Methods for model optimisation
based upon the criterion of the model output fitting the tomographic image. The
problems that we anticipate are:
• How do we characterise the behaviour of a physical model?
• How do we select the model input parameters to generate a
representative sample of solutions?
• How to deal with global/local minima?
• Once we have found a model run that matches the tomographic image
(in fact a tomographic movie) how unique is that
model solution?
• Do those model inputs really represent reality? Is the knowledge of the physics behind the model
adequate? If not can we identify its
lack and improve it?
Model-image coupling