Second, despite the potential for SRM approaches to offset warming in a globally averaged sense, local imbalances in radiative forcing could still lead to regional climate shifts, and the impact of SRM on precipitation and the hydrologic cycle is not very well understood. Short-term volcanic eruptions are not a good direct analog of long-term deployments, yet they provide valuable tests of our process understanding and ability to simulate the climate response to such forcings. Currently climate models underestimate the magnitude of the observed global land precipitation response to 20th-century volcanic forcing (Hegerl and Solomon, 2009) as well as human-induced aerosol changes (Gillett et al., 2004; Lambert et al., 2005), suggesting that these models may not reliably predict the simultaneous effect of SRM approaches on both precipitation and temperature (Caldeira and Wood, 2008). Some modeling studies (Robock et al., 2008) indicate that sulfate aerosol injection could decrease rainfall in the Asian and African monsoons, thereby affecting food supplies. Observational studies also reported that the Ganges and Amazon rivers both experienced very low flows immediately following the eruption of Mount Pinatubo (Trenberth and Dai, 2007). With regard to cloud-based options, it is also unclear if changes to cloud properties in one region could lead to “downwind” changes in the hydrologic cycle, including changes to precipitation.

For the injection of sulfate aerosols, an additional concern exists: the potential for increased concentrations of stratospheric aerosols to enhance the ability of residual chlorine, left from the legacy of chlorofluorocarbon use, to damage the ozone layer, especially in the early spring months at high latitudes. A sudden increase in stratospheric sulfate aerosol could strongly enhance chemical loss of stratospheric polar ozone for several decades, especially in the Arctic (Tilmes et al., 2008). There is also some evidence, however, that sulfate injection, by scattering some of the sunlight that does reach the Earth’s surface, could actually boost ecosystem productivity and crop yields—this could disturb natural ecosystems but be an unintended co-benefit for agricultural systems (Gu et al., 2003; Roderick et al., 2001).

Finally, many SRM approaches require continuous intervention with the climate system in order to offset the forcing associated with GHGs. At some point in the future, if geoengineering were abandoned following its deployment, the adjustment of the climate system to the accumulated GHGs could involve warming on the order of several degrees Fahrenheit per decade (Matthews and Caldeira, 2007), a rate far greater than that estimated for the planet in the absence of geoengineering.