Alan Robock continues to misrepresent geoengineering using stratospheric aerosols. On one hand, he agrees that "it must be geoengineering plus mitigation, and never geoengineering instead of mitigation"; but on the other hand, he has run a climate model experiment that's precisely "geoengineering instead of mitigation." While his work identifies possible risks associated with the "instead of" idea, these risks cannot be transferred to the realistic "plus mitigation" case. It's a bit like pointing out the risks of skydiving without a parachute and saying they apply to the more sensible "with parachute" case. The same problem applies to the work on stratospheric ozone by Simone Tilmes and her colleagues.
Yes, it's useful to show that "geoengineering instead of mitigation" isn't a good idea. But we already knew that. What alarms me is that these studies strongly mislead unless given suitable caveats, as inexpert readers may think that the risks identified by these idealized sensitivity studies apply equally to realistic "geoengineering plus mitigation” scenarios.
From the start, I've advocated for the need to consider geoengineering in conjunction with mitigation as the only tenable option. Nevertheless, the implications of my work haven't sunk in, so I'll take this opportunity to explain them further.
I consider two cases: (1) a standard monotonic pathway to carbon-dioxide concentration stabilization at 450 parts per million (ppm)--the WRE450 case from my 1996 Nature article; and (2) an overshoot pathway where concentrations rise to 540 ppm in 2090 before declining to eventual stabilization at 450 ppm. The overshoot case gives us more time to reverse our emissions trajectory, which would be less costly and more technically feasible than a monotonic trajectory that requires stringent mitigation policies such as an immediate moratorium on building coal power plants.
So where does geoengineering come in? If we need extra time to stabilize atmospheric carbon dioxide at 450 ppm but don't want to risk a dangerous rise in temperature by overshooting it, then geoengineering could provide a negative radiative forcing equal to the difference that would occur between the monotonic and overshoot pathways. It's a trivial exercise to calculate this negative forcing--it would rise slowly from zero to peak at around -1 watt per square meter (the equivalent of turning off a Christmas tree light over every square meter of Earth's surface) in 2090 and then decline slowly back to zero.
The extreme cases considered by Alan, Tilmes, and their colleagues show that the negative climate and ozone consequences of geoengineering in the 450 ppm stabilization mitigation-plus-geoengineering scenario would be negligible.
For stratospheric ozone, the aerosol loadings for much of this century in the 450-ppm case would be about an order of magnitude less than those considered by Tilmes. By the time aerosol loadings reached an appreciable amount, the ozone layer would have essentially healed.
As for precipitation, the consequences of geoengineering would actually be beneficial. Alan brings up the issue that precipitation is more sensitive to short-wave (stratospheric aerosol) forcing than to long-wave (greenhouse gas) forcing. Thus, if one were to try to offset a large carbon-dioxide forcing with geoengineering, the positive precipitation increases due to enhanced carbon dioxide would be more than fully compensated and the net effect would be a (possibly detrimental) precipitation decrease.
But let's look at a realistic case. Suppose we follow a monotonic pathway to atmospheric carbon-dioxide stabilization at 450 ppm. Globally, this would lead to a precipitation increase relative to today of, say, X percent. The overshoot pathway would lead to still greater precipitation increases, say (X+Y) percent. Geoengineering to compensate for additional warming in the overshoot case would reduce precipitation by Z percent, where Z is greater than Y. The net effect of overshoot-plus-geoengineering would therefore be to increase precipitation relative to today by W percent, where W=X+Y-Z. W must be less than X. In other words, the combined scenario would lead to future precipitation that's actually closer to today than in the mitigation-only case. The differential short-wave/long-wave effect on precipitation actually works in our favor, contrary to what might be inferred from Alan's work and claims.
He sees the possibility that "the promise of geoengineering will delay the implementation of mitigation" as a primary danger. I disagree. As scientists who communicate with policy makers and politicians, the challenge is to ensure that this doesn't happen. Presenting scary extreme scenarios as relevant to the future isn't going to advance the mitigation cause. Indeed, crying wolf may well damage our credibility. Our responsibility is to explain why geoengineering can only be a complement to mitigation and evaluate the implications, costs, and technological challenges of scenarios that combine geoengineering and mitigation.
It's time to move on and assess the combined mitigation/geoengineering option in detail.