No abstract

The effect of uncertainty in the parameters that define the experimental geometry is investigated to determine the performance of matched-field inversion for geoacoustic properties. The investigation is carried out using simulated data for a vertical line array in a North Pacific environment. The results indicate that there can be significant degradation of inversion performance for small errors (∼1/3 wavelength) in source depth, depth and tilt of the array, ocean depth and source/array range. The relatively high sensitivity to mismatch in experimental geometry suggests that uncertainties of this type can be tolerated in the inversion by including the geometrical parameters in the global search algorithm. The simulations show that performance is improved by inverting for both the ocean bottom properties and the geometrical parameters simultaneously.