CN103901466A - Three-dimensional seismic data interpolation method - Google Patents

Abstract

The invention is a three-dimensional seismic data interpolation method which can be used to save the acquisition cost. The method comprises the following steps: transform along the time direction is carried out to obtain frequency-space domain data, and odd and even seismic data is divided into two groups; the odd and even seismic data is transformed to a frequency-wave number domain to obtain odd and even wave number domain data; division is carried out on the odd and even wave number domain data to obtain a prediction operator; the prediction operator is multiplied by even frequency data, and inverse Fourier transform is carried out on the obtained product to obtain a time-space domain interpolation result; and mean processing, linear interpolation or cubic function interpolation processing is carried out on a current computation window and a previous computation window, thus completing interpolation. Compared with a two-dimensional frequency-wave number domain interpolation, the interpolation result has better consistency, is clear in event, has no obvious jitter and has natural background, and the interpolation result is superior to that of two-dimensional frequency-wave number domain interpolation in signal and background.

Description

A kind of 3D seismic data interpolation method

Technical field

The present invention relates to geophysical exploration technology, belong in geological data processing for encrypting a kind of 3D seismic data interpolation method of geological data, raising spatial sampling rate and signal to noise ratio (S/N ratio), saving acquisition cost.

Background technology

In the gatherer process of geological data, due to the restriction of the objective condition such as construction environment and workload, usually there is the larger problem of track pitch in some direction in the seismologic record collecting, makes spatial sampling rate wretched insufficiency, and the meetings such as migration imaging processing are produced and had a strong impact on.Address this problem the most direct method and be naturally while collection in the wild and reduce spatial sampling interval by increasing survey line quantity; But the thing followed is increasing substantially of acquisition cost, and some work area is due to the restriction of surface condition practically, is difficult to improve the spatial sampling rate of field work.

Seismic trace interpolation value is not increase under the prerequisite of acquisition cost, reduces the Perfected process of track pitch.Frequency-Space Domain Trace Interpolation technology is utilized the predictability of plane wave, utilizes least square method to try to achieve predictive operator and carry out interpolation by solving Hermite system of linear equations at Frequency-Space Domain.Frequency-wavenumber domain Trace Interpolation technology, by the development of Frequency-Space Domain interpolation technique, need to not solve predictive operator at Frequency-Space Domain, and Interpolation Process all completes at frequency-wavenumber domain, thereby can increase substantially interpolation efficiency.

Existing frequency-wavenumber domain interpolation technique is two-dimensional interpolation, only processes single section when each interpolation.During for 3-D data volume application two-dimensional interpolation, only utilize the information on single section, Interpolation Process only to consider the data variation on single direction at every turn, and ignored the globality of geological data, easily caused interpolation result unstable.

Goal of the invention

The object of the present invention is to provide that a kind of interpolation result is stable, lineups are clear, the natural three-dimensional frequency-wavenumber domain geological data interpolation method of background.

Summary of the invention

The present invention realizes as follows:

1) gather 3D seismic data, do a point time, spatial window processing;

2) in current calculating window, do Fourier transform along time orientation, obtain Frequency-Space Domain data, and odd number, even number road geological data are divided into two groups;

3) by step 2) in odd numbered trace, even number road earthquake data transformation to frequency-wavenumber domain, obtain odd numbered trace, even number road wavenumber domain data; Frequency-Space Domain data transformation to frequency-wavenumber domain is obtained to even number frequency data;

4) in frequency-wavenumber domain, the odd numbered trace obtaining in step 3), even number road wavenumber domain data are divided by and are obtained predictive operator;

5) the even number frequency data that obtain in the predictive operator obtaining in step 4) and step 3) are multiplied each other, and this product is obtained to temporal-spatial field interpolation result by inverse fourier transform;

6) current calculating window and last calculating window are carried out to average value processing, linear interpolation or cubic function interpolation processing;

7) current calculating window completes interpolation, Output rusults if most end is calculated window; Calculating window if not most end skips to next calculating window and returns to step 2) circulation, until most end is calculated window.

The present invention is with respect to two-dimentional frequency-wavenumber domain interpolation, and interpolation result has better consistance, and interpolation result lineups are clear, obviously shake of nothing, background nature; Interpolation result is all better than two-dimentional frequency-wavenumber domain interpolation in signal and background.

Accompanying drawing explanation

Fig. 1: processing flow chart of the present invention.

Fig. 2: somewhere geological data section (300CMP)

Fig. 3: the present invention to this data interpolating after gained section (300CMP).

Fig. 4: two-dimentional frequency-wavenumber domain interpolation method to this data interpolating after gained section (800LINE).

Fig. 5: the present invention to this data interpolating after gained section (800LINE).

Specific embodiments

The present invention implements as follows, according to treatment scheme Fig. 1:

1) step S1, gathers 3D seismic data, is designated as s _mn(t), m=0,1,2 ..., M, n=1,2 ..., N, N represents directions X line number, and M represents Y-direction road number, and t represents the time.

