Publication

2007-07

English

http://archimer.ifremer.fr/doc/00000/11030/8978.pdf (Publisher's official version, 1.80 Mo)

2007 IEEE

Nsiri Benayad^{1, 2}, Chonavel Thierry³, Boucher Jean^{3, 4}, Nouze Herve⁵

1 : Univ Hassan 2, Dept Phys, Ain Chock Fac Sci, Casablanca 20101, Morocco.
2 : Ecole Natl Super Telecommun Bretagne, F-29200 Brest, France.
3 : Ecole Natl Super Telecommun Bretagne, Dept Signal & Commun, F-29200 Brest, France.
4 : National Scientific Research Center Laboratory, Traitement Algorithmique et Matériel de la Communication, de l'Information et de la Connaissance, UMR 2872, Brest 29238
5 : IFREMER, Ctr Brest, Marine Geosci Dept, F-29280 Plouzane, France.

Ieee Journal Of Oceanic Engineering (0364-9059) (Ieee-inst Electrical Electronics Engineers Inc), 2007-07 , Vol. 32 , N. 3 , P. 729-743

3

Physical Oceanography

Bernoulli Gaussian BG process, blind deconvolution, Gibbs sampler, maximum likelihood ML, maximum posterior mode MPM, Monte Carlo Markov chains MCMCs methods, Prony algorithm, seismic deconvolution, stochastic expectation maximization SEM

In seismic deconvolution, blind approaches must be considered in situations where reflectivity sequence, source wavelet signal, and noise power level are unknown. In the presence of long source wavelets, strong interference among the reflectors contributions makes the wavelet estimation and deconvolution more complicated. In this paper, we solve this problem in a two-step approach. First, we estimate a moving average (MA) truncated version of the wavelet by means of a stochastic expectation-maximization (SEM) algorithm. Then, we use Prony's method to improve the wavelet estimation accuracy by fitting an autoregressive moving average (ARMA) model with the initial truncated wavelet. Moreover, a solution to the wavelet initialization problem in the SEM algorithm is also proposed. Simulation and real-data experiment results show the significant improvement brought by this approach.