It’s again time for one of those puzzling results that if they turn out to be true, would have some very important implications and upset a lot of relatively established science. The big issue of course is the “if”. The case in question relates to some results published this week in Nature by Joanna Haigh and colleagues. They took some ‘hot off the presses’ satellite data from the SORCE mission (which has been in operation since 2003) and ran it through a relatively complex chemistry/radiation model. These data are measurements of how the solar output varies as a function of wavelength from an instrument called “SIM” (the Spectral Irradiance Monitor).
It has been known for some time that over a solar cycle, different wavelengths vary with different amplitudes. For instance, Lean (2000) showed that the UV component varied by about 10 times as much as the total solar irradiance (TSI) did over a cycle. This information (and subsequent analyses) have lent a lot of support to the idea that solar variability changes have an important amplification via changes in stratospheric ozone (Shindell et al (2001), for instance). So it is not a novel finding that the SIM results in the UV don’t look exactly like the TSI. What is a surprise is that for the visible wavelengths, SIM seems to suggest that the irradiance changes are opposite in sign to the changes in the TSI. To be clear, while the TSI has decreased since 2003 (as part of the descent into the current solar minimum), SIM seems to indicate that the UV decreases are much larger than expected, while irradiance in visible bands has actually increased! This is counter to any current understanding of what controls irradiance on solar cycle timescales.
What are the implications of such a phenomena? Well, since the UV portion of the solar input is mostly absorbed in stratosphere, it is the visible and near-IR portions of the irradiance change that directly influence the lower atmosphere. Bigger changes in the UV also imply bigger changes in stratospheric ozone and temperature, and this influences the tropospheric radiative forcing too. Indeed, according to Haigh’s calculations, the combination of the two effects means that the net radiative forcing at the tropopause is opposite in sign to the TSI change. So during a solar minimum you would expect a warmer surface!
Much of the longer term variance in solar output has been hypothesised to follow what happens over the solar cycle and so if verified, this result would imply that all current attributions to solar variability of temperature changes in the lower atmosphere and surface ocean would be of the wrong sign. Mechanisms elucidated in multiple models from multiple groups would no longer have any validity. It would be shocking stuff indeed.
Conceivably, there might be another missing element (such as a cosmic-ray/cloud connection) that would counteract this physics and restore the expected sign of the change, but no-one has succeeded in finding any mechanism that would quantitatively give anything close the size of effect that would now be required (see our previous posts on the subject).
So is this result likely to be true? In my opinion, no. The reason why has nothing to do with problems related to the consequences, but rather from considerations of what the SIM data are actually showing. This figure gives a flavour of the issues:
(courtesy Judith Lean). Estimates of irradiance in three bands are given in each panel, along with the raw measurements from various satellite instruments over the last 30 years. The SIM data are the purple dots in the third panel. While it does seem clear that the overall trend from 2003 to 2009 is an increase, closer inspection suggests that this anti-phase behaviour only lasts for the first few years, and that subsequently the trends are much closer to expectation. It is conceivable, for instance, that there was some undetected or unexpected instrument drift in the first few years. The proof of the pudding will come in the next couple of years. If the SIM data show a decrease while the TSI increases towards the solar maximum, then the Haigh et al results will be more plausible. If instead, the SIM data increase, that would imply there is an unidentified problem with the instrument.
In the meantime, this is one of those pesky uncertainties we scientists love so much…