CN103472481B - A kind of GPU of utilization carries out the method that angular-trace gather is extracted in reverse-time migration - Google Patents
A kind of GPU of utilization carries out the method that angular-trace gather is extracted in reverse-time migration Download PDFInfo
- Publication number
- CN103472481B CN103472481B CN201210184592.8A CN201210184592A CN103472481B CN 103472481 B CN103472481 B CN 103472481B CN 201210184592 A CN201210184592 A CN 201210184592A CN 103472481 B CN103472481 B CN 103472481B
- Authority
- CN
- China
- Prior art keywords
- wave field
- angular
- reverse
- trace gather
- time
- 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.)
- Active
Links
Landscapes
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention provides a kind of method that GPU of utilization carries out the extraction of reverse-time migration angular-trace gather, belong to seismic prospecting and the development field of petroleum industry.The inventive method comprises the following steps: the continuation of (1) shot point wave field forward; (2) executed in parallel reverse-time migration and angle information statistics; (3) angular-trace gather extracts.Present approach provides a kind of reverse-time migration angular-trace gather extracting method of fast and stable, the ability of its tool reverse-time migration and angle information statistics parallel practice, be more suitable for the realization of GPU program, solve the problem of reverse-time migration in the past and the necessary serial execution of angular-trace gather extraction, solve counting yield low, storage demand is excessive, and fault-tolerant ability is weak, not easily the shortcoming of repetition test.
Description
Technical field
The invention belongs to seismic prospecting and the development field of petroleum industry, be specifically related to a kind of GPU of utilization and carry out the method that angular-trace gather is extracted in reverse-time migration, extract for the common picture point road collection in the migration imaging software systems of artificial earthquake record.
Background technology
The English full name of GPU is GraphicProcessingUnit, translator of Chinese is " graphic process unit ", it be from CPU (central processing unit) be derived a concept, it is the heart of video card, carry internal memory with video card plate and form a subsystem, the common graphics process performance determining PC system.Increasing graphical application makes GPU become more and more important in the computing machine in modern times, this chip CPU dominate PC performance move towards decades after nearly 2 years in emerge rapidly, its Floating-point Computation ability has reached ten times even tens times of the CPU same period, at high-performance computing sector, almost reach the status on an equal footing with CPU.
Graphic process unit (GPU) has been become a hot issue at present for the problem solving general-purpose computations (GPGPU) and related fields thereof.In fact, whole IT industry has all appreciated GPU general-purpose computations observantly will bring revolutionary change to PC, and then have influence on the development of CPU.GPU aims at image procossing design, and its storage system is actually a two-dimentional fragmented storage space, comprises a segment number (therefrom reading images) and two-dimensional address (X in image, Y-coordinate).In addition, without any indirect write command.Write address output is determined by raster processor, and can not by routine change.This is great challenge for the algorithm of NATURAL DISTRIBUTION among storer, and the access efficiency of storer is the key issue of boosting algorithm efficiency.Utilized GPU to do high-performance calculation is very difficult work in the past, and the appearance of the high-level programming languages such as later CUDA and OpenCL in 2007 makes programming new hand also can grasp the calculated performance advantage of GPU.
In the application of reverse-time migration, basic research work is exactly the extraction of angular-trace gather.Since entering 21 century, wave equation exports reflection angle image set altogether just becomes the focus of industry research.Its implementation can be divided into two large classes: a class is the extracting method (as shown in Figure 1) before imaging, hereinafter be called method one, the method needs to calculate wave field angle in real time in the process of reverse-time migration, and export angular-trace gather in real time, the indivisible realization of three (as shown in Figure 5, Fig. 5 is the refinement flow process of Fig. 1, wherein angular-trace gather extraction and wave field extrapolation are coupling in a module and perform), and needing the memory headroom of a large amount of (TB level) as guarantee, is the implacable condition of current computer technology; Another kind of is extracting method (as shown in Figure 2) after imaging, is hereinafter called method two, and the method adopts translation or time shift effect condition first to generate offset distance picture point road collection altogether, then waits method migration to generate angular-trace gather by Tau-P conversion.Extracting method after imaging is suitable for offset module in the past, only needs to do less change and follow-up conversion process just can obtain angular-trace gather, is also to use more method at present.But, because the transformational relation of usual offset distance and angle can not accomplish one_to_one corresponding, so equally also there is illusion in the angular-trace gather generated.
It is even four-dimensional that seismic data acquisition of today has developed into three-dimensional, its data volume is very huge especially, reached TB level even the prestack list big gun data of 1OTB level, the computing time of so huge data and IO time are also the difficult problems that reverse-time migration angular-trace gather extraction work must face.
Due to the develop rapidly of GPU high-performance calculation, the problem of reverse-time migration and angular-trace gather calculating can be increased substantially, but still do not solve storage problem in the implementation process of prior art.
Summary of the invention
The object of the invention is to solve the difficult problem existed in above-mentioned prior art, provide a kind of GPU of utilization to carry out the method for reverse-time migration angular-trace gather extraction, angular-trace gather can be extracted in fast and stable ground.
The present invention is achieved by the following technical solutions:
Utilize GPU to carry out a method for reverse-time migration angular-trace gather extraction, said method comprising the steps of:
(1) shot point wave field forward continuation;
(2) executed in parallel reverse-time migration and angle information statistics;
(3) angular-trace gather extracts.
Described step (1) comprises the following steps:
(A1) artificial wavelet focus is put in sp location;
(A2) utilize the wave field of finite difference method to the seismic event that the artificial wavelet epicenter excitation in step (A1) goes out to carry out forward continuation along time orientation, obtain not shot point wave field in the same time;
Described step (A2) completes on GPU.
In described step (2), described reverse-time migration comprises the