IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group I: The Physical Science Basis Estimating Past Solar Radiative Forcing Reconstructions of past variations in solar irradiance

Long-term solar irradiance changes over the past 400 years may be less by a factor of two to four than in the reconstructions employed by the TAR for climate change simulations. Irradiance reconstructions such as those of Hoyt and Schatten (1993), Lean et al. (1995), Lean (2000), Lockwood and Stamper (1999) and Solanki and Fligge (1999), used in the TAR, assumed the existence of a long-term variability component in addition to the known 11-year cycle, in which the 17th-century Maunder Minimum total irradiance was reduced in the range of 0.15% to 0.3% below contemporary solar minima. The temporal structure of this long-term component, typically associated with facular evolution, was assumed to track either the smoothed amplitude of the solar activity cycle or the cycle length. The motivation for adopting a long-term irradiance component was three-fold. Firstly, the range of variability in Sun-like stars (Baliunas and Jastrow, 1990), secondly, the long-term trend in geomagnetic activity, and thirdly, solar modulation of cosmogenic isotopes, all suggested that the Sun is capable of a broader range of activity than witnessed during recent solar cycles (i.e., the observational record in Figure 2.16). Various estimates of the increase in total solar irradiance from the 17th-century Maunder Minimum to the current activity minima from these irradiance reconstructions are compared with recent results in Table 2.10.

Each of the above three assumptions for the existence of a significant long-term irradiance component is now questionable. A reassessment of the stellar data was unable to recover the original bimodal separation of lower calcium (Ca) emission in non-cycling stars (assumed to be in Maunder-Minimum type states) compared with higher emission in cycling stars (Hall and Lockwood, 2004), which underpins the Lean et al. (1995) and Lean (2000) irradiance reconstructions. Rather, the current Sun is thought to have ‘typical’ (rather than high) activity relative to other stars. Plausible lowest brightness levels inferred from stellar observations are higher than the peak of the lower mode of the initial distribution of Baliunas and Jastrow (1990). Other studies raise the possibility of long-term instrumental drifts in historical indices of geomagnetic activity (Svalgaard et al., 2004), which would reduce somewhat the long-term trend in the Lockwood and Stamper (1999) irradiance reconstruction. Furthermore, the relationship between solar irradiance and geomagnetic and cosmogenic indices is complex, and not necessarily linear. Simulations of the transport of magnetic flux on the Sun and propagation of open flux into the heliosphere indicate that ‘open’ magnetic flux (which modulates geomagnetic activity and cosmogenic isotopes) can accumulate on inter-cycle time scales even when closed flux (such as in sunspots and faculae) does not (Lean et al., 2002; Y. Wang et al., 2005).

A new reconstruction of solar irradiance based on a model of solar magnetic flux variations (Y. Wang et al., 2005), which does not invoke geomagnetic, cosmogenic or stellar proxies, suggests that the amplitude of the background component is significantly less than previously assumed, specifically 0.27 times that of Lean (2000). This estimate results from simulations of the eruption, transport and accumulation of magnetic flux during the past 300 years using a flux transport model with variable meridional flow. Variations in both the total flux and in just the flux that extends into the heliosphere (the open flux) are estimated, arising from the deposition of bipolar magnetic regions (active regions) and smaller-scale bright features (ephemeral regions) on the Sun’s surface in strengths and numbers proportional to the sunspot number. The open flux compares reasonably well with the cosmogenic isotopes for which variations arise, in part, from heliospheric modulation. This gives confidence that the approach is plausible. A small accumulation of total flux (and possibly ephemeral regions) produces a net increase in facular brightness, which, in combination with sunspot blocking, permits the reconstruction of total solar irradiance shown in Figure 2.17. There is a 0.04% increase from the Maunder Minimum to present-day cycle minima.

Prior to direct telescopic measurements of sunspots, which commenced around 1610, knowledge of solar activity is inferred indirectly from the 14C and 10Be cosmogenic isotope records in tree rings and ice cores, respectively, which exhibit solar-related cycles near 90, 200 and 2,300 years. Some studies of cosmogenic isotopes (Jirikowic and Damon, 1994) and spectral analysis of the sunspot record (Rigozo et al., 2001) suggest that solar activity during the 12th-century Medieval Solar Maximum was comparable to the present Modern Solar Maximum. Recent work attempts to account for the chain of physical processes in which solar magnetic fields modulate the heliosphere, in turn altering the penetration of the galactic cosmic rays, the flux of which produces the cosmogenic isotopes that are subsequently deposited in the terrestrial system following additional transport and chemical processes. An initial effort reported exceptionally high levels of solar activity in the past 70 years, relative to the preceding 8,000 years (Solanki et al., 2004). In contrast, when differences among isotopes records are taken into account and the 14C record corrected for fossil fuel burning, current levels of solar activity are found to be historically high, but not exceptionally so (Muscheler et al., 2007). Implications for solar radiative forcing

