CN107102355A - The parallel Marchenko imaging methods of low-frequency reconfiguration - Google Patents
The parallel Marchenko imaging methods of low-frequency reconfiguration Download PDFInfo
- Publication number
- CN107102355A CN107102355A CN201710284885.6A CN201710284885A CN107102355A CN 107102355 A CN107102355 A CN 107102355A CN 201710284885 A CN201710284885 A CN 201710284885A CN 107102355 A CN107102355 A CN 107102355A
- Authority
- CN
- China
- Prior art keywords
- frequency
- low
- marchenko
- green
- function
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000003384 imaging method Methods 0.000 title claims abstract description 96
- 238000011084 recovery Methods 0.000 claims abstract description 6
- 230000006870 function Effects 0.000 claims description 86
- 238000000034 method Methods 0.000 claims description 66
- 238000001914 filtration Methods 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 13
- 238000004088 simulation Methods 0.000 claims description 12
- 230000000694 effects Effects 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000001228 spectrum Methods 0.000 claims description 7
- 238000000354 decomposition reaction Methods 0.000 claims description 5
- 238000013461 design Methods 0.000 claims description 3
- 238000011835 investigation Methods 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 4
- 230000035515 penetration Effects 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 6
- 238000005314 correlation function Methods 0.000 description 5
- 230000011514 reflex Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000004613 tight binding model Methods 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000036039 immunity Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
- 238000000547 structure data Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
- G01V1/30—Analysis
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Acoustics & Sound (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The present invention relates to a kind of parallel Marchenko imaging methods of low-frequency reconfiguration, it is that the passive source low-frequency information reconstruct active source low-frequency information technology based on frequency dominance has been applied in Marchenko imagings, passive source low-frequency noise can be prevented to the influence of the broadband seismic data reconstructed to greatest extent, recovery has obtained broadband seismic data, the HFS of active source data is intactly remained again, this is than artificially controlling the weight with adjustment low frequency signal in broadband seismic data more to meet reality, broadband seismic data minute gun parallel iteration after reconstruct solves focus function, complete Green's function, upload Green's function, under pass Green's function, the substantial amounts of time is saved, significantly improve the computational efficiency of Marchenko imagings.Finally in broadband seismic data Marchenko imagings, reflecting interface is more focused on, cancellation of the present invention using the stronger shielded layer of high velocity penetration capacity of low-frequency information and archenko imaging algorithms to interbed multiple is acted on, and obtains high-precision subsurface structure imaging.
Description
Technical field
The present invention relates to the passive source low-frequency reconfiguration of a kind of subsurface imaging method of seismic prospecting, especially seismic interference method
Earthquake surmounts the parallel direct wave estimation of interferometric method active source data, minute gun and data drive surface shields at a high speed subterranean to complicated
The high-precision Marchenko imagings made.
Background technology：
In seismic imaging refutation process, low-frequency information is particularly important for the structure imaging of underground medium, its direct shadow
Ring the efficiency of inverse process of underground medium fine structure.Low frequency signal has the stronger ability for penetrating shielded layer of high velocity, it is possible to use
Low frequency signal reduces scattering and shielding action of the shielded layer of high velocity to seismic signal, improves the signal to noise ratio of deep layer signal.It is actual
In, because passive source data has intrinsic low frequency advantage, often through seismic interference method to passive source data at
Reason, obtains useful low-frequency information.Seismic interference is theoretical to be proposed by Claerbout in nineteen sixty-eight, and Calvert is based in 2004
Seismic interference theory proposes " image source method ", focus can be reset to geophone station position, this method is in seismic exploration data
It is widely applied.Wapenaar demonstrated seismic interference technology in 2012 to difference with Integral Theory and reciprocal theorem
Non- attenuation medium and different focus under the conditions of set up, and verified with theoretical and real data.It is used as earthquake
The expansion of interferometric method, earthquake surmounts interferometric method and carried earliest by Kees Wapenaar and Filippo Broggini in 2012
Go out, the method is realized in the case where that need not know underground medium information in the new of the virtual focus of non-detector position structure
Method, i.e., the position of virtual focus can be distributed in the optional position of underground medium.Earthquake is surmounted interferometric method and received using earth's surface
To active source reflex response and the direct wave estimated on inaccurate smoothing model reconstruct comprising primary wave and many subwaves
Green's function.Using the obtained upload of reconstruct and pass Green's function down and carry out Marchenko imagings, can without
The accurate medium to target area of situation for solving overlying medium construction carries out accurately image.
The Green's function at geological data reconstruct underground medium lacked based on low frequency can cause final
Occur in Marchenko imaging results due to low frequency missing secondary lobe caused by subsurface reflective boundary can not focus on the problem of, this meeting
The explanation that latter earthquake is explored is impacted.There are two kinds, Yi Zhongshi to the reconstruct method that geological data low frequency is lacked in practice
Low-frequency information is obtained from the mode of mathematical computations：Han was based on compressed sensing and sparse inversion method weight in 2012 and 2014
Structure data low-frequency information；The long wave that Huang has recovered underground medium p-and s-wave velocity in 2015 with elastic wave envelope method grows up to
Point.Such low-frequency reconfiguration method may produce deceptive information in earthquake record in actual applications, final influence imaging knot
Really.Another is to carry out seismic interference method processing to passive source data, then carries out low-frequency reconfiguration to active source：Zhang in
Active source and passive source data are matched from energy point of view (amplitude) within 2015, even if the high s/n ratio after denoising is low
Still there is noise in frequency, this limitation also can produce influence to final imaging results in.In addition, such a passive source
Low frequency supplements active source data method based on manual gain's coefficient to complete, subjective, and can not be driven based on data
It is dynamic.