2) step S2, data s step S1 being read according to space calculating window, Time Calculation window parameter _{m, n}(t) divide window processing, be designated as

i, j is representation space window, time window numbering respectively.

3) step S3 does Fourier transform along time orientation in current calculating window, obtains Frequency-Space Domain data

f represents frequency; By Frequency-Space Domain data odd numbered trace, even number road be divided into two groups, be designated as respectively

with

$m=\mathrm{0,1},\·\·\·,\frac{M}{2};$ n＝1，2,…,N。

4) step S4, by odd numbered trace, even number road geological data

with

be converted into frequency-wavenumber domain, obtain odd numbered trace, even number road wavenumber domain data, be designated as respectively

with

by Frequency-Space Domain data

be converted into frequency-wavenumber domain and obtain frequency-wavenumber domain data, be designated as

obtain even number frequency data along frequency direction interval value again, be designated as

k _x, k _yrepresent respectively the wave number of directions X and Y-direction.

5) step S5, by the odd numbered trace wavenumber domain data that obtain in step S4

divided by even number road wavenumber domain data obtain predictive operator, be designated as ?

$P_{m,n}^{i,j}(f,k_x,k_y)=S_{2m+1,n}^{i,j,\mathrm{odd}}(f,k_x,k_y)/S_{2m,n}^{i,j,\mathrm{even}}(f,k_x,k_y).$

6) step S6, by the predictive operator obtaining in step S5

with the even number frequency data that obtain in step S4

multiply each other, obtain frequency-wavenumber domain interpolation result

$S_{m,n}^{i,j,I}(f,k_x,k_y)=P_{m,n}^{i,j}(f,k_x,k_y)\·S_{m,n}^{i,j}(2f,k_x,k_y);$

Again by frequency-wavenumber domain interpolation result

obtain temporal-spatial field interpolation result by inverse fourier transform, be designated as

7) step S7, to current calculating window interpolation result and last calculating window interpolation result lap process, lap result is designated as

can adopt average value processing, that is,

$S_{m,n}^{i,j,\mathrm{overlap}}\left(t\right)=(S_{m,n}^{i,j,I}\left(t\right)+S_{m,n}^{i-1,j,I}\left(t\right))/2;$

Linear interpolation, that is,

$S_{m,n}^{i,j,\mathrm{overlap}}\left(t\right)=\mathrm{\α}\·S_{m,n}^{i,j,I}\left(t\right)+(1-\mathrm{\α})\·S_{m,n}^{i-1,j,I}\left(t\right);$

Or cubic function interpolation, that is,

$S_{m,n}^{i,j,\mathrm{overlap}}\left(t\right)={\mathrm{\α}}^3\·S_{m,n}^{i,j,I}\left(t\right)+(1-{\mathrm{\α}}^3)\·S_{m,n}^{i-1,j,I}\left(t\right)$

Here α is scale-up factor.

8) step S8, current calculating window i, j window completes interpolation if most end is calculated window, to step S9 Output rusults; Calculate window if not most end and skip to next calculating window i+1, j window also returns to step S3 circulation, until most end is calculated window.

Fig. 2, Fig. 3, Fig. 4 and Fig. 5 are the specific embodiment of the invention.Relatively can find out by Fig. 2, Fig. 3, interpolation result of the present invention and raw data have high consistency, and lineups are clear, background nature.

Can find out by Fig. 4, Fig. 5 contrast, with respect to two-dimentional frequency-wavenumber domain interpolation, interpolation result of the present invention has better consistance, and interpolation result lineups are clear, obviously shake of nothing, background nature; Interpolation result of the present invention is all better than two-dimentional frequency-wavenumber domain interpolation in signal and background.

Claims (1)

1. a 3D seismic data interpolation method, realize as follows:

1) gather 3D seismic data, do a point time, spatial window processing;

2) in current calculating window, do Fourier transform along time orientation, obtain Frequency-Space Domain data, and odd number, even number road geological data are divided into two groups;

3) by step 2) in odd numbered trace, even number road earthquake data transformation to frequency-wavenumber domain, obtain odd numbered trace, even number road wavenumber domain data; Frequency-Space Domain data transformation to frequency-wavenumber domain is obtained to even number frequency data;

4) in frequency-wavenumber domain, the odd numbered trace obtaining in step 3), even number road wavenumber domain data are divided by and are obtained predictive operator;

5) the even number frequency data that obtain in the predictive operator obtaining in step 4) and step 3) are multiplied each other, and this product is obtained to temporal-spatial field interpolation result by inverse fourier transform;

6) current calculating window and last calculating window are carried out to average value processing, linear interpolation or cubic function interpolation processing;

7) current calculating window completes interpolation, Output rusults if most end is calculated window; Calculating window if not most end skips to next calculating window and returns to step 2) circulation, until most end is calculated window.

Images