High-resolution prestack seismic inversion using a hybrid FISTA least-squares strategy
A new inversion method to estimate high-resolution amplitude-versus-angle attributes (AVA) attributes such as intercept and gradient from prestack data is presented. The proposed technique promotes sparse-spike reflectivities that, when convolved with the source wavelet, fit the observed data. The i...
| Authors: | , , |
|---|---|
| Format: | article |
| Status: | Published version |
| Publication Date: | 2013 |
| Country: | Argentina |
| Institution: | Universidad Nacional de La Plata |
| Repository: | SEDICI (UNLP) |
| Language: | English |
| OAI Identifier: | oai:sedici.unlp.edu.ar:10915/99562 |
| Online Access: | http://sedici.unlp.edu.ar/handle/10915/99562 |
| Access Level: | Open access |
| Keyword: | Geofísica Algorithm Inversion Amplitude Variation with Offset (AVO) Least squares Seismic attributes |
| Summary: | A new inversion method to estimate high-resolution amplitude-versus-angle attributes (AVA) attributes such as intercept and gradient from prestack data is presented. The proposed technique promotes sparse-spike reflectivities that, when convolved with the source wavelet, fit the observed data. The inversion is carried out using a hybrid two-step strategy that combines fast iterative shrinkagethresholding algorithm (FISTA) and a standard least-squares (LS) inversion. FISTA, which can be viewed as an extension of the classical gradient algorithm, provides sparse solutions by minimizing the misfit between the modeled and the observed data, and the l1-norm of the solution. FISTA is used to estimate the location in time of the main reflectors. Then, LS is used to retrieve the appropriate reflectivity amplitudes that honor the data. FISTA, like other iterative solvers for l1-norm regularization, does not require matrices in explicit form, making it easy to apply, economic in computational terms, and adequate for solving large-scale problems. As a consequence, the FISTA+LS strategy represents a simple and cost-effective new procedure to solve the AVA inversion problem. Results on synthetic and field data show that the proposed hybrid method can obtain highresolution AVA attributes from noisy observations, making it an interesting alternative to conventional methods. |
|---|