Improving climate models and predictions

You may find this difficult to believe, but climate science is not conducted within the pages of the Guardian or Daily Mail, or on the websites of the world’s more opinionated bloggers. But for those without ready access to the ongoing debate among scientists, one could be forgiven for thinking that it is. Even the sagacious Clive James is at it now (“I’m not a climate scientist, but…”).

Climate predictions rely on models of the Earth’s atmosphere, and models are, owing to the complexity of the natural world and the limited computational power at our disposal, idealised representations of reality. However, simplified though they can sometimes be, climate models are subject to continuous development and refinement as we learn more about the Earth system. Uncertainties in the predictions made with the aid of models are thereby reduced.

Contrary to the claims of anthropogenic climate change sceptics and outright deniers, the science of global warming is settled. What continues to be debated are the finer details of the processes involved and their complex interactions. These are of course crucial, and there is much discussion within the Earth science community about uncertainties in climate models, and the impact these have on predictions which will have huge social and economic ramifications.

In a recent issue of the American Geophysical Union’s house journal Eos, Columbia University researcher Lisa Goddard and others discuss among other things the issue of uncertainty, and the difficulty in making predictions with climate models.

“Although scientific effort never will eliminate uncertainty, it can better estimate uncertainty. This is especially true at the time scales of seasons to decades, which are the most actionable time scales. These are the time scales at which incremental adaptation can evolve with the help of appropriate state-of-the-art climate predictions.”

Goddard and her colleagues go on to stress that climate science is inherently probabilistic in nature, and, as such, individual models cannot be used for point-specific climate forecasting. This crucial point must be hammered into the dense skulls of pundits and policymakers who cannot seem to cope with anything other than simple, binary logic. What is needed, say the authors, is for decision makers to learn how to treat information about future climate as a range of possibilities.

Eos is available only to members and affiliates of the AGU, and owing to copyright restrictions I cannot reproduce the entire Goddard et al. article here. This is unfortunate, as in my opinion it would help raise the public’s awareness and understanding of climate change to have such discussions among experts visible in the public domain.

Further reading

Goddard et al., “The Urgent Need for Improved Climate Models and Predictions”, Eos 90, 343 (2009)