Each Green's function put of Marchenko imaging algorithms dependent on targeted imaging region, the grid of very little is big
It is small to cause huge amount of calculation.Traditional single-shot serial computing far can not meet the requirement of computational efficiency.
The content of the invention：
The purpose of the present invention improves low-frequency reconfiguration effect there is provided one kind for above-mentioned the deficiencies in the prior art and imaging is imitated
The parallel object-oriented data-driven Marchenko imaging methods of rate.
The purpose of the present invention is achieved through the following technical solutions：
The core of the parallel object-oriented data-driven Marchenko imaging methods of low-frequency reconfiguration is based on frequency dominance
The passive source data of seismic interference method is surmounted interferometric method active source data low-frequency reconfiguration to earthquake, managed based on PBS batch jobs
The parallel earthquake of system surmounts the reconstruct of interferometric method Green's function and the Marchenko towards complicated subsurface structure is imaged.To intricately
The accurate Marchenko imaging results of lower construction need wide band geological data as input, are primarily based on seismic interference method
The low frequency advantage of passive source data surmounts interferometric method to earthquake and carries out low-frequency reconfiguration, is then based on normalization and adds and basis afterwards
Active source data peak swing carries out energy recovery, obtains wide band geological data.Iteration asks for focus function, Complete Lattice
Woods function, the amount of calculation for uploading Green's function and passing Green's function down are huge, will be calculated using PBS batch jobs management system
Platform is divided into a primary processor and some from processor.Direct wave minute gun is estimated on the whole smoothing model of main processor monitors
In the forward simulation stage, forward modelling is assigned to each by primary processor and carried out from processor, calculated after completing again by leading
Processor collects complete Green's function reconstruction result, complete Green's function decomposition result, then submits being based on cross-correlation
Marchenko imaging processing operation calculated to from processor, now main processor monitors entirely be based on upload and under pass
Green's function calculates the virtual hypocentral location reflectance factor stage, calculates after completing, and primary processor collects obtained target area
The reflectance factor of the virtual hypocentral location in each underground, presses residing further according to the position of the virtual focus in underground to reflectance factor
Observation system position is arranged in order, and obtains final Marchenko imaging results.Underground intricately texture was so both solved
The high-precision imaging problem made, and given full play to based on the parallel forward modeling of minute gun the excellent in performance of current calculating platform, significantly contracting
The calculating time of short Marchenko imagings.
The parallel Marchenko imaging methods of low-frequency reconfiguration, comprise the following steps：
A, Torque2.4.16 is installed in Ubuntu14.04 system platforms, builds PBS batch jobs management systems；
B, the passive focus earthquake signal to actual acquisition carry out seismic interference method processing, based on cross-correlation method, reconstruct and master
The passive focus earthquake data of the dynamic identical observation system in source；
C, to passive focus earthquake data carry out multiple domain iterated denoising, obtain the passive source data compared with high s/n ratio.Compensation is gone
Passive source data energy after making an uproar, finds passive source low frequency end and intersection point of the active source low frequency end on frequency spectrum, and determine Jiao
The frequency of point position is dominant frequency；
D, with reference to low pass filter LPF is done at dominant frequency to active source, passive source respectively；
E, design matched filter, calculate matched filtering operator, the active source low-frequency information and low pass after LPF
Filtered passive source low-frequency information matched filtering, obtains the active source low-frequency information after matched filtering；
F, the active source low-frequency information and active source after matched filtering are normalized respectively, then the two is done plus and,
Active source data peak swing before being then based on active source normalization carries out energy recovery, the broadband after being synthesized
Geological data；
G, to earthquake data prediction, meet the requirement of Marchenko iterative Green's functions, the step is divided to two
The situation of kind：
In A, numerical simulation example, calculated by the method for numerical computations direct wave in broadband seismic data,
The many subwaves of Free Surface.
B, actual seismic exploration in, the cancellation of many subwaves of Free Surface can be reached by SRME methods, direct wave
Calculating can be calculated by shallow-layer investigation；
H, carry out forward modeling on inaccurate model and obtain estimating between the wave detector that underground medium point is laid to earth's surface
The direct wave response of meter.Broadband seismic data after low-frequency compensation can be imaged to underground complex geological district.
Selected underground medium due to the shielding action similar to high impedance stratum cause weaker, the conventional earthquake of reflected seismic energy into
Not accurate enough the area of image space method is used as imaging region.Assuming that imaging region has the virtual hypocentral locations of m, then be accomplished by into
M minute gun forward modeling of row.Assuming that having N number of from processor participation calculating in calculate node, then the operation of parallel computation is submitted to
It is N number of after PBS operating systems to carry out parallel computation from processor, after the forward modeling of a big gun terminates, meter is terminated from processor
Calculate, now, PBS primary processors distribute next forward modelling operation from processor in order for this, until m forward modeling is complete
Portion's result；
I, Marchenko imaging need solve imaging region all imaging points upload Green's function and under pass Green
Function.The position of the virtual focus of imaging region m correspond to m iterative Green's function.PBS primary processors will be by pre-
The direct wave that forward modeling is estimated in broadband seismic data and h steps after processing is assigned to N number of among processing.
