CN1797029A - Method for pressing regular noise in 3D earthquake data before superposition - Google Patents

Abstract

The invention is a method for eliminating regular noises from 3D prestack seismic data to raise the data S/N ratio in the seismic data processing course, firstly analyzing channel head information in seismic data; adopting multichannel information analysis and single-channel one-point calculation modes and scan- stacking measure to identify the regular noises; compressing the identified regular noises; smoothening the regular noises. It adopts dichotomy classification scan-superposing technique to identify noises and adopts point-by-point automatic identification method, able to automatically adapt to any change of visual speed of the regular noises and identify and eliminate the regular noises.

Description

The drawing method of organized noise in the three-dimensional earthquake data before superposition

Technical field

The present invention relates to the drawing method of organized noise in a kind of three-dimensional earthquake data before superposition, belong to and improve data signal to noise ratio (S/N ratio) technology category in the geological data processing procedure.

Background technology

In the seismic exploration data that gather in the area of surface conditions more complicated, often comprise stronger organized noise.In two-dimension earthquake data big gun collection or common detector gather, organized noise is linear lineups form with various tilt profiles; In 3D seismic data big gun collection record, the lineups of organized noise are the hyperbolic curve form and distribute.

When geological data was handled, the organized noise of strong energy directly influenced deconvolution effect, the analysis precision that underspeeds, and finally reduce the signal to noise ratio (S/N ratio) of superposition of data.Although can suppress organized noise effectively by stack,, when its energy is very strong, still can residual stronger organized noise in the poststack data.This is because when relying on stack to suppress organized noise and expectation fully to obtain desirable effect, requires to have higher degree of covering, and in practice, especially this point often is difficult to reach in 3D seismic data.Therefore, for the regular very noisy that exists in the original seismic data,, can reduce the signal to noise ratio (S/N ratio) of superposition of data greatly if can not before stack, effectively suppress.

Usually eliminated organized noise by F-K filtering at prestack, but the spinoff that this method produces is bigger in the past.In addition, this method also has harsh application prerequisite, concentrates number variation insufficient and that organized noise is subjected to stratigraphic dip to influence generation in road all can obviously reduce its effect as change sight, road.Particularly when the prestack road concentrates the energy of organized noise more intense, when distribution range is relatively wider, the effect of using F-K filtering is undesirable.

The lineups of organized noise are the hyperbolic curve form in the 3-D seismics record, and especially in the bigger arrangement of perpendicular offset, Hyperbolic Feature is more obvious.

If still the conventional method at organized noise in the two-dimentional prestack seismogram of employing is suppressed the organized noise in the three-dimensional earthquake data before superposition, then apparent velocity can not be subjected to effective compacting than higher organized noise in the near offset trace.

Summary of the invention

Technical matters to be solved by this invention is that organized noise is considered as having Hyperbolic Feature, and a kind of method that can effectively detect and suppress organized noise in the three-dimensional prestack seismogram is provided, and improves the processing quality of 3D seismic data.

The detection and the drawing method of organized noise in the three-dimensional earthquake data before superposition that the present invention relates to, concrete treatment step comprises:

(1) at first analyzes the road header of geological data;

(2) the measure recognition rule noise that adopts multiple tracks information analysis, single track single-point calculation mode and scanning to superpose;

(3) suppress the organized noise that identifies;

(4) organized noise is carried out smoothing processing.

The detection and the drawing method of organized noise in the three-dimensional earthquake data before superposition of the present invention are defined as organized noise and have Hyperbolic Feature.

The detection and the drawing method of organized noise in the three-dimensional earthquake data before superposition of the present invention are considered the influence of recording geometry to the organized noise distributional pattern when the identification noise.For the two-dimensional observation system, organized noise is similar in earthquake data before superposition and is the linear distribution feature; And for stereo observing system, owing to there is the influence of the non-advance of shot point, organized noise is the hyperbola distribution feature in earthquake data before superposition.

The detection and the drawing method of organized noise in the three-dimensional earthquake data before superposition of the present invention, the organized noise of described compacting identification is the method that adopts intermediate value or predictive filtering.

The detection and the drawing method of organized noise in the three-dimensional earthquake data before superposition of the present invention, adopted dichotomy scanning superimposing technique identification noise, adopt the pointwise method of identification automatically, can adapt to the variation of organized noise apparent velocity automatically, organized noise is identified and deducts.

