A comparison of model fitting using separation ratios, epsilon fitting and phase matching
Ian Roxburgh (Queen Mary University of London)

I discuss the properties of 3 surface layer independent model fitting techniques - comparison of model and observational data on separation ratios, comparison of epsilon values and phase matching. I conclude that epsilon matching is the simplest to implement. Both separation ratios and epsilon matching only require knowledge of model and observed frequencies which are stored with a model set, and both only interpolate in model values, but the errors in separation ratio fitting are strongly correlated requiring the evaluation of covariance matrices, whereas the errors are uncorrelated in epsilon matching. Phase matching is in principle similar to epsilon matching but requires the calculation of the inner pause shift of the models at the observed frequencies. I give some examples.