Comment on “Estimated solar contribution to the global surface warming using the ACRIM TSI satellite composite” by N. Scafetta and B. J. West


[1] Whereas the Intergovernmental Panel on Climate Change [2001] attributes most of recent surface warming to human activities, Scafetta and West [2005] claim that the Sun contributed at least 10–30% to global surface warming of 0.40 ± 0.04 K from 1980 to 2002. But their claim depends crucially on the solar irradiance time series adopted for this period; nor is it reproduced by a multivariate linear regression analysis of the solar, anthropogenic and climate data.

[2] Scafetta and West [2005] use the composite of contemporary solar irradiance observations that Willson and Mordvinov [2003] compiled by cross calibrating published individual data sets. In that composite, shown by the (red) symbols in Figure 1a, irradiance levels during the 1996 minimum of the 11-year solar activity cycle exceed the level of the prior minimum in 1986, an increase which Scafetta and West [2005] “minimally interpret as a 22-year square waveform modulation” of amplitude 0.45 Wm−2. But by comparing overlapping, concurrent irradiance time series, Fröhlich and Lean [1998, 2004] had already identified significant drifts amongst the various data sets attributable to changes in radiometric sensitivities rather than to real solar variability. When the irradiance data sets are corrected for these instrumental effects, the resultant composite, also shown in Figure 1a as a solid (green) line, differs from that of Willson and Mordvinov [2003], in that irradiance levels during the 1986 and 1996 solar minima are similar. Evidence for an upward irradiance trend is negligible, as is a secular solar contribution to recent surface warming [Fröhlich and Lean, 1998]. Furthermore, Scafetta and West [2005, Figure 4] characterize the amplitude of their assumed 22-year cycle as increasing steadily between ∼1987 and 2002. But a step-like jump, rather than a steady increase, is the source of the apparent upward trend in the Willson and Mordvinov [2003] composite relative to that of Fröhlich and Lean [2004]. As Figure 1b shows, the increase occurs only over 3 years, from mid 1989 to mid 1992. Between January 1980 and June 1989, and again after July 1992, the trends in the composites are not significantly different. Nor can cycle amplitude differences account for the 22-year effect since peak irradiances in even-numbered cycle 22 and odd-numbered cycle 23 agree to within ±0.06 Wm−2 (0.004%) in both the PMOD and ACRIM composites, as shown by the 15-month smoothed curves in Figure 1a.

Figure 1.

(a) Compared are the irradiance composites of Willson and Mordvinov [2003] (WM2003, red symbols) and Fröhlich and Lean [2004] (FL2004, green solid line). (b) Their differences have no significant trend between January 1981 and June 1989, nor after July 1992. The ∼3-year jump from mid 1989 to mid 1992 produces higher irradiance levels in 1996 relative to 1986, which Scafetta and West [2005] assume to be an upward “trend between minima during solar cycles 21–23” associated with a 22-year solar cycle.

[3] Scafetta and West's [2005] claim that the Sun has minimally contributed ∼10–30% of recent surface warming is not reproduced when the solar and anthropogenic contributions are isolated by multivariate linear regression, even using the Willson and Mordvinov [2003] solar irradiance composite. Figure 2 shows the results of such an analysis, using the same global surface temperature anomalies as Scafetta and West [2005] (from the University of East Anglia Climatic Research Unit (CRU) [Jones and Moberg, 2003]). The volcanic and El Niño Southern Oscillation (ENSO) signals identified in the surface temperature anomalies are shown in Figure 2b, and an empirical model of their combined effect is indicated in Figure 2a. The time series of volcanic aerosols are from Sato et al. [1993] and the Multivariate ENSO Index is from Wolter and Timlin [1998], updated to 2004.

Figure 2.

