Signature Estimation and Drop-Out Implications for a Triple Source Marine Seismic Survey

We demonstrate the compatibility of modern far-field signature estimation with triple source acquisition and the resilience of this estimation in the face of drop-outs. Our results are based on data acquired with a two-string, 2965 cubic inch source array, and with a de-tuned 2415 cubic inch variant of that array which exhibits reduced peak amplitude and lower primary-to-bubble ratio. Modelled signatures were compared with those derived by least-squares inversion of near-field hydrophone measurements, according to the method discussed in Hargreaves et al. (2015, 2016) and, after examining different choices of ghosting parameters, we derive de-signature operators and apply these to seismic shot records. We find that the de-tuned 2415 array gives comparable results to the full array after de-signature, which is encouraging for robustness of signature estimation in the face of drop-outs. We note improved de-signature results using a frequency-dependent sea-surface reflection coefficient which is smaller than predicted based on the sea state reported during acquisition. This suggests energy loss mechanisms are significant in the vicinity of the source array.