For each single Marchenko iterative from processor, iterative is three-dimensional first
Marchenko equations obtain focus function, then the relation between complete Green's function and focus function 2, wideband data,
Complete Green's function is tried to achieve, the relation between Green's function and focus function 1 is passed according to upload, down afterwards, tries to achieve after decomposition
Upload Green's function and under pass Green's function.
When complete Green's function of Marchenko iteratives, upload Green's function and pass Green's function down and terminate,
Primary processor will continue to distribute subsequent job to from processor processing, until whole ends of job；
J, Green's function is passed by the upload Green's function obtained after decomposition and be down used for carrying out Marchenko imagings.Into
As region m virtual hypocentral locations correspond to the upload Green's function of m single virtual hypocentral location with passing Green's function down
Cross-correlation function is calculated and the reflectance factor of the position is asked for.PBS primary processors are first by the 1st, 2,3 ..., and N number of calculate is made
Industry is sequentially allocated to the 1st, 2,3 ..., N number of to be calculated from processor.When the reflectance factor of single virtual hypocentral location is calculated
After obtaining, PBS primary processors collect the reflectance factor for recording the virtual hypocentral location and record the virtual focus is in observation
Coordinate in system, then continues to distribute operation, until whole Activity Calculations completes to be idle from processor；
K, after system tries to achieve in imaging region all reflectance factors of virtual hypocentral locations.According to observation system, foundation
The sequencing of grid point coordinates in imaging region, is serially successively aligned to reflectance factor the position of mesh point, final to obtain
The data-driven Marchenko imaging results in the object-oriented region to after based on low-frequency reconfiguration.
Beneficial effect：Passive source low-frequency information based on frequency dominance is successfully reconstructed active source low frequency and believed by the present invention
Breath technology has been applied in Marchenko imagings, the active source low-frequency reconfiguration method base based on passive source low frequency advantage of proposition
Passive source low-frequency noise can be prevented to greatest extent to the influence of the broadband seismic data reconstructed in matched filter,
Both recover to have obtained broadband seismic data based on normalized energy restoration methods simultaneously, and intactly remain active source number
According to HFS, this than artificially controlling more to meet reality with weight of the adjustment low frequency signal in broadband seismic data,
And this method is based on data-driven.Being solved based on the broadband seismic data minute gun parallel iteration after reconstruct for proposing focuses on letter
Several, complete Green's function, upload Green's function, pass Green's function down, saved the substantial amounts of time, significantly improved
Marchenko imaging computational efficiency, finally broadband seismic data Marchenko imaging in subsurface reflective boundary more
Focus on, relatively conventional low-frequency compensation method, present aspect does not introduce the reflecting interface of falseness.More preferable imaging effect is obtained
Really.Present invention mainly solves problems with：
1st, seismic interference method and earthquake surmount interferometric method based on different it is assumed that of the invention respectively from the angle of numerical simulation
Degree and the angle of actual production, give the method that this two kinds of data are combined.It is imaged for passive source data in Marchenko
In application lay a good foundation.
2nd, the low frequency signal contained in the active source geological data of conventional Christmas is often weaker than its high-frequency signal, broadband
Geological data need to obtain based on low-frequency reconfiguration.Relative to pure values low frequency continuation, the low frequency signal source used in the present invention
In real geological data, so the broadband seismic data that compensation is obtained are relatively reliable；Relative to traditional based on energy
Compensation method can produce false reflection interface in final imaging results, frequency dominance proposed by the present invention ground reconstructing method
The low frequency energy of active source, and passive source can be reconstructed well on the basis of active source high-frequency signal is completely retained
Noise signal final imaging results will not be impacted.
3rd, traditional seismic method to the region containing high impedance superstratum often due to strong scattering, at a high speed shielding make
With causing, illumination deficiency, imaging effect are poor.The present invention make use of the stronger shielded layer of high velocity penetration capacity of low-frequency information simultaneously
The cancellation of interbed multiple is acted on Marchenko imaging algorithms, high-precision subsurface structure imaging results are obtained.
4th, each Green's function put of Marchenko imaging algorithms dependent on targeted imaging region, the grid of very little
Size can cause huge amount of calculation, and calculating operation is assigned to by PBS batch jobs management system and carried out from processor
Calculate, linear speed-up ratio can be obtained.
Brief description of the drawings
The parallel Marchenko imaging methods flow chart of Fig. 1 low-frequency reconfigurations.
Fig. 2 illustratons of model
(left side) rate pattern,
(right side) density model
Illustraton of model inaccurate Fig. 3
Rate pattern after (left side) is smooth,
Density model after (right side) is smooth
Fig. 4 targeted imaging regions schematic diagram
(left side) targeted imaging region,
Targeted imaging region after (right side) amplification
Single shot record after the passive source cross-correlation seismic interference method processing of Fig. 5
Broadband single shot record after Fig. 6 active sources single shot record and low-frequency reconfiguration
(left side) active source single shot record,
Broadband single shot record after (right side) low-frequency reconfiguration.
