This last step is probably the most critical and important in the entire code of practice as without it our scientific analysis will not be used by anyone else beside ourselves. The discussions above have been concerned more with the meaning of quantitative estimates of uncertainty, but it is important to recognise that in all model applications, those quantitative estimates are necessarily based on qualitative choices and assumptions at earlier steps in the modelling process. A Code of Practice would need to provide a framework for the justification and testing of those assumptions, but most users will not wish to delve into that background material, in the same way as they do not normally wish to query the numerical methods and coding methods used in, say, a flood inundation model. Users require model predictions to make decisions that are based on an assessment of potential future outcomes and the risk (or impacts) associated with those outcomes. This is a natural product of uncertainty estimation.

Communicating Results to Users --WCarpenter?, Sat, 20 Jan 2007 16:20:13 +0000