Bit of a tease SA
Guess we will have to wait and see then -are we about to open a 'book' on this ?. My thoughts are that a physical model would have to be piecewise linear, so I am doubtful about quadratic curve fits as a basis of extrapolation. The manifold variables make the signal to noise ratio poor, and as has been suggested, filtering of the input data is probably necessary. Trouble is I am all theory and no practice so these words are probably only adding to the noise too !
AUV-ee, thanks for the link. Have skimmed it and am a bit surprised. Throwing Kalman filters around is all well and good but I wonder if that isn't just convoluting the noise. Why not get a pristine galley section, inflict equivalent damage, lob it in the sea, measure currents and winds, watch its motion and then apply the estimated environment.
Last edited by Mr Optimistic; 6th Dec 2010 at 19:21.