The Earth’s carbon cycle is influenced by a rich diversity of lifeforms and complex ecological systems. Representing this behaviour as a series of mathematical functions remains a work in progress. "The models we analyzed did not even agree on the sign of the cumulative carbon flux by 2100," Nicole Lovenduski of the University of Colorado, US, told environmentalresearchweb.

The study helped clarify the picture, however. "By considering a number of models you gain a better sense of the full range of uncertainty and the sensitivity of the uncertainty to the representation of these physical and biological processes," said Lovenduski.

Together with Gordon Bonan of the US National Center for Atmospheric Research, Lovenduski weighted the models based on the ability of these mathematical expressions to reproduce the observed change in carbon accumulations from 1959–2005. However, this approach failed to close the gap between predicted values when the weighted system was applied from 2006 to 2100.

The researchers found that uncertainty in the projection of global terrestrial carbon accumulation is driven primarily by model structure. They made a number of recommendations: areas to focus on include raising the accuracy of observations and improving process understanding.

"Net terrestrial carbon flux is a complex quantity that depends on many different processes that we are yet to fully understand – for example, how will climate change affect fire?" said Lovenduski. "In addition, our present day observations of terrestrial carbon uptake are highly uncertain themselves, both on global and regional scales."

Improvements in model structure require deep knowledge of real-world processes. Based on their analysis the scientists believe that a multitude of approaches will be required to capture the full range of possible outcomes. "The focus on reducing multi-model spread does not necessarily make carbon cycle projections more reliable and may, in fact, limit scientific progress," they conclude in their paper in Environmental Research Letters (ERL).

Related links

Related articles