Waveform is taken out in Fig. 7 low-frequency reconfiguration Hou Chou road waveforms and Whole frequency band forward modeling
Broadband single-shot frequency spectrum and Whole frequency band forward modeling single-shot frequency spectrum after Fig. 8 low frequencies missing single-shot frequency spectrum, low-frequency reconfiguration
The Green's function of Fig. 9 active source data reconstructions
The Green's function reconstructed after Figure 10 active source low-frequency reconfigurations
Figure 11 refers to Green's function
The Green's function that is reconstructed after the Green's functions of Figure 12 active source data reconstructions, active source low-frequency reconfiguration and refer to lattice
Woods function takes out trace comparison
Figure 13 Marchenko imaging results
(left side) low frequency lacks the Marchenko imaging results of active source geological data,
The Marchenko imaging results of broadband seismic data after (right side) low-frequency reconfiguration.
Figure 14 same area reverse-time migration imaging results
Embodiment
Below in conjunction with the accompanying drawings with example detailed description further to the present invention
The parallel Marchenko imaging methods of low-frequency reconfiguration, comprise the following steps：
A, program are to write completion based on C language, parallel in order to carry out minute gun, are pacified in Ubuntu14.04 system platforms
Torque2.4.16 is filled, PBS batch jobs management systems are built；
B, the passive focus earthquake signal to actual acquisition carry out seismic interference method processing, based on following cross-correlation method, reconstruct
The passive focus earthquake data of observation system identical with active source, in order to strengthen the signal to noise ratio of geological data after reconstruct, it is necessary to long
Temporally receive passive source signal.
WhereinObserved wave field at respectively A, B location,Represent xBFor focus, xA
For the Green's function of wave detector,For power spectrum,Real part is represented, c and ρ represent spread speed and density respectively；
C, matched filtering need to input the geological data of high s/n ratio, it is necessary to change to passive focus earthquake data progress multiple domain
The passive source data compared with high s/n ratio is obtained for denoising.The passive source data energy after denoising is compensated, passive source low frequency is found
End and intersection point of the active source low frequency end on frequency spectrum, and determine the frequency of focal position for dominant frequency fcut；
D, with reference to low pass filter active source is done LPF obtain the active source geological data after low frequency filtering, it is right
LPF is done at dominant frequency and obtains the passive focus earthquake data after low frequency filtering in passive source：
E, design matched filter, calculate matched filtering operators m (t), the active source low-frequency information after LPF with
Passive source low-frequency information matched filtering after LPF, obtains the active source low-frequency information after matched filtering：
F, the active source low-frequency information and active source after matched filtering are normalized respectively, then the two is done plus and,
Active source data peak swing before being then based on active source normalization carries out energy recovery, the broadband after being synthesized
Geological data:
G, to earthquake data prediction, meet the requirement of Marchenko iterative Green's functions, the step is divided to two
The situation of kind：
A, in numerical simulation example, calculated by the method for numerical computations through in broadband seismic data
The many subwaves of ripple, Free Surface:
Assuming that F1(t) it is reflex response that earth's surface has absorbing boundary condition (such as PML), it comprises primary wave, through
Ripple and interbed multiple.F2(t) it is that the direct wave obtained in uniform dielectric numerical simulation is responded.So many subwave F of Free Surface3
(t) it can be drawn by following formula：
F3(t)=F (t)-F1(t)
Now meet the reflex response R of Marchenko imagings0It can be expressed as：
R0=F (t)-F3(t)-F2(t)
B, actual seismic exploration in, the cancellation of many subwaves of Free Surface can be reached by SRME methods, direct wave
Calculating can be calculated by shallow-layer investigation；
H, on inaccurate model (accompanying drawing 3) carry out forward modeling obtain underground medium point to earth's surface laying wave detector
Between estimation direct wave response.Estimate it is worth noting that, model now does not need accurate model to carry out direct wave
Meter.Broadband seismic data after low-frequency compensation can be imaged to underground complex geological district.Selected underground medium
Because the seismic imaging method that the shielding action similar to high impedance stratum causes reflected seismic energy weaker, conventional is not smart enough
True area is used as imaging region (accompanying drawing 4).Assuming that imaging region has m virtual hypocentral locations, then be accomplished by carrying out m times
Minute gun forward modeling.Assuming that there is N to participate in calculating from processor in calculate node, then the operation of parallel computation is submitted to PBS and made
It is N number of after industry system to carry out parallel computation from processor, after the forward modeling of a big gun terminates, terminate to calculate from processor, this
When, PBS primary processors distribute next forward modelling operation from processor in order for this, until m forward modeling is all tied
Really；
I, Marchenko imaging need solve imaging region all imaging points upload Green's function and under pass Green
Function.Upload Green's function G-, pass Green's function G down+Relation with focus function and seismic reflection data is
Wherein R is seismic reflection data, f1 +And f1 -Respectively upload and under pass focus function, t represents the time.
According to above-mentioned expression formula, it is known that focus function, can ask for upload Green's function and under pass Green's function.Focus on
Function needs to ask for three-dimensional Marchenko equations by iteration to obtain.General iterations is set to 4-15 times.
The position of the virtual focus of imaging region m correspond to m iterative Green's function and its decomposition.PBS main process tasks
The direct wave that forward modeling is estimated in broadband seismic data after pretreatment and h steps is assigned to N number of from processing by device
It is central.
