Alan Robock of Rutgers University, US, is "still not convinced" that testing technologies for stratospheric geoengineering can be done. "I just don't think small scale testing is practical," he told environmentalresearchweb. "It wouldn't tell you anything because it wouldn't put enough of a signal into the climate system to detect among the noise of natural variability."

According to Robock, his fellow scientist David Keith of the University of Calgary, Canada, is working with aerospace companies on methods for adding material to the stratosphere from planes. "I suppose that's okay but that will only test the initial phase of putting gas or particles in the stratosphere, it still won't allow you to test what the final composition of the cloud would be, unless you do it in an environment where there is already a cloud," he said.

It's this need for injection of particles into an existing aerosol cloud that makes it impossible to test such geoengineering on a local scale, because the cloud cannot be confined to one area. Knowing the final composition of the cloud is important because the aerosol droplets will grow over time, reducing how much solar radiation they scatter and prevent from reaching the Earth's surface.

Aerosol growth also makes it hard to test geoengineering over a large area but at a low level by injecting only a small amount of aerosol. It means that further injections would be required to maintain cooling levels; the researchers say such effects could not be tested except at full scale.

What's more, because of the natural variability in Earth's climate, trials of geoengineering would need to run for at least 10 years at a large scale to determine whether the technique was working, and whether models had correctly predicted its effect on climate.

"It's not clear to me that anybody could agree to that," said Robock. "That would be actually implementing geoengineering and then it would be hard to stop it. There'll be an interest in pursuing it and continuing it."

Rainfall worries

If geoengineering tests were to weaken the African and Asian summer monsoons, as some models have indicated, water and food supplies to more than two billion people would be under threat. Following the eruption of Mount Pinatubo in 1991, which ejected 20 Tg of sulphur dioxide into the stratosphere, the planet cooled by around 0.5 °C and the Asian monsoon in the summer of 1992 was considerably reduced. The cooling effect lasted around a year, until the volcanic cloud fell out of the atmosphere. But smaller volcanic eruptions since then that released only a couple of teragrammes of sulphur dioxide did not create a change in climate large enough to be distinguished from natural variability.

"The Indian monsoon failed last year and nobody really knows why," said Robock. "We need a better understanding of what the cause is of variations now. We need a lot more information before we can make a decision whether we should do any geoengineering testing in the real world."

For the next few years, Robock recommends that scientists test geoengineering scenarios in models. Indeed, researchers have come together to agree international standardized geoengineering tests for climate models, as part of the CMIP5 Coupled Model Intercomparison Project. Data from these geoengineering runs should be available for the next IPCC report, and will enable direct comparison of the outputs from different models.

"If policymakers are ever going to consider using geoengineering they have to know if it's possible, if it would be effective, how much it would cost and what the negative aspects would be," said Robock. "And so not doing any research would be a mistake because then people would try to discuss it from a situation of ignorance. But a lot more work needs to be done on climate response, on technical issues and on things like governance, how you would decide whether to pursue it or not."

Robock and colleagues Martin Bunzl and Ben Kravitz, both of Rutgers University, and Georgiy Stenchikov of King Abdullah University of Science and Technology, Saudi Arabia, reported their views in Science.