely explanation of the divergence of slopes during the ACRIM gap is uncorrected ERBS degradation. The use of ERBS to relate ACRIM1 and ACRIM2 in the PMOD composite would therefore yield systematically lower results following the ACRIM gap and a smaller trend between solar activity minima.
They show the trends before, during, and after the ACRIM gap here:
However, they provide no substantiation for the claim that “The ACRIM gap occurred during a time of increasing and maximum magnetic activity during solar cycle 22.” One of the surest indicators of (proxies for) both TSI and magnetic activity is the sunspot index. Here are the Zurich sunspot counts during the ACRIM gap:
As you can see, the trend during this time interval is decreasing (in agreement with ERBS) rather than increasing (as indicated by HF/Nimbus 7). Therefore I find Willson & Mordvinov’s claim not credible; the sunspot index agrees with the result of ERBS but contradicts Nimbus 7.
Frohlich has compared the results of all three composites to a proxy reconstruction of TSI based on magnetograms observed at the Kitt Peak solar observatory (by Wenzler 2005, Ph.D. thesis, ETH Nr 16199, Eidgenössische Technische Hochschule, Zürich):
The PMOD composite correlates with the magnetogram-reconstructed TSI with common variance 0.83, significantly better than ACRIM (0.62) or IRMB (0.63).
I find the extreme range of variation in the residuals of the ACRIM composite, which is greater than the range of variation during the solar cycle itself, somewhat suspicious. I likewise find the extreme decline in the earliest part of the ACRIM composite (more than a full W/m^2) suspicious. Most important, the fact that the bulk of the increase shown in the ACRIM composite, nearly 0.7 W/m^2 relative to PMOD, occurs exactly during the ACRIM gap, makes me very skeptical of the correctness of the ACRIM composite.
Only time will tell (if at all) with certainty which composite is closer to the truth. But my examination of the published reports gives me far greater confidence in the correctness of the PMOD composite, than ACRIM.
6 responses so far ↓
nanny_govt_sucks // July 27, 2007 at 6:16 pm
Can you address this please? Why, in the PMOD reconstruction, were such liberties taken with the only available data? Both ACRIM and IRMB show high values at the very beginning of the series, but PMOD does not. Why?
[Response: When time permits (my employer demands my attention) I'll update the post to address this question.]
EliRabett // July 28, 2007 at 2:34 am
You might take a look at SOLAR IRRADIANCE VARIABILITY SINCE 1978 Revision of the PMOD Composite during Solar Cycle 21 by Froehlich which discusses
george // July 28, 2007 at 9:01 pm
This strikes me as one of those cases where people should be very careful about making claims about “upward trends of 0.04%” when there are so many other more likely possibilities (instrument errors, no continuous record)
The fact that there was no continuous record should give one all the more reason for caution.
The increase in TSI should be the very last possibility to be considered after every other possibility has been ruled out.
And even if there had been an increase over a couple cycles, I just don’t see how one could call that a “trend”.
But then as was pointed out to me by Willson in the other thread, I don’t know much about solar physics (just about common sense).
nanny_govt_sucks // July 29, 2007 at 3:32 am
Why didn’t the IRMB folks go for this argument?
george // July 30, 2007 at 12:08 pm
According to a paper by A. Krivova and S. K. Solanki (Max-Planck-Institut f¨ur Aeronomie, 37191 Katlenburg-Lindau, Germany)
“An additional argument against the secular increase of the total irradiance is provided by the evolution of the UV irradiance. Neither the Mg II core-to-wing ratio, which is the standard proxy of UV irradiance, nor the UV-irradiance reconstruction show the increase from one minimum to another (Cebula et al., 1992; Viereck & Puga, 1999; Fligge & Solanki,
2000).”
SOLAR TOTAL AND SPECTRAL IRRADIANCE: MODELLING AND A POSSIBLE IMPACT ON CLIMATE
The drop in sunspot activity from 1989-1992 (the period over which Willson claimed increasing magnetic activity) is very evident in this NOAA graph of annual mean sunspot numbers and also in this Zurich graph of monthly sunspot numbers.
It is true that the single month that had the maximum number of sunspots (200) was Aug, 1990, but the activity in the months to either side was significantly less and the count for that month was only slightly higher than the monthly number for June 1989 (196.2) as you can see by looking at the monthly mean counts accessed from this NOAA page.
The year that had the maximum annual mean count for the years 1989-1992 was actually 1989 (157.6) as opposed to 1990 (142.6), 1991 (145.7) and 1992 (94.3) which you can see in a table accessed from the same page above.
With regard to piecing the various instrument data together to bridge the ACRIM gap, according to Frohlich
“The ACRIM composite obviously neglects the HF correction during the ACRIM gap and thus concludes that the Sun was increasing by 403 ppm between the two minima. The difference of 403 + 33 ppm can be explained by 376 ± 71 ppm derived from the correction over the corresponding period of time.”
In other words, all of what Willson calls an increase in TSI can be accounted for simply by including this correction.
Frohlich explains the various corrections and methods for making the composite in Total Solar Irradiance since late 1978Claus FröhlichPhysikalisch-MeteorlogischesObservatoriumDavos,
Finally, with regard to the match between the solar irradiance curve constructed from the magnetograms based on solar models, Can Digging into the Details Settle this Debate? says that
“The deviation of his data set from the solar models doesn’t trouble Willson, who says that while the solar proxy models are useful for trying to describe solar activity back in time before scientists had actual observations of total solar irradiance , “the models are not competitive in accuracy or precision with even the worst total solar irradiance satellite observations.”
“Joe Gurman, US Project Scientist for SOHO, isn’t so quick to dismiss the models. “These models have done a very good job in simulating the solar activity over recent solar cycles as well as more pronounced, long-term changes in solar output that can be linked to historically documented changes in climate,” he says. “If you accept Willson’s conclusion that total solar irradiance has increased over the last two solar cycles, then not only do you have to explain why the models don’t show it, but you also have to explain why no other single instrument shows a similar increase, and why none of our other solar indicators, like total magnetic flux or ultraviolet light output, has shown a similar increase.”
John V Magisano // September 6, 2007 at 9:01 pm
Is seems you guys are getting to caught up with the 1989-1992 sunspot data. This was the period of solar cycle number 22, the solar max itself. This is always a period of high activity with large swings in activity. You need the long term trend of sun spot data to show the clear correlation between earth average temperature and solar activities. We only have 2+ solar cycles of data for irradiance since the satellites have only been up there that long (< 30 years) but again here the data generally shows an upward trend in irradiance. SOHO constantly surprises us with higher than expected energy outputs from the sun. Remember we have not seen increases in global warming over the past 5 years or so. Hopefully the sun has peaked but you never know. However we can also hope another Maunder Minimum isn’t coming.
[Response: This post is specifically about the satellite data, so of course it doesn't address the long-term data. That was the topic of this post. As for the claim that "... again here the data generally shows an upward trend in irradiance," no it doesn't. See this post and this post.]
Leave a Comment