For each single Marchenko iterative from processor, iterative is three-dimensional first
Marchenko equations obtain focus function, then by uploading, passing Green's function and focus function f down1 +、f1 -Between relation, ask
Must upload Green's function and under pass Green's function.
Terminate when a Marchenko iterative uploads Green's function and passes Green's function down, primary processor will continue
Subsequent job is distributed to from processor processing, until whole ends of job；
J, by the upload Green's function G obtained after decomposition-Lower biography Green's function G+For carry out Marchenko into
Picture.The virtual hypocentral location of imaging region m correspond to the upload Green's function G of m single virtual hypocentral location-Lattice are passed with
Woods function G+Cross-correlation function calculate and the reflectance factor of the position is asked for.Cross-correlation function is expressed as：
Wherein, x "0ForWave detector on face.In frequency domain, cross-correlation image-forming condition is cross-correlation function in all frequencies
Integration.
PBS primary processors are first by the 1st, 2,3 ..., and N number of calculating operation is sequentially allocated to the 1st, 2,3 ..., N number of from
Reason device is calculated.After the reflectance factor of single virtual hypocentral location, which is calculated, to be obtained, PBS primary processors collect record should
The reflectance factor of virtual hypocentral location simultaneously records the coordinate of the virtual focus in observation system, then to be idle from processing
Device continues to distribute operation, until whole Activity Calculations is completed；
K, after system tries to achieve in imaging region all reflectance factors of virtual hypocentral locations.According to observation system, foundation
The sequencing of grid point coordinates in imaging region, is serially successively aligned to reflectance factor the position of mesh point, final to obtain
The data-driven Marchenko imaging results in the object-oriented region to after based on low-frequency reconfiguration.
Embodiment 1
The calculating requirement being imaged according to seismic prospecting, Torque2.4.16 is installed under Ubuntu14.04 systems,
Carry out building for PBS parallel tables.
Tested using the speed and density model (accompanying drawing 2) that shield folded strata at a high speed containing high impedance.Model is joined
Number is as follows：
Model meshes size is 801 × 2401, and grid is away from dz=dx=2.5m, and lateral separation is 6.0km, longitudinal depth
For 2.0km, seimic wave velocity scope is from 1.9km/s to 2.8km/s in model, and geophone is placed in model surface, altogether
601 between wave detector, and wave detector at intervals of 10m, the same detector position in active source forward simulation surface seismic source position,
Have 601 focus.The wave detector sampling interval is 0.004s.
In order to meet cross-correlation method reconstruct it is assumed that active source and the earth's surface in passive source are a Free Surfaces.First
Passive source is simulated on model, in order to obtain the passive source reconstruction result of high s/n ratio, the simulation noise source of underground 4500 is random
It is distributed in laterally (- 2900m, 2900m), in the range of longitudinal direction (1350m, 1700m), wave detector receives 1200s passive source number
According to.Passive source reconstruct is carried out by the reconstruct of conventional cross-correlation, the earthquake record of the virtual focus of earth's surface (0m, 0m) position is for example attached
Fig. 5.Then active source simulation is carried out, simulation obtains the weaker active source geological data (accompanying drawing 6 of below 20Hz low frequency signals
It is left).According to frequency dominance low-frequency reconfiguration method proposed by the present invention, the energy compensating to passive source is first passed around, 20Hz is determined
For dominant frequency.With reference to low pass filter, 20Hz LPFs are carried out to active source and passive source signal respectively, respectively obtained
Active source data and passive source data after low frequency filtering.Active source data now is gone to match passive source data：It is first
Matched filtering operator is first solved, the active source data for then acting on matched filtering operator after LPF is reconstructed
The active source low-frequency information gone out.Then in conjunction with method for normalizing, energy recovery is carried out to geological data, obtain containing direct wave,
The broadband seismic data (accompanying drawing 6 is right) of many subwaves of Free Surface, interbed multiple, from accompanying drawing 7, the wideband after reconstruct
The geological data obtained with geological data and Whole frequency band forward modeling meets ground very well.To the broadband after active source data, reconstruct
Shake data and Whole frequency band forward modeling data carry out spectrum analysis (accompanying drawing 8) and understand that the radio-frequency component in original active source is obtained
It is effectively maintained, while understanding that low frequency signal has obtained effective reconstruct by the spectral trends of low frequency end.
It is not include direct wave, the data of many subwaves of Free Surface that earthquake, which surmounts interferometric method reconstruct Green's function to input,
Need to pre-process data.Based on numerical simulation preprocess method proposed by the present invention, broadband seismic data are carried out
Surmount the input of interferometric method after pretreatment as earthquake.Estimation to direct wave carries out (accompanying drawing 3) on inaccurate model.
