---

Seismic Denoising and Stacking Using Non-Linear Principal Component Analysis.

Ramesh A, Satyavani N.

GEOPHYSICAL PROSPECTING

https://doi.org/10.1111/1365-2478.70148 

Conventional linear approaches may struggle with nonlinearity and imperfect NMO corrections, therefore in this study, Non-Linear PCA (NLPCA)- based similarity-weighted stacking method is introduced using auto-associative neural networks to better capture the complex behaviour of seismic signals. The proposed approach demonstrates Improved SNR, Enhanced reflector continuity, Better fault imaging in complex geology, and reduced information loss in NMO-corrected gathers.

Fig:(a) Conventional NMO stack, (b) PCA stack, and (c) NLPCA stack for complex geology.

---