Can we quantify uncertainties in complex models of the complete Earth System?

The models we are using to base our judgements on are all imperfect. We must acknowledge this and attempt to address quantitatively how these uncertainties affect the judgements we are able to make on the basis of these models. Without uncertainty estimates (error bars if you like) any decision making will be arbitrary. With them it is more complex, remains subjective but at least can be justified and explained. This will be essential if the hard choices that we face are to be effectively communicated and acted on. Addressing uncertainty is central to all aspects of modelling, from the dynamic climate models to models of ecological systems, society and economics. Coupling such models makes uncertainty quantification even more essential. I believe this is a fundamental question to address before we can even start to answer specific questions about the Earth System.

2 comments

1. Alastair says:

   at

   The answer to the question is ‘No’! Since the uncertainties are unknown we cannot quantify them. For instance, climate sensitvity was supposed to be between 1.5C and 4C. ClimatePrediction.net got values of over 10C. There is no way to tell who is right, far less quantify the uncertainty.
2. Kalense says:

   at

   This question is, indeed, fundamental. Behind it sits the bigger question – how can we gather enough data on the complete Earth system to be able to construct and validate even the simplest models? Complex systems do unexpected things. We need huge amounts of data on how our social and economic systems, the living world and our inanimate environment interact, before we can construct plausible models. Data-poor models are always dangerous, expecially if they are based on data that do not include rare events. It is the unusual outliers that may provoke the most important system responses.