(a) Shown are monthly mean global surface temperatures and an empirical model of the variations arising from ENSO and volcanic influences, lagged by 7 and 9 months, respectively. (b) Subtracting the ENSO and volcanic signals from the observational record produces (c) the residual surface temperature anomalies. Compared with these residuals is the combined influences of solar variability (according to the Willson and Mordvinov [2003] composite) and anthropogenic gases (the net forcing of greenhouse gas warming and tropospheric aerosol cooling). Solar irradiance is lagged by one month relative to the observed surface temperatures. (d) Shown are the individual solar and net anthropogenic contributions to the residual surface temperature anomalies in Figure 2c.

[4] Removing the ENSO and volcanic contributions to the CRU global surface temperature anomalies produces the time series of residual surface temperature anomalies shown in Figure 2c. The upward trend in these residuals of 0.0166 ± 0.0008 K per year produces a warming of 0.365 ± 0.02 K between mid 1980 and mid 2002. Note that the warming in this residual time series is smaller than the 0.38 ± 0.02 K warming in the actual surface temperatures (Figure 2a) over the same period. This latter warming is to be compared with Scafetta and West's [2005] reported surface warming of 0.4 ± 0.04 K since they did not remove ENSO and volcanic effects.

[5] Under debate is the relative solar and anthropogenic contribution to the residual surface temperature anomalies in Figure 2c. Shown in Figure 2d are the solar and anthropogenic signals identified in multiple regression analysis of the surface temperature anomalies. The solar irradiance time series used is that of Willson and Mordvinov [2003] and the net anthropogenic forcing is the combined time series of CO2 and aerosol forcings, in which the greenhouse gas forcing since 1850 is +2.44 Wm−2 and the tropospheric aerosol forcing since 1850 is −0.6 Wm−2 [Hansen et al., 2002]. The resultant solar plus anthropogenic effect is compared with the residual surface temperature anomalies in Figure 2c. The increase in the Willson and Mordvinov [2003] irradiance composite between consecutive 11-year cycle minima (mid 1986 and mid 1996) produces 0.017 K warming, which is 4.7% of the 0.365 K warming in the residual surface temperature anomalies between 1980 and 2002. If, as Scafetta and West [2005] argue, climate sensitivity is a factor of 1.5 higher for 22-year forcing relative to the stronger 11-year forcing (which dominates the multiple regression results) the 0.017 K may be amplified to 0.026 K, which is 7% of the 0.365 K surface warming between 1980 and 2002.

[6] In an alternative approach, the multiple regression analysis is repeated using the Fröhlich and Lean [2004] irradiance composite and the net anthropogenic forcing to which is added an additional radiometric forcing equivalent to the difference between the Willson and Mordvinov [2003] and Fröhlich and Lean [2004] irradiance composites (i.e., the time series in Figure 1b multiplied by 0.7 and divided by 4). In this case the surface warming between 11-year cycle minima in mid 1986 and mid 1996 is 0.18 K. With just the anthropogenic forcing alone (i.e., without the additional forcing corresponding to the differences between the two irradiance composites) the surface warming between cycle minima is 0.155 K. The difference of 0.025 K between these separate analyses represents the climate response to the upward trend in the Willson and Mordvinov [2003] irradiance composite relative to that of Fröhlich and Lean [2004], estimated with the scaling coefficient (i.e., sensitivity) of secular anthropogenic forcing (i.e., at lower frequencies than the 11-year cycle). The 0.025 K difference is 7% of the 0.365 K warming in the residual surface temperature anomalies between 1980 and 2002, and is likely an upper limit to the surface warming associated with the upward trend in the Willson and Mordvinov [2003] composite, since climate sensitivity to secular forcing is expected to exceed that for 22-year cyclic forcing.

[7] In summary, rather than minimally contributing ∼10–30% of global mean surface warming from 1980 to 2002, the Willson and Mordvinov [2003] irradiance composite accounts for 7% or less of this warming. Since secular change in solar irradiance is negligible in recent decades according to the Fröhlich and Lean [2004] composite, it is entirely possible that the Sun has contributed little to the recent global warming trend.


[8] NASA funded this work. Ongoing collaboration with Claus Fröhlich is appreciated. Data were obtained from,,, and and