Then targeted imaging region (accompanying drawing 4) asked for based on parallel form iteration complete Green's function, upload Green's function, under
Pass Green's function.Given iterations is now needed, general iterations is set as between 4-15.(0,1000) master at place
The Green's function (accompanying drawing 10) that is reconstructed after the Green's function (accompanying drawing 9) of dynamic source data reconstruct, active source low-frequency compensation, with reference to lattice
Woods function (accompanying drawing 11) is as shown in drawings.Trace comparison is taken out as shown in the figure to this three kinds of Green's functions, it can be seen that low-frequency reconfiguration it
Green's function afterwards with reference to Green's function with more conforming to.Marchenko imaging results and low frequency weight are lacked by contrasting low frequency
Structure Marchenko imaging results, it will be seen that the reflecting interface of low-frequency reconfiguration more focuses on (accompanying drawing 13 is right), model
In the interface of fold with shielding at a high speed obtained good imaging, deep layer imaging of interface is clear.And in low frequency missing
In Marchenko imagings (accompanying drawing 13 is left), because low frequency is lacked in the reflex response of input, side lobe effect is caused to have impact on
Marchenko imaging effects.Final result verification effectiveness of the invention.Due in final imaging results not by
The false reflection interface that low frequency matched filtering is introduced, so the method proposed in the present invention has noise immunity.Meanwhile, in phase
With region and traditional imaging method imaging results contrasted knowable to the method that is proposed of this method it is excellent in imaging
Gesture.
The performance detection environment of table 1
The monokaryon of table 2 and multi-core parallel concurrent computational efficiency comparing result
| From processor core calculation | Calculate time-consuming | CPU usage | Maximum uses internal memory (MB) |
| 1 | 24 points 29 seconds | 26.1% | 91.5MB |
| 2 | 14 points 53 seconds | 52.1% | 184MB |
| 4 | 7 points 10 seconds | 98.6% | 370MB |
Using the test environment based on table 1, we have carried out three groups of tests to Marchenko imaging processes respectively.Respectively
Calculate the Marchenko imaging results of the virtual hypocentral location in 11 undergrounds.Test result is shown in Table 2, as can be seen from Table 2, in bag
The process of Green's function and the imaging based on cross-correlation function is estimated, solved containing direct wave, with the increase of core amounts,
Under the test environment shown in table 1, obtain close to linear speed-up ratio.
Fig. 1 is the parallel Marchenko Irnaging procedures figure of whole low-frequency reconfiguration, from flow chart it can be seen that low-frequency reconfiguration simultaneously
Row Marchenko methods are divided into following steps：(1) observation system identical with active source is reconstructed based on cross-correlation method
Passive focus earthquake data；(2) the passive source data energy after compensation denoising, determines dominant frequency f_cut；(3) low pass is combined
Filtering carries out active source low frequency and passive source low frequency matched filtering；(4) normalized energy matching synthesis broadband seismic data；
(5) data prediction is carried out, eliminating Free Surface using method for numerical simulation repeatedly involves itself and estimation after direct wave
The virtual focus in underground to the direct wave between earth's surface wave detector as input, and based on Marchenko Equation Iterative Parallel implementation lattice
Woods function；(6) cross-correlation image-forming condition is finally based on, the reflectance factor of all imaging point positions of target area is asked in circulation
Complete Marchenko imagings.
Claims (2)
1. a kind of parallel Marchenko imaging methods of low-frequency reconfiguration, it is characterised in that the passive source low-frequency reconfiguration of seismic interference method
Shake surmounts the parallel direct wave estimation of interferometric method active source data, minute gun and data drive surface shields at a high speed subsurface structure to complicated
High-precision Marchenko imagings, comprise the following steps：
A, Torque2.4.16 is installed in Ubuntu14.04 system platforms, builds PBS batch jobs management systems；
B, the passive focus earthquake signal to actual acquisition carry out seismic interference method processing, based on cross-correlation method, reconstruct and active source
The passive focus earthquake data of identical observation system；
C, to passive focus earthquake data carry out multiple domain iterated denoising, obtain the passive source data compared with high s/n ratio；Compensate denoising it
Passive source data energy afterwards, finds passive source low frequency end and intersection point of the active source low frequency end on frequency spectrum, and determine focus position
The frequency put is dominant frequency；
D, with reference to low pass filter LPF is done at dominant frequency to active source, passive source respectively；
E, design matched filter, calculate matched filtering operator, the active source low-frequency information and LPF after LPF
Passive source low-frequency information matched filtering afterwards, obtains the active source low-frequency information after matched filtering；
F, the active source low-frequency information and active source after matched filtering are normalized respectively, then the two is done plus and, then
Active source data peak swing before being normalized based on active source carries out energy recovery, the broadband seismic number after being synthesized
According to；
G, to earthquake data prediction, meet the requirement of Marchenko iterative Green's functions；
H, carry out on inaccurate model that forward modeling obtains estimating between the wave detector that underground medium point is laid to earth's surface it is straight
Up to ripple response, the broadband seismic data after low-frequency compensation are imaged to underground complex geologic body, now, the main places of PBS
Reason device distributes next forward modelling operation from processor in order for this, until the whole results of m forward modeling；
I, Marchenko imaging need solve imaging region all imaging points complete Green's function, upload Green's function and
Under pass Green's function, the virtual hypocentral location of imaging region m correspond to m iterative Green's function, as a Marchenko
The complete Green's function of iterative, upload Green's function and pass Green's function down and terminate, primary processor will continue to distribute follow-up make
Industry is given from processor processing, until whole ends of job；
J, Green's function is passed by the upload Green's function obtained after decomposition and be down used for carrying out Marchenko imagings, the main places of PBS
Reason device collects the reflectance factor for recording the virtual hypocentral location and records the coordinate of the virtual focus in observation system, Ran Houwei
Idle continues distribution operation from processor, until whole Activity Calculations is completed；
K, after system tries to achieve in imaging region all reflectance factors of virtual hypocentral locations, according to observation system, according to imaging
The sequencing of grid point coordinates in region, is serially successively aligned to reflectance factor the position of mesh point, finally gives base
Data-driven Marchenko imaging results in the object-oriented region after low-frequency reconfiguration.