Specific implementation principle of the present invention is as follows:

Prestack is write down (i more arbitrarily ₀, j ₀), the scanning stack power is

$E_{j_0}\left(k\right)=\underset{j=j_0-\frac{M}{2}}{\overset{j=j_0+\frac{M}{2}}{\mathrm{\Σ}}}\underset{i=i_0-\frac{L}{2}}{\overset{i_0+\frac{L}{2}}{\mathrm{\Σ}}}A(i,j,k)$ k∈[k1，k2]

In the formula, (i, j k) are amplitude to A, and (k1, k2) is the The dip sweep scope, and i is the time-sampling point, and j is a Taoist monastic name corresponding in the scope of search volume.Calculate a series of scanning energy value E (k) corresponding to different angle.Get

E _max＝max{E(k)} k∈[k1，k2]

Then the K value of Emax correspondence is point (i ₀, j ₀) the lineups inclination angle located.

The lineups inclination angle scope of supposing given organized noise is [a1, a2], order

$P=\left\{\begin{array}{cc}1& K_{\max }\∈\[a1,a2\]\\ 0& K_{\max }\∉\[a1,a2\]\end{array}\right.$

When P=1, be illustrated in point (i ₀, j ₀) detected lineups are organized noise; When P=0, be illustrated in point (i ₀, j ₀) do not detect organized noise.

Determine that K is point (i ₀, j ₀) behind the lineups inclination angle of organized noise, can select for use the method for medium filtering or predictive filtering to eliminate organized noise.

The method principle of medium filtering of the present invention is as follows.

Inclination direction along the K correspondence is got intermediate value, has

$V_m=\mathrm{MED}\left\{A(i_0,j-\frac{k}{2},\…\…,j+\frac{k}{2})\right\}$

Order

$A\text{'}(i_0,j_0)=\left\{\begin{array}{cc}A(i_0,j_0)& p=0\\ A(i_0,j_0)-V_m& p=1\end{array}\right.$

A ' (i ₀, j ₀) be the result behind the filtering organized noise.

The lineups of organized noise are the distribution of hyperbolic curve form because stereo observing system causes in the 3-D seismics record, and this can't change the feature that organized noise is had in the prestack record.If unification is considered according to stereo observing system when the recognition rule noise, promptly when the identification inclination angle, consider the time difference of following formula, the two-dimension earthquake data are a special case of 3D seismic data, this moment h=0, then the organized noise in the three-dimensional prestack seismogram can be considered as also have linear feature, and can carry out the identification and the compacting of noise according to this feature.

$\mathrm{\δt}=k^{*}(\sqrt{h^2+x^2}-x)$

In the formula, k is the apparent slowness of organized noise, and x is the projection of geophone offset on line direction, and h is non-advance.

The present invention adopts the mode of three-dimensional space-time territory prediction hyperbolic curve rule noise to discern and suppress organized noise in the three-dimensional prestack seismogram, has a significant effect, and mainly shows as:

(1) the present invention can effectively predict the hyperbolic curve rule noise in the 3-D seismics record;

(2) the present invention designs according to the characteristics of three-dimensional pre stack data, has originality;

(3) the present invention is very little to the influence of usable reflection signal in application.

Description of drawings

Fig. 1 considers the influence of recording geometry to the organized noise form

Fig. 1 (a) is the synoptic diagram of organized noise in the two dimension record;

Fig. 1 (b) is the synoptic diagram of organized noise in the three-dimensional record;

Fig. 2 is the The trial result of the theoretical three-dimensional single shot record that synthesizes,

Wherein (a) synthesizes three-dimensional single shot record, comprise organized noise and reflection line-ups two parts, (b) carry out result after the organized noise compression process, be the organized noise that the hyperbolic curve form distributes and obtained effective compacting, (c) organized noise of Qu Chuing (a and b's is poor);

Fig. 3 is the The trial result of certain actual 3-D seismics big gun collection record,

Wherein (a) is original single shot record, wherein there are the stronger organized noises of various energy such as refraction, secondary disturbances, ground roll, (b) for carrying out the result after organized noise is suppressed, the strong energy rule noise that is the distribution of hyperbolic curve form has obtained effective compacting basically, (c) organized noise (a and b's is poor) for removing;

Fig. 4 is a stacked section making the preceding Inline direction of three-dimensional prestack organized noise compacting, owing to be subjected to the influence of strong organized noise, its signal to noise ratio (S/N ratio) is very low;

Fig. 5 is the stacked section of doing after the prestack organized noise is suppressed, owing to before stack strong energy rule noise is compared effective compacting, stacked section is obviously weakened by the influence of organized noise, and the signal to noise ratio (S/N ratio) of the entire profile is significantly improved.

