Feb 13, 2012
'Quality control' of climate models can help predict future rainfall
Will cotton farmers in India still be able to rely on the south-west monsoon by 2020? Will the Corn Belt in the Midwestern United States become wetter or drier? Are European rivers more likely to burst their banks and cause significant floods? These are important questions that need answering to prepare for climate change. So can climate models provide the answers?
A new study suggests that, when it comes to the global water cycle, the answer is no: current climate models are not providing trustworthy predictions. However, with greater "quality control" measures in place, climate models should be able to make precipitation predictions that we can have confidence in.
Beate Liepert from NorthWest Research Associates in Washington State, US and Michael Previdi from the Lamont-Doherty Earth Observatory at Columbia University, studied data from 18 different climate models. All but two of the models predict an atmospheric transport of around 5% more moisture towards land by the end of this century, assuming a moderate increase in greenhouse gases over the same period.
"This increase in moisture is approximately equivalent to a major river discharge, like the Nile," Liepert told environmentalresearchweb.
Traditionally, climate scientists have averaged the results from a number of models, to iron out individual errors. But when it comes to the water cycle, multi-mean models (as the average result is known) are misleading. "We show that some models don't get the basic physics right, and these models can tilt the multi-model mean in one direction," said Liepert.
In the case of the 18 climate models that Liepert and Previdi studied, global mean precipitation exceeded evaporation in 13 (the atmosphere "leaked" water), while in the remaining five evaporation exceeded precipitation (the atmosphere "flooded"). "The fluxes of water in and out of the atmosphere do not add up correctly – we describe it as a 'ghost' sink/source," explained Liepert.
Even small leaks have big implications for the energy budget of the atmosphere. Water exists in three phases in the atmosphere (water vapour gas, liquid rain droplets and solid ice), and turning from one phase to the other consumes or releases very large amounts of energy.
"A model leak of moisture from the atmosphere establishes an artificial positive perturbation of the energy budget, since less energy is required for the transition," said Liepert. "This spurious extra energy can then be utilized elsewhere in the atmosphere and hence constitutes a 'ghost' forcing of climate. The ghost forcing strengths range from −1 to +6 watts per square metre, which is in the size range of forcings by non-CO2 greenhouse gases."
A handful of the models that Liepert and Previdi studied handled the moisture budget correctly (and presumably got the physics right). By identifying the outliers the researchers believe it is possible to narrow the uncertainty range and make more accurate predictions about the future behaviour of the global water cycle. Their findings are published in Environmental Research Letters (ERL).
Rather than continually increasing the resolution of climate models, and adding to the complexity (by coupling more and more physical, biological and chemical processes), Liepert and Previdi say that greater "quality control" – checking the self-consistency of models – is required. "Climate research institutions worldwide aggressively pursue the strategies of bigger, faster and more complex," said Liepert. "They are under pressure to produce results for the next IPCC report. Policymakers are forced to take action and sponsoring bigger, faster, more complex climate models is the easiest thing to do, easier than dealing with greenhouse-gas emissions and implementing sustainable energy policies." Checking, checking and checking again might be boring, but vital if we want to be prepared for whatever the future may hold.
- Beate Liepert
- Inter-model variability and biases of the global water cycle in CMIP3 coupled climate models, Beate G Liepert and Michael Previdi 2012 Environ. Res. Lett. 7 014006
About the author
Kate Ravilious is a contributing editor for environmentalresearchweb.