As previously indicated, Akasofu believes the changes in the sun to have been the primary mechanism for increased temperatures, but this is a theory that runs into immediate problems. Consider the following graphs:
(Original source, IPCC AR4)
(Lean 2000 TSI reconstruction overlaid by the HadCRU temperature index, source: Tamino)
(Image created by Robert A. Rohde / Global Warming Art. Follow link for key.)
As you can see, there are wide variations in estimates of changes in TSI and temperature over the course of the LIA. For my initial discussion, I will use the values indicated by Lean, 2000; and by Moberg et al., 2005 (the bright red line in the temperature reconstructions). Partly, this is because these reconstructions provide figures closest to those accepted by Akasofu.
Close examination of the charts shows a change of temperature between 1620 (the coldest period of the LIA) and 1850 of around 0.5 degrees C. There is a further increase of temperature of 0.3 degrees between 1850 and 1940; and from 1940 to 2004, another increase of just over 0.4 degrees C. That is reasonably close to the same increase over each interval.
In contrast, the increase in TSI over the same intervals are, respectively, 1.5 w/m^2, 1 w/m^2, and 0.05 w/m^2. Allowing that the increase in TSI is entirely responsible for the increase in temperature between 1620 and 1850, then we expect a 0.33 increase in temperature per 1w/m^2 increase in TSI. In other words, we would expect a 0.33 degree increase between 1850 and 1940, close enough to the correct value; but only a 0.17 degree increase in temperature between 1940 and 2004. So, <b>if we assume changes in TSI are responsible for all increases in temperature up to 1950, then that demonstrates that they cannot be responsible for increases in temperature post 1950</b>.
This argument does not depend on any assumptions about solar modulation of cosmic rays, or any other speculative theory yet to be proposed. The effects of such modulations, if real, would have included in the temperature increases from 1620 to 1850, as also those from 1850 to 1940. Further, the temperature rise cannot be explained as the impact of multi-decadal oscillations. Both 1940 and 2004 were at, or near the peak for both the Pacific Decadal Oscilation and the Atlantic Multidecadal Oscillation; and both were El Nino years. Consequently, any temperature difference between them due to oceanic oscillations should be minimal, and of no consequence.
As noted previously, there is significant uncertainty in reconstructions of LIA temperatures and TSI. It may occur to some people that Akasofu can escape the logic of my analysis by the appropriate choice of a temperature reconstruction and TSI reconstruction. However, the problem is not that easily solved. Taking Wang et al., 2005 for his temperature reconstruction, for example still leaves him with a 0.6 w/m^2 difference between 1620 and 1850, a 0.4 w/m^2 difference between 1850 and 1940, and at best a 0.2 w/m^2 difference between 1940 and 2004. That predicts a temperature difference between 1940 and 2004 or only 0.17 degrees C.
In fact, direct measurements show that there has been no increase in TSI since 1975, showing that no selection of TSI reconstructions can both explain temperature changes in the LIA, and temperatures changes in the late 20th century:
(Image created by Robert A. Rohde / Global Warming Art)
(Return to Part 1)
(Continue to Part 3)