Considerable progress has been made in understanding the volcanic response problem, but the attempts to reconcile simulations with observations underscore clearly that the present capability for simulating stratospheric aerosols and the climate response to the associated radiative forcing is in a relatively primitive state. As discussed in Chapter 5, the current understanding of albedo modification is insufficient to permit accurate assessment of the likely effects of climate intervention by deliberate alteration of stratospheric aerosols, let alone to plan for deployment. This section highlights some recent work on understanding the climate’s response to volcanic eruptions and discusses prospects for future research directions.
OBSERVATION AND SIMULATION OF RESPONSE TO VOLCANIC ERUPTIONS: PAST STUDIES
There are many different approaches to simulation of volcanic response, which can be used to shed light on the processes involved. The approaches differ in the choice of what is calculated in the model versus what is imposed as boundary conditions based on observations. At the extreme end of the spectrum of forcing models with observations, one can specify the sea surface temperature and sea ice patterns and impose observed volcanic radiative perturbations to the atmosphere, and then see how well the observed changes in land surface temperature and atmospheric circulation patterns can be simulated (as in Graf et al., 1993). As a variant on this approach, different sea surface temperature patterns (e.g., El Niño vs La Niña) or initial circulation states of the stratosphere can be imposed in order to assess which aspects of the observed posteruption climate are due to the aerosol-related radiative forcing versus natural variability which may or may not have been influenced by the eruption (Kirchner et al., 1999; Stenchikov et al., 2004; Thomas et al., 2009a, b). If one is interested primarily in testing aerosol chemistry and microphysics, one can instead impose the observed stratospheric temperature and circulation pattern and see how well the observed aerosol properties can be modeled. At the opposite limit of simulation approaches, models can be driven by estimates of the observed injection of volcanic sulfur dioxide and other substances; both the resulting aerosol and ozone distribution and the
ocean-atmosphere circulation and associated sea ice changes are simulated using a fully coupled model. This approach requires a coupled ocean-atmosphere model with a full representation of stratospheric dynamics and chemistry and is very demanding. It is the kind of simulation that most closely mimics what would be required for assessment of climate intervention actions, but very few simulations of this type have so far been conducted in the context of volcanic response. Various intermediate combinations of the approaches have appeared in the literature.
The complexity of the atmosphere’s response to volcanic eruptions serves as a stark reminder of the challenges confronting any attempt to engineer the climate through deliberate modification of stratospheric aerosols. Aerosol characteristics and the length of time the aerosols remain in the atmosphere depend on the latitude at which the volcanic sulfur dioxide is injected. The aerosols absorb incoming solar infrared and thermal infrared upwelling from below, in addition to keeping some sunlight from reaching the surface, and the infrared effects lead to stratospheric heating that warms the stratosphere. This heating affects stratospheric circulations, which via a range of complex fluid mechanical processes affect the climate of the lower parts of the atmosphere, including surface temperature. The character of the response to the aerosol-induced stratospheric heating is sensitive to interannual variations in the state of the stratosphere at the time the injection occurs, in particular to the state of the Quasi-Biennial Oscillation (Stenchikov et al., 2004; Thomas et al., 2009a). Most attempts to simulate the effects of stratosphere-based climate intervention crudely represent the effect of the engineered aerosols by simply reducing the amount of solar energy hitting the top of the atmosphere; simulations of this sort do not represent the important dynamical and chemical effects of the aerosol-induced stratospheric heating and can lead to severe distortions of the climate response (Tilmes et al., 2009).
As a result, the volcanic response is not a simple cooling of the planet. Large eruptions lead to severe reductions in rainfall over land, especially in the tropics (Trenberth and Dai, 2007). Furthermore, though eruptions cool the following summers, the first winter following an eruption exhibits pronounced high-latitude warming (Robock and Mao, 1992). This winter warming, as well as many other regional aspects of the volcanic response, cannot be accounted for as a response to the blocking of sunlight but instead results as an indirect effect of stratospheric heating; it requires accurate calculation of the aerosol and radiative processes leading to the heating, a well-resolved stratosphere, and a good representation of the interaction between the stratosphere and the lower parts of the atmosphere. Models that incorporate stratospheric heating, either by calculation or by imposing it from observations, can yield a winter warming pattern that has some resemblance to observations, but accurately reproducing the magnitude of the response has proved problematic.
The discussion in Chapter 3 (“Observations and Field Experiments of Relevance to SAAM”) summarizes a recent assessment of the ability of coupled ocean-atmosphere models to reproduce the winter volcanic response as found in the study by Driscoll et al. (2012); see Figure 3.11. There have also been a number of simulation studies aimed at testing models of aerosol evolution rather than climate response (English et al., 2013; Kravitz et al., 2010, 2011b), and these highlight the considerable remaining difficulties both in observing and modeling aerosol properties. Arfeuille et al. (2013) argued that even with accurate observationally based specification of aerosol properties, existing radiative transfer codes could not accurately reproduce the stratospheric heating.
VOLCANIC RESPONSE IS FAR FROM AN EXACT ANALOGY FOR CLIMATE INTERVENTION BY STRATOSPHERIC AEROSOL MODIFICATION
It has been argued that the climate response to engineered stratospheric aerosol modification would have much in common with that from volcanic eruptions, but the volcanic response should nonetheless not be taken as an exact analogue for climate intervention (Robock et al., 2010, 2013). From a microphysical standpoint, the key difference is that eruptions inject sulfur dioxide into a relatively clean stratosphere, whereas engineered injections would add sulfur dioxide to a stratosphere that already has a considerable burden of aerosols. This changes various aspects of the physics determining droplet size growth and coalescence of smaller droplets to form larger ones, both of which affect the residence time of aerosols and their effects on albedo. Engineered injection may also involve a different range of altitudes, and the latitudinal distribution would probably also be different; it is generally assumed that climate intervention would produce a more spatially uniform distribution of aerosols than point-source volcanic eruptions, but it is not yet known how well the actual distribution of aerosols can be controlled. Furthermore, volcanic eruptions inject a range of substances, such as ash, that would not be present in an engineered injection.
From the standpoint of climate response, the chief difference between volcanic and engineered injection is that volcanic eruptions give rise to a short-lived radiative forcing perturbation (at most a few years), which is sufficient to yield a strong climate response over land in the case of large eruptions but does not last long enough for the ocean temperature to be much affected, and insofar as the ocean is affected at all it is only the uppermost layers of the ocean that are involved; sustained aerosol forcing due to climate intervention action would involve a considerably deeper part of the ocean, and a larger ocean response. The probable difference in land-sea temperature contrast between engineered and volcanic stratospheric aerosol injection has impli-
cations for all atmospheric circulations driven by land-sea thermal contrast, notably monsoons and diversions of the midlatitude jet streams. Response of sea ice is sensitive to subtle changes in the ocean circulation, and probably cannot be adequately tested by examination of volcanic response. This is a particular concern, since there are indications that multiple closely spaced eruptions—a rare occurrence such as happened at the time of the Little Ice Age—which approximate the sustained cooling resulting from engineered aerosol modification, can switch the North Atlantic over into an icy mode that can persist for centuries (Miller et al., 2012).
Despite these shortcomings of the volcanic analogue vis-à-vis engineered modification of stratospheric aerosols, the volcanic response engages almost all of the same aspects of atmospheric chemistry, physics, and dynamics as does the climate intervention problem and, therefore, serves as a useful test of the simulation capabilities that would be needed to assess the effects of deployment of climate intervention schemes involving stratospheric aerosol modification.