Polynomial Filter and its Application to Seismic Data

Research output: ThesisDoctoral Thesis

Standard

Bibtex - Download

@phdthesis{c2592b81c7124c95a487aae61b71736e,
title = "Polynomial Filter and its Application to Seismic Data",
abstract = "One of the main problems in land seismic datasets is that primary energy, which are reflections, are covered by ground roll. It is a non trivial task to separate primaries from ground roll, when the amplitudes of the reflections should stay unaffected. This is important, because for reservoir geophysics, the inversion of amplitudes play a key role when inferring petrophysical properties, like porosity, permeability, pore pressure and fluid saturation. Until now, simple bandpass filter or frequency - wavenumber filter (f-k) are applied to remove the ground roll from a seismic section. One of the main difficulties involved in applying those Fourier techniques to the processing of seismic data is the energy associated with the Gibbs error. This work introduces the concept of discrete polynomial basis functions to filter seismic data as global polynomial approximation and local polynomial approximation as an alternative to convential Fourier techniques. Gram polynomials were chosen, as their properties support polynomial preserving filtering, ringing free filtering and offer an ideal noise behaviour. The derivation of the Gram basis is shown and that the principle of superposition holds. For discrete local polynomial approximation, the concept of Savitzky-Golay smoothing was extended to the application to seismic data. Wavelength selective filters were implemented and tested on a 2D-3C seismic dataset. It is demonstrated, that the ground roll can be suppressed as a 1D trace by trace algorithm and as a 2D approach. For the 1D approach, the new concept of ringing free polynomial preserving filtering introduces a powerful tool for proper ground roll removal in the time domain. In the past, undersampled or highly unequally spaced seismic data was difficult to process. With this new approach, a ringing free and relative amplitude preserving filtering is possible, also for undersampled or unequally spaced data. The application of Savitzky-Golay smoothing in 2D introduces this new technique in detecting and removing waveforms in both time and spacial domain. The separation of reflective energy from ground roll was shown. The principle of discrete local polynomial smoothing was extended to high frequency noise removal and, that the method can be modified to other approaches than ground roll removal.",
keywords = "seismic, ground roll, discrete polynomial filtering, Gibbs error, reservoir characterization, amplitude preserving, Seismik, Ground Roll, Polynom Filter, Gibbs Energie, Reservoir Charakterisierung, amplitudenerhaltend",
author = "Claudia Steiner-Luckabauer",
note = "no embargo",
year = "2011",
language = "English",

}

RIS (suitable for import to EndNote) - Download

TY - BOOK

T1 - Polynomial Filter and its Application to Seismic Data

AU - Steiner-Luckabauer, Claudia

N1 - no embargo

PY - 2011

Y1 - 2011

N2 - One of the main problems in land seismic datasets is that primary energy, which are reflections, are covered by ground roll. It is a non trivial task to separate primaries from ground roll, when the amplitudes of the reflections should stay unaffected. This is important, because for reservoir geophysics, the inversion of amplitudes play a key role when inferring petrophysical properties, like porosity, permeability, pore pressure and fluid saturation. Until now, simple bandpass filter or frequency - wavenumber filter (f-k) are applied to remove the ground roll from a seismic section. One of the main difficulties involved in applying those Fourier techniques to the processing of seismic data is the energy associated with the Gibbs error. This work introduces the concept of discrete polynomial basis functions to filter seismic data as global polynomial approximation and local polynomial approximation as an alternative to convential Fourier techniques. Gram polynomials were chosen, as their properties support polynomial preserving filtering, ringing free filtering and offer an ideal noise behaviour. The derivation of the Gram basis is shown and that the principle of superposition holds. For discrete local polynomial approximation, the concept of Savitzky-Golay smoothing was extended to the application to seismic data. Wavelength selective filters were implemented and tested on a 2D-3C seismic dataset. It is demonstrated, that the ground roll can be suppressed as a 1D trace by trace algorithm and as a 2D approach. For the 1D approach, the new concept of ringing free polynomial preserving filtering introduces a powerful tool for proper ground roll removal in the time domain. In the past, undersampled or highly unequally spaced seismic data was difficult to process. With this new approach, a ringing free and relative amplitude preserving filtering is possible, also for undersampled or unequally spaced data. The application of Savitzky-Golay smoothing in 2D introduces this new technique in detecting and removing waveforms in both time and spacial domain. The separation of reflective energy from ground roll was shown. The principle of discrete local polynomial smoothing was extended to high frequency noise removal and, that the method can be modified to other approaches than ground roll removal.

AB - One of the main problems in land seismic datasets is that primary energy, which are reflections, are covered by ground roll. It is a non trivial task to separate primaries from ground roll, when the amplitudes of the reflections should stay unaffected. This is important, because for reservoir geophysics, the inversion of amplitudes play a key role when inferring petrophysical properties, like porosity, permeability, pore pressure and fluid saturation. Until now, simple bandpass filter or frequency - wavenumber filter (f-k) are applied to remove the ground roll from a seismic section. One of the main difficulties involved in applying those Fourier techniques to the processing of seismic data is the energy associated with the Gibbs error. This work introduces the concept of discrete polynomial basis functions to filter seismic data as global polynomial approximation and local polynomial approximation as an alternative to convential Fourier techniques. Gram polynomials were chosen, as their properties support polynomial preserving filtering, ringing free filtering and offer an ideal noise behaviour. The derivation of the Gram basis is shown and that the principle of superposition holds. For discrete local polynomial approximation, the concept of Savitzky-Golay smoothing was extended to the application to seismic data. Wavelength selective filters were implemented and tested on a 2D-3C seismic dataset. It is demonstrated, that the ground roll can be suppressed as a 1D trace by trace algorithm and as a 2D approach. For the 1D approach, the new concept of ringing free polynomial preserving filtering introduces a powerful tool for proper ground roll removal in the time domain. In the past, undersampled or highly unequally spaced seismic data was difficult to process. With this new approach, a ringing free and relative amplitude preserving filtering is possible, also for undersampled or unequally spaced data. The application of Savitzky-Golay smoothing in 2D introduces this new technique in detecting and removing waveforms in both time and spacial domain. The separation of reflective energy from ground roll was shown. The principle of discrete local polynomial smoothing was extended to high frequency noise removal and, that the method can be modified to other approaches than ground roll removal.

KW - seismic

KW - ground roll

KW - discrete polynomial filtering

KW - Gibbs error

KW - reservoir characterization

KW - amplitude preserving

KW - Seismik

KW - Ground Roll

KW - Polynom Filter

KW - Gibbs Energie

KW - Reservoir Charakterisierung

KW - amplitudenerhaltend

M3 - Doctoral Thesis

ER -