Detecting deep seismic reflectors aboard the R/V atlantis II

Multiple reflections in the water column often make the detection of deep seismic reflectors in oceanographic exploration very difficult. A novel method of modelling the unwanted feedback generating these multiples is introduced. This model formulates the removal of the multiples as a system identification problem, and the maximum likelihood method of statistics is used to derive an algorithm for estimating the model parameters. This algorithm can be implemented on a small on board computer and yields an enhanced performance over comparable methods previously used. Experimental results of applying the algorithm using data taken aboard the R/V Atlantis II of Woods Hole Oceanographic Institute´s IDOE cruise to the Southwest African coast are presented.