Automatic Tracking of Faults by Slope Decomposition

• US 20150117151A1
• Filed: 09/15/2014
• Published: 04/30/2015
• Est. Priority Date: 10/31/2013
• Status: Active Grant

• Current Assignee: Ethan Nowak, Fuxian Song, William Curry
• Original Assignee: Ethan Nowak, Fuxian Song, William Curry
• Inventors: Nowak, Ethan, Curry, William, Song, Fuxian
• Granted Patent: US 10,393,899 B2
• Time in Patent Office: Days
• US Class Current: 367/43
• CPC Class Codes: G01V 1/30 Analysis G01V1/50 takes precedence
• CPC Class Codes: G01V 1/301 for determining seismic cross-sections or geostructures
• CPC Class Codes: G01V 2210/60 Analysis

Abstract

Method for locating fault lines or surfaces in 2-D or 3-D seismic data based on the fact that fault discontinuities in the space domain span a wide range in a local slowness (slope) domain, whereas other dipping events in the space domain data, such as noise, tend to be coherent, and hence will appear focused in the slowness dimension. Therefore, the method comprises decomposing the seismic data (102) by a transformation to the local slowness domain, preferably using Gaussian slowness period packets as the local slowness or slope decomposition technique, thereby avoiding problems with the data stationary assumption. In the local slowness domain, faults may be identified (104) using the principle mentioned above, i.e. that faults are represented as a truncation in the space domain data, hence they will appear broadband in the slowness dimension.

Claims

1. A computer-implemented method for automatically tracking faults in a 2-D or 3D seismic data volume or cross-section, comprising:

(a) decomposing the seismic data into local slownesses for seismic data recorded in time, or slopes if the seismic data have undergone depth migration, using a programmed computer; and

(b) identifying voxels in the seismic data which when decomposed into slownesses or slopes span a broader range of slownesses or slopes than other voxels in the seismic data, thereby forming a fault-highlighted data volume or cross-section.

2. The method of claim 1, further comprising:

(c) selecting one or more initial seeds for fault surfaces or fault lines within the fault-highlighted data volume or cross-section based on the fault highlights;

(d) generating one or more fault contours in the fault-highlighted data volume or cross-section starting from the initial seeds; and

(e) outputting a connected, smooth fault surface or line based on the one or more fault contours.

3. The method of claim 2, wherein the generating of fault contours is performed by an active contouring method.

4. The method of claim 2, wherein (a)-(e) are performed using a programmed computer, with the initial seeds being selected and input by a user.

5. The method of claim 1, wherein the decomposing of the seismic data is performed using one of curvelets, a local radon decomposition, and wave atoms.

6. The method of claim 1, wherein the decomposing of the seismic data is performed using a cascade of 1D Gaussian filters modulated by complex exponentials.

7. The method of claim 6, wherein the decomposing of the seismic data comprises:

using a cascade of filtering operations to decompose the seismic data into components in a frequency-wavenumber domain wherein the seismic data are represented in terms of tiled windows.

8. The method of claim 7, wherein the cascade of filtering operations is a series of 1D Gaussian filters modulated by complex exponentials, creating a series of Gaussian windows in the frequency-wavenumber domain.

9. The method of claim 8, wherein the series of modulated 1D Gaussian filters are designed to create a series of Fourier-domain windows that when combined captures a selected frequency range and slowness range of frequency and wavenumber space.

10. The method of claim 9, wherein the windows are spaced equally in frequency and slowness within said selected frequency range and slowness range.

11. The method of claim 8, wherein the seismic data are expressed in terms of depth rather than time, and the using of a cascade of 1D modulated Gaussian filters creates a decomposition that when combined captures a selected wavenumber range of k_x-k_y-k_z space.

12. The method of claim 7, wherein the tiled windows are Gaussian slowness-period packets g that may be expressed mathematically as

13. The method of claim 12, wherein said normalization factors computed to produce a flat impulse response may be expressed mathematically as