2. reconstruct parallel Marchenko imaging methods according to the frequency described in claim 1, it is characterised in that pair described in g steps
Geological data is pre-processed, and meets the requirement of Marchenko iterative Green's functions, there is two kinds of situations：
In A, numerical simulation example, direct wave in broadband seismic data, freedom are calculated by the method for numerical computations
Surface-related multiple；
B, actual seismic exploration in, the cancellation of many subwaves of Free Surface is reached by SRME methods, and the calculating of direct wave is
Calculated by shallow-layer investigation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710284885.6A CN107102355B (en) | 2017-04-27 | 2017-04-27 | The parallel Marchenko imaging methods of low-frequency reconfiguration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710284885.6A CN107102355B (en) | 2017-04-27 | 2017-04-27 | The parallel Marchenko imaging methods of low-frequency reconfiguration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107102355A true CN107102355A (en) | 2017-08-29 |
| CN107102355B CN107102355B (en) | 2017-12-05 |
Family
ID=59656692
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710284885.6A Expired - Fee Related CN107102355B (en) | 2017-04-27 | 2017-04-27 | The parallel Marchenko imaging methods of low-frequency reconfiguration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107102355B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109490425A (en) * | 2018-12-18 | 2019-03-19 | 浙江大学 | A kind of passive material bass reflex coefficient measuring method based on Green's function reconfiguration technique |
| CN109738952A (en) * | 2019-01-24 | 2019-05-10 | 吉林大学 | The direct offset imaging method in passive source based on full waveform inversion driving |
| CN110058300A (en) * | 2018-10-30 | 2019-07-26 | 南方科技大学 | Primary wave reconstruction method and device, terminal equipment and storage medium |
| CN111505718A (en) * | 2020-04-28 | 2020-08-07 | 西安交通大学 | High-resolution underground structure amplitude-preserving imaging method |
| CN112444867A (en) * | 2019-08-30 | 2021-03-05 | 中国石油化工股份有限公司 | Seismic data frequency band expansion processing method based on compressed sensing |
| CN112987099A (en) * | 2021-04-19 | 2021-06-18 | 吉林大学 | Low-frequency seismic data reconstruction method based on multi-seismic-source convolutional neural network |
| CN113391351A (en) * | 2021-06-10 | 2021-09-14 | 中国地质科学院 | Method for analyzing and extracting ore-concentrated area structure based on passive source seismic wave field |
| CN113534243A (en) * | 2021-07-15 | 2021-10-22 | 中国地质科学院 | Marchenko imaging method and system for passive source |
| CN113866821A (en) * | 2021-09-26 | 2021-12-31 | 吉林大学 | Passive source interference offset imaging method and system based on illumination direction constraint |
| CN118033746A (en) * | 2024-04-07 | 2024-05-14 | 吉林大学 | Object-oriented marine streamer seismic data Marchenko imaging method |
| CN118131322A (en) * | 2024-05-06 | 2024-06-04 | 吉林大学 | Micro-seismic source positioning method based on Marchenko imaging |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110255371A1 (en) * | 2009-01-09 | 2011-10-20 | Charlie Jing | Hydrocarbon Detection With Passive Seismic Data |
| CN104335072A (en) * | 2012-02-06 | 2015-02-04 | 离子地球物理公司 | Integrated passive and active seismic surveying using multiple arrays |
| CN106291684A (en) * | 2015-06-06 | 2017-01-04 | 中国石油化工股份有限公司 | The seismic response of a kind of blind focus earthquake wave field recovers and virtual source road collection construction method |
-
2017
- 2017-04-27 CN CN201710284885.6A patent/CN107102355B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110255371A1 (en) * | 2009-01-09 | 2011-10-20 | Charlie Jing | Hydrocarbon Detection With Passive Seismic Data |
| CN104335072A (en) * | 2012-02-06 | 2015-02-04 | 离子地球物理公司 | Integrated passive and active seismic surveying using multiple arrays |
| CN106291684A (en) * | 2015-06-06 | 2017-01-04 | 中国石油化工股份有限公司 | The seismic response of a kind of blind focus earthquake wave field recovers and virtual source road collection construction method |
Non-Patent Citations (2)
| Title |
|---|
| ZHANG PAN等: "Passive-source multitaper-spectral method based low-frequency data reconstruction for active seismic sources", 《APPLIED GEOPHYSICS》 * |
| 张盼等: "主动源与被动源地震数据插值及联合数据成像", 《地球物理学报》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110058300A (en) * | 2018-10-30 | 2019-07-26 | 南方科技大学 | Primary wave reconstruction method and device, terminal equipment and storage medium |
| CN109490425B (en) * | 2018-12-18 | 2020-05-19 | 浙江大学 | Passive material low-frequency reflection coefficient measuring method based on Green function reconstruction technology |
| CN109490425A (en) * | 2018-12-18 | 2019-03-19 | 浙江大学 | A kind of passive material bass reflex coefficient measuring method based on Green's function reconfiguration technique |
| CN109738952A (en) * | 2019-01-24 | 2019-05-10 | 吉林大学 | The direct offset imaging method in passive source based on full waveform inversion driving |