In terms of plausible physical understanding, the most likely secular increase in total irradiance from the Maunder Minimum to current cycle minima is 0.04% (an irradiance increase of roughly 0.5 W m–2 in 1,365 W m–2), corresponding to an RF[11] of +0.1 W m–2. The larger RF estimates in Table 2.10, in the range of +0.38 to +0.68 W m–2, correspond to assumed changes in solar irradiance at cycle minima derived from brightness fluctuations in Sun-like stars that are no longer valid. Since the 11-year cycle amplitude has increased from the Maunder Minimum to the present, the total irradiance increase to the present-day cycle mean is 0.08%. From 1750 to the present there was a net 0.05% increase in total solar irradiance, according to the 11-year smoothed total solar irradiance time series of Y. Wang et al. (2005), shown in Figure 2.17. This corresponds to an RF of +0.12 W m–2, which is more than a factor of two less than the solar RF estimate in the TAR, also from 1750 to the present. Using the Lean (2000) reconstruction (the lower envelope in Figure 2.17) as an upper limit, there is a 0.12% irradiance increase since 1750, for which the RF is +0.3 W m–2. The lower limit of the irradiance increase from 1750 to the present is 0.026% due to the increase in the 11-year cycle only. The corresponding lower limit of the RF is +0.06 W m–2. As with solar cycle changes, long-term irradiance variations are expected to have significant spectral dependence. For example, the Y. Wang et al. (2005) flux transport estimates imply decreases during the Maunder Minimum relative to contemporary activity cycle minima of 0.43% at 200 to 300 nm, 0.1% at 315 to 400 nm, 0.05% at 400 to 700 nm, 0.03% at 700 to 1,000 nm and 0.02% at 1,000 to 1,600 nm (Lean et al., 2005), compared with 1.4%, 0.32%, 0.17%, 0.1% and 0.06%, respectively, in the earlier model of Lean (2000).

Table 2.10. Comparison of the estimates of the increase in RF from the 17th-century Maunder Minimum (MM) to contemporary solar minima, documenting new understanding since the TAR.

Reference Assumptions and Technique RF Increase from the Maunder Minimum to Contemporary Minima (W m–2)a Comment on Current Understanding 
Schatten and Orosz (1990)  Extrapolation of the 11-year irradiance cycle to the MM, using the sunspot record. ~ 0  Irradiance levels at cycle minima remain approximately constant. 
Lean et al. (1992)  No spots, plage or network in Ca images assumed during MM.  0.26  Maximum irradiance increase from a non-magnetic sun, due to changes in known bright features on contemporary solar disk. 
Lean et al. (1992)  No spots, plage or network and reduced basal emission in cell centres in Ca images to match reduced brightness in non-cycling stars, assumed to be MM analogues.  0.45  New assessment of stellar data (Hall and Lockwood, 2004) does not support original stellar brightness distribution, or the use of the brightness reduction in the Baliunas and Jastrow (1990) ‘non-cycling’ stars as MM analogues. 
Hoyt and Schatten (1993)b  Convective restructuring implied by changes in sunspot umbra/penumbra ratios from MM to present: amplitude of increase from MM to present based on brightness of non-cycling stars, from Lean et al. (1992). 0.65  As above  
Lean et al. (1995)  Reduced brightness of non-cycling stars, relative to those with active cycles, assumed typical of MM. 0.45  As above  
Solanki and Fligge (1999)b  Combinations of above.  0.68  As above  
Lean (2000)  Reduced brightness of non-cycling stars (revised solar-stellar calibration) assumed typical of MM. 0.38  As above  
Foster (2004) Model  Non-magnetic sun estimates by removing bright features from MDI images assumed for MM.  0.28  Similar approach to removal of spots, plage and network by Lean et al. (1992).  
Y. Wang et al. (2005)b  Flux transport simulations of total magnetic flux evolution from MM to present.  0.1  Solar model suggests that modest accumulation of magnetic flux from one solar cycle to the next produces a modest increase in irradiance levels at solar cycle minima. 
Dziembowski et al. (2001)  Helioseismic observations of solar interior oscillations suggest that the historical Sun could not have been any dimmer than current activity minima. ~ 0    


a The RF is the irradiance change divided by 4 (geometry) and multiplied by 0.7 (albedo). The solar activity cycle, which was negligible during the Maunder Minimum and is of order 1 W m–2 (minimum to maximum) during recent cycles, is superimposed on the irradiance changes at cycle minima. When smoothed over 20 years, this cycle increases the net RF in the table by an additional 0.09 W m–2.

b These reconstructions extend only to 1713, the end of the Maunder Minimum.


Figure 2.17. Reconstructions of the total solar irradiance time series starting as early as 1600. The upper envelope of the shaded regions shows irradiance variations arising from the 11-year activity cycle. The lower envelope is the total irradiance reconstructed by Lean (2000), in which the long-term trend was inferred from brightness changes in Sun-like stars. In comparison, the recent reconstruction of Y. Wang et al. (2005) is based on solar considerations alone, using a flux transport model to simulate the long-term evolution of the closed flux that generates bright faculae.

  1. ^  To estimate RF, the change in total solar irradiance is multiplied by 0.25 to account for Earth-Sun geometry and then multiplied by 0.7 to account for the planetary albedo (e.g., Ramaswamy et al., 2001). Ideally this resulting RF should also be reduced by 15% to account for solar variations in the UV below 300 nm (see Section and further reduced by about 4% to account for stratospheric absorption of solar radiation above 300 nm and the resulting stratospheric adjustment (Hansen et al., 1997). However, these corrections are not made to the RF estimates in this report because they: 1) represent small adjustments to the RF; 2) may in part be compensated by indirect effects of solar-ozone interaction in the stratosphere (see Section; and 3) are not routinely reported in the literature.