10,000 climate models predict significant global temperature rise

A distributed computing project to model the earth's atmosphere has ended, yielding a prediction that the temperature of the atmosphere will rise between 1.4C and 3C by 2050.
The study included an individual climate simulation (supplied by the UK Met Office) run on almost 10,000 home computers, under the auspices of the BBC's Climate Change Experiment at climateprediction.net.
Like the infamous butterfly flapping its wings and causing a hurricane, tiny inaccuracies in the measurement of the initial conditions can magnify into vast changes as a simulation progresses. However, by running multiple climate forecasts, each with slightly different initial conditions, it's possible to work out which broad trends are likely to persist and which just arise from hiccups in the imprecise data fed in.
The process, known as ensemble forecasting, has been used in regular weather forecasting for many years, but its use in climate forecasting is still rare. The results suggest that the world will warm by between 1.4C and 3C by the middle of the century, even under a mid-range emissions scenario, where some cuts are enacted. That rate is somewhat faster than that predicted by other climate models.
"It's only by running such a large number of simulations -- with model versions deliberately chosen to display a range of behaviour -- that you can get a handle on the uncertainty present in a complex system such as our climate," said Dan Rowlands of Oxford University's Department of Physics, lead author of a paper describing the research in Nature Geoscience. "Our work was only possible because thousands of people donated their home computer time to run these simulations."
Myles Allen of the School of Geography and Environment and Department of Physics, Oxford University, an author of the paper and principal investigator of the climateprediction.net project, told Wired.co.uk: "Most forecasts of global warming are based on the range of results that different groups around the world happen to contribute to a model comparison. These groups don't set out to explore the full range of uncertainty, which is why studies like ours are needed."

Thank you : Source :  www.wired.co.uk