| CN109738952B (en) * | 2019-01-24 | 2020-05-08 | 吉林大学 | Passive source direct offset imaging method based on full waveform inversion driving |
| CN112444867A (en) * | 2019-08-30 | 2021-03-05 | 中国石油化工股份有限公司 | Seismic data frequency band expansion processing method based on compressed sensing |
| CN111505718B (en) * | 2020-04-28 | 2021-11-19 | 西安交通大学 | High-resolution underground structure amplitude-preserving imaging method |
| CN111505718A (en) * | 2020-04-28 | 2020-08-07 | 西安交通大学 | High-resolution underground structure amplitude-preserving imaging method |
| CN112987099A (en) * | 2021-04-19 | 2021-06-18 | 吉林大学 | Low-frequency seismic data reconstruction method based on multi-seismic-source convolutional neural network |
| CN113391351A (en) * | 2021-06-10 | 2021-09-14 | 中国地质科学院 | Method for analyzing and extracting ore-concentrated area structure based on passive source seismic wave field |
| CN113391351B (en) * | 2021-06-10 | 2024-02-13 | 中国地质科学院 | Method for extracting mine collection area structure based on passive source seismic wave field analysis |
| CN113534243A (en) * | 2021-07-15 | 2021-10-22 | 中国地质科学院 | Marchenko imaging method and system for passive source |
| CN113534243B (en) * | 2021-07-15 | 2022-02-15 | 中国地质科学院 | Marchenko imaging method and system for passive source |
| CN113866821A (en) * | 2021-09-26 | 2021-12-31 | 吉林大学 | Passive source interference offset imaging method and system based on illumination direction constraint |
| CN113866821B (en) * | 2021-09-26 | 2022-08-02 | 吉林大学 | Passive source interference offset imaging method and system based on illumination direction constraint |
| CN118033746A (en) * | 2024-04-07 | 2024-05-14 | 吉林大学 | Object-oriented marine streamer seismic data Marchenko imaging method |
| CN118033746B (en) * | 2024-04-07 | 2024-07-09 | 吉林大学 | Object-oriented marine streamer seismic data Marchenko imaging method |
| CN118131322A (en) * | 2024-05-06 | 2024-06-04 | 吉林大学 | Micro-seismic source positioning method based on Marchenko imaging |
| CN118131322B (en) * | 2024-05-06 | 2024-08-13 | 吉林大学 | Micro-seismic source positioning method based on Marchenko imaging |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107102355B (en) | 2017-12-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107102355B (en) | The parallel Marchenko imaging methods of low-frequency reconfiguration | |
| CN108037526B (en) | Reverse-time migration method based on all-wave wave field VSP/RVSP seismic data | |
| CN108345031B (en) | Full waveform inversion method for elastic medium active source and passive source mixed mining seismic data | |
| US8953411B2 (en) | Apparatus and method for imaging a subsurface using frequency-domain elastic reverse-time migration | |
| US9910174B2 (en) | Seismic imaging apparatus and method for performing iterative application of direct waveform inversion | |
| Pan et al. | Love-wave waveform inversion in time domain for shallow shear-wave velocity | |
| CN100545679C (en) | The earthquake poststack forward modeling method of relief surface | |
| CN111158049B (en) | Seismic reverse time migration imaging method based on scattering integration method | |
| CN106526677B (en) | A kind of wideband reverse-time migration imaging method of marine adaptive compacting ghost reflection | |
| US9766357B2 (en) | Seismic image dip decomposition estimation and recomposition | |
| CN107894613B (en) | Elastic wave vector imaging method, device, storage medium and equipment | |
| CN107765308B (en) | Reconstruct low-frequency data frequency domain full waveform inversion method based on convolution thought Yu accurate focus | |
| CN107505654A (en) | Full waveform inversion method based on earthquake record integration | |
| CN113534243B (en) | Marchenko imaging method and system for passive source | |
| CN108845317B (en) | Frequency domain reverse time migration method based on layered medium Green function | |
| CN104199088B (en) | Incident angle gather extraction method and system | |
| Xu et al. | Noniterative least-squares reverse time migration based on an efficient asymptotic Hessian/point-spread function estimation | |
| CN111505714A (en) | Elastic wave direct envelope inversion method based on rock physical constraint | |
| CN110967734A (en) | Virtual source reconstruction method and system based on fast Fourier transform | |
| CN115993641B (en) | Method for extracting passive source surface wave dispersion curve | |
| CN113866821B (en) | Passive source interference offset imaging method and system based on illumination direction constraint | |
| US8391564B2 (en) | Providing an imaging operator for imaging a subterranean structure | |
| CN115421195B (en) | Method, device, equipment and storage medium for generating velocity field in seismic exploration | |
| CN106199692A (en) | A kind of wave equation inverse migration method based on GPU | |
| Yang et al. | Frequency-domain multi-scale early-arrival waveform tomography with a time-domain wavefield modeling engine |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171205 Termination date: 20200427 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |