Seismic Wavelet Estimation Based on ARMA Model via Cumulants and SVD-TLS
Yongshou Dai, Yuanyuan Li, Wei Dong Lei, Zhiyong Huo · 2006
On the assumption that the reflection coefficient series is a non-Gaussian, stationary and statically independent random process, ARMA model was introduced to solve the mixed phase seismic wavelet estimation. In this paper, a cumulant-based SVD-TLS (singular value decomposition and total least squares) algorithm with less computational price was employed. Numerical simulations demonstrate that the ARMA model provides a parsimonious and effective signal model in fitting seismic trace. The cumulant-based SVD-TLS algorithm is not sensitive to colored Gaussian noise, but it strongly relies upon the accuracy of the trace cumulant estimates. If the estimated error and variance of trace cumulant are moderate, the cumulant-based SVD-TLS algorithm combined with the ARMA model description of the seismic trace is appropriate for seismic wavelet estimation. The real seismic data examples demonstrate the practicability of the method in seismic data processing