The sensor will sit on board the $270 m Orbiting Carbon Observatory, which is due for launch in December 2008.

"We know where the sources are but the sinks are a big mystery right now," Dave Crisp of JPL told environmentalresearchweb at the American Geophysical Union Fall Meeting in San Francisco, US. "People want us to predict the effects of carbon dioxide but we don't know how much will stay in the atmosphere."

Crisp and his team are working on a sensor that will measure carbon dioxide levels with an accuracy of 1 part per million, or around 0.3-0.5% - roughly three times better than any other spaceborne trace gas measurement system to date. Today's ozone measurement sensors, for example, have an accuracy of 1%.

The sensor will use three grating spectrometers with a resolving power of 20,000 to analyse reflected sunlight, measuring whole column atmospheric levels of both carbon dioxide and oxygen. "This is a very simple technology pushed to its edge," explained Crisp.

The oxygen data will act as a reference gas to account for factors such as topography and air pressure. While carbon dioxide levels vary significantly near the Earth's surface, when integrated over a whole column, concentrations of the gas differ by only around 8 ppm worldwide. Crisp chose a whole-column technique as it provides information about gas concentrations near the surface - carbon dioxide detected higher in the atmosphere may have moved a significant distance away from the source or sink affecting its concentration.

Currently undergoing testing at JPL in California, the sensor is due for launch into a low-earth orbit in December 2008 and will be moved into its final position by February 2009. The satellite will fly in the A-train, a series of satellites taking data from 705 km above the Earth: the mission is scheduled to last for two years.

Once the system is up and running, the data it provides will be cross-checked with surface measurements taken by the US National Oceanic and Atmospheric Administration and ground-based spectrometer measurements of whole column carbon dioxide made by the Total Carbon Column Observing Network. According to Crisp, it's important to be able to verify the measurements as they could provide contentious information as to how well a country is performing at reducing its carbon emissions.

The system does have a few limitations - it can only take data during the day when sunlight is available, while carbon dioxide levels change significantly between night and daytime. And the sensor cannot take data points when clouds are present.

As a bonus, the oxygen measurements will yield surface pressure measurements with an accuracy of 1 millibar, which will help improve knowledge about the weather. Crisp says that there is currently a distinct lack of surface pressure data, particularly over the oceans.

Data from the Orbiting Carbon Observatory will be available to all in a publicly accessible archive.