US 4,543,632 AGranted

Robust estimation method for determining when subsequent data processing can include sign-bit representations of full-waveform seismic traces

AssigneeChevron Research Company

InventorsRaymond A. Ergas, Francis Muir

Life of this patent

Estimating the term…

Loading patent…

Abstract

A method is disclosed for determining if sign-clipped variations of full-waveform traces can be used in subsequent cross-correlation processing without undue hardship. In accordance with one aspect, the present invention examines skewness of the amplitude spectrum of the full waveform. If the former significantly differs from conventional Gaussian distribution over the frequency range of interest, i.e., the amplitude spectrum is non-Gaussian, sign-clipped versions thereof can be used in subsequent cross-correlation processing without undue loss in processing accuracy. In another aspect, the present invention is used to edit a series of traces, one at a time, gathered by a large multichannel collection system. In this aspect, the different spectral noise estimates, including cross-correlation estimates generated from sign-clipped versions of the full-waveform traces, are advantageously used.

Claims11

Dependent claims

A robust method of examining field traces T0, T1, T2 . . . T associated with a collection survey providing N fold subsurface common depth point coverage of an earth formation, to determine if sign-clipped versions of said traces T0 . . . T can be used in generating cross-correlation estimates in subsequent seismic processing irrespective of the fact that full-waveform versions of said traces T0 . . . T are available for such processing:

(i) determining a time window i for examining each trace T0 . . . T, said window defining a trace segment thereof containing a plurality of sample points associated with a dependent variable that varies as a function of one or more independent variables;
(ii) determining skewness of said dependent variable of said trace T0 . . . T as a function of said one or more independent variables, over said time window i one trace-at-a-time;
(iii) comparing said determined skewness of step (ii) with a Gaussian distribution of said same variable as a function of said same one or more independent variables;
and
(iv) if said skewness of step (iii) deviates substantially from that of said Gaussian distribution, sign clipping said trace T0 . . . T over at least said window i wherein a resulting sign-clipped version can be used in generating cross-correlation estimates in a subsequent seismic processing method within undue loss in accuracy, said substantial deviation of said skewness of step (ii) from that of said Gaussian distribution being such as to define a shift in magnitude of the peak independent variable of at least 50% from that associated with said Gaussian distribution.

2 more elements

A robust method of editing field traces T0, T1, T2 . . . T associated with a large multichannel seismic collection survey collecting M channels of data per collection cycle so as to provide N fold subsurface common depth point coverage of an earth formation using part sign-clipped versions of said traces T0 . . . T without need of either full trace examination along the full-time scale of said trace, or without examination of adjacent traces associated with each trace T0 . . . T, comprising:

(i) determining a time window for examining each trace T0 . . . T, said window defining a full-wave trace segment thereof containing a plurality of sample points defining a signal-to-noise ratio below unity and being associated with a dependent variable that varies as a function of one or more independent variables;
(ii) sign clipping said independent variable associated with said full-wave trace segment over said time window i thereby forming a sign-clipped version thereof;
(iii) generating a series of statistical functions including cross-correlation estimates F1 . . . F associated with said sign-clipped independent variable of step (ii) over said time window i for said trace T0 . . . T;
(iv) comparing said generated statistical functions of step (iii) with a series of similar values associated with a standard trace which experience indicates if the limits thereof are exceeded, is unacceptable for further processing;
and
(v) if said generated limits of step (iv) exceed said standard limits of said standard trace, removing said trace from further seismic processing, said removal occurring without the need of examining other time segments of said trace and without examining adjacent traces associated with said trace T0 . . . T.

3 more elements

Dependencies are inferred from the claim text.