If reducing carbon emissions fails to stop climate change, we may one day have the option of sending mirror-supporting satellites into space or filling the stratosphere with light-reflecting particles to block the sun's rays.

According to a new study, such measures could significantly cool Earth. But researchers caution that if they if they do not work or are suddenly halted, they could make matters worse.

"As far as I know, this is the first century-scale, time-dependent simulation of a geoengineering scheme," says Ken Caldeira, a climate scientist at the Carnegie Institution of Washington's Department of Global Ecology in Stanford, Calif., and senior author of the study published this week in Proceedings of the National Academy of Sciences USA. Geoengineering refers to activities designed to alter Earth's climate that may include blocking the sun, large-scale reforestation and sequestering carbon dioxide in the ocean.

First, Caldeira and his colleague, ecologist Damon Matthews, constructed a model to determine what would happen to the global climate if carbon dioxide emissions continued to increase at their current rate. Their findings: by 2100 Earth's surface temperature would have risen by an average of 3.5 degrees Celsius (6.3 degrees Fahrenheit) from what it had been in 1900, with temperatures in some parts of the Arctic jumping by as much as by as much as six degrees C (10.8 degrees F).

Next, the pair simulated the effect of adding a geoengineering scheme sometime between the years 2000 and 2075, which would result in the amount of radiation hitting the Earth being uniformly decreased. The authors' model predicts that temperatures in tropical regions would dip slightly (by 0.35 degree C or 0.63 degree F) and climb by a degree C (1.8 degrees F) in the Arctic. In other words, the planet's average temperature would not change much from what it had been back in 1900.

"The positive result is you can ramp up geoengineering along with the carbon dioxide," Caldeira says. "Within a decade, you would get most of the cooling effect of a geoengineering scheme, and in two decades you'd get all of it."

The climate mediation model also indicated that the amount of rainfall could also be manipulated. Sans intervention, steadily warming oceans will lead to more rainfall over the ocean, whereas increased carbon dioxide in the atmosphere will diminish the levels of water evaporated from plants, leading to decreased rainfall in the tropics. If the sun's rays are partially deflected, the ocean will not warm, and only tropical rains will be affected, lowering overall precipitation about ten times as much as in the warming scenario.

Caldeira and Matthews also modeled the scenario if a geoengineering scheme were halted or failed after initially cooling things off. They found that the effects would be similar to what would happen if an artificial cap were temporarily slapped on Earth's thermostat. "If the earth wanted to be four degrees warmer," Caldeira says, and then the "system fails, the rapid response time of the climate system means that you would get hit with all that warming." In this case, he warns, it could result in the rapid warming of the planet at a clip of up to 20 times the current rate of warming (0.2 degree C, or 0.36 degree F per decade).

"The missing stat [here] is to assume that we will make our policies depend solely on climate models," says Gregory Benford, a physicist at the University of California, Irvine, who studies geoengineering. He advocates doing more experimentation with these types of schemes to determine the risks and benefits, noting that he does not believe the consequences would necessarily be grave were a geoengineering scheme to fail or be abruptly halted for political or other reasons. "You've been holding off a differential, so the warming rate will accelerate," Benford says. "However, you just had 20 years when you could try to do these other measures, such as cutting carbon emissions."

Caldeira says that, in any case, geoengineering is probably not the first option in controlling climate change. "The risk-averse strategy," he says, "is to reduce emissions."