Embodiment

The detection and the drawing method of organized noise in the three-dimensional earthquake data before superposition of the present invention are defined as organized noise and have Hyperbolic Feature.

Analyze the road header of geological data earlier, consider the influence of recording geometry, determine the characteristics that organized noise has, if the two-dimension earthquake data are just calculated according to the characteristics of organized noise in the two-dimension earthquake data to the organized noise distributional pattern; If 3D seismic data just calculates according to the characteristics of organized noise in the 3D seismic data.

According to Fig. 1, carry out multi-trace statistical analysis in the horizontal, on vertical (time), carry out the single track single-point and calculate, promptly get multiple tracks (2n+1) road and carry out statistical study, interim orbit (n+1) road is calculated, the road on both sides as a reference, to carry out the stack of dichotomy energy scan to the result of calculation twice adjacent in middle (n+1) road, judge whether to meet the feature of organized noise then, if meet with both sides, it then is organized noise, it is identified, if do not meet, then is effective information.Library track rolls forward with calculating the road then, repeats said method recognition rule noise, identify organized noise after, adopt the method for medium filtering or predictive filtering to suppress, and organized noise carried out smoothing processing.

When implementing the compacting organized noise, at first sample along its distribution arrangement in earthquake data before superposition.When adopting the medium filtering mode, the value of these sampling points is carried out the intermediate value analysis, the value with sampling point deducts intermediate value as new sample value then.When adopting the mode of predictive filtering, the value of these sampling points is carried out predictive filtering, get the predictive filtering error as new sample value.

Shown in Figure 6 is a test data.

At first in the data more arbitrarily, utilize the mode of scanning stack power and curve fitting to carry out multi-trace statistical analysis, promptly get 7 roads and carry out statistical study, interim orbit (the 4th road) is for calculating the road, each 3 road, both sides are as the reference road.With interim orbit (the 4th road) is calculation level, and 6 adjacent roads of both sides carry out the stack of dichotomy energy scan, promptly can identify the inclination angle of this point of interim orbit.All roads all roll forward together then, repeat aforementioned calculation, can identify two groups of inclination angles in the data, i.e. inclination angle 1 and inclination angle 2.

This test data is set the inclination angle scope in advance and be organized noise in-8ms/ road to 1ms/ road, and remaining is useful signal.The actual measurement situation is: inclination angle 1 is-the 2ms/ road that inclination angle 2 is the 1.5ms/ road.Therefore, inclination angle 1 belongs to organized noise, should suppress.

1 carries out medium filtering along the inclination angle, promptly to data along the inclination angle 1 direction value, and calculate the intermediate value of these sample value, again sample value is deducted intermediate value as new sample value.

Parameter and amplitude analysis result that this test figure is adopted are as follows:

The input data: theoretical test figure is total up to 64 roads, 512 sampling points of per pass.

Processing parameter: the compacting time difference range is-3.5-1 (millisecond/road); Getting 7 roads adds up

Calculate; Adopt the mode of medium filtering to calculate.

Amplitude analysis: 0.11% of the original amplitude of residue amplitude average out at inclination angle 1; Remaining of inclination angle 2

99.83% of the original amplitude of surplus amplitude average out to.

Claims (5)

1, the drawing method of organized noise in the three-dimensional earthquake data before superposition, treatment step comprises:

(1) at first analyzes the road header of geological data;

(2) the measure recognition rule noise that adopts multiple tracks information analysis, single track single-point calculation mode and scanning to superpose;

(3) suppress the organized noise that identifies;

(4) organized noise is carried out smoothing processing;

2, the drawing method of organized noise in the three-dimensional earthquake data before superposition as claimed in claim 1 is characterized in that, considers the influence of recording geometry to the organized noise distributional pattern when the identification noise.

3, the drawing method of organized noise in the three-dimensional earthquake data before superposition as claimed in claim 2, it is characterized in that, organized noise in the 3-D seismics record is defined as has Hyperbolic Feature, organized noise in the two-dimension earthquake record is the special case of organized noise in the 3-D seismics record, promptly has linear feature.

4, the drawing method of organized noise in the three-dimensional earthquake data before superposition as claimed in claim 1 is characterized in that the organized noise that described compacting identifies is the method that adopts intermediate value or predictive filtering.

5, the drawing method of organized noise in the three-dimensional earthquake data before superposition as claimed in claim 1 is characterized in that having adopted dichotomy scanning superimposing technique identification noise.

Images