Our recent paper on how rainfall processes are represented in climate models (‘Afternoon rain more likely over drier soils’ published online in Nature this week) led to some interesting headlines around the globe including "Shock. Climate models are too alarmist" (Andrew Bolt’s column in the Melbourne Herald Sun). This is not exactly how I’d express the meaning of the paper!
A key result of our work is that the current models are getting the relationship between rainfall and local soil moisture incorrect, and therefore can end up in a vicious circle of drying soil and reduced rainfall. Representing the processes by which clouds and rain develop within computer models in order to make climate projections over the entire globe for decades hence is tremendously difficult. Until now we didn't have a global observational basis for testing how well the models were predicting this particular aspect of climate. The paper attempts to do this using data from across six continents. Both the paper itself and our press release go to considerable efforts to provide a balanced report on the research.
We have a saying in the UK that you shouldn't throw the baby out with the bath-water. There are plenty of people around who jump on any evidence that the models aren't as good as we would like, and instantly move to reject the whole enterprise of climate modelling. We all know the models have to get better, and to get better, scientists need to identify aspects where the models fail. Sometimes communicating that process isn't easy.
Christopher M. Taylor, Richard A. M. de Jeu, Françoise Guichard, Phil P. Harris & Wouter A. Dorigo ‘Afternoon rain more likely over drier soils’ was published in Nature on 12 September 2012. DOI 10.1038/nature11377
