The title of this blog is from a quote by the statistician George E. P. Box (b1919). It is a mantra of modellers everywhere:

“essentially, all models are wrong, but some are useful”

I’m a climate modeller at the University of Bristol. My research is in using models of the earth system – climate, ice sheets, and vegetation – to recreate the past and predict the future, and in trying to quantify the inevitable uncertainties. For more details, see my university and academia.edu pages.

Quantifying uncertainties is important in all scientific research: without an estimate of confidence, a result cannot be placed in context, cannot be given meaning.  But it is essential in climate science because climate is, by definition, the statistics of weather: and statistics is the science of uncertainty.

Even more important are the stakes. The global reach of the earth’s climate – everything under the sky – and the intimate, complex connections between humans, society and the environment mean that climate scientists must work hard to understand the range of possible futures we face.

Climate scientists have been accused of many things, of which two are insufficient consideration or communication of uncertainty, and insufficient transparency. People will disagree on the extent to which we deserve these accusations. But there will always be room for improvement in both, and this blog is my small contribution to the conversation.

All models are undoubtedly ‘wrong’, because we cannot precisely simulate every breath of wind, every raindrop, or every worm turning over the soil. But this does not preclude their usefulness as tools to explore the broad consequences of known physical laws. These abstract representations help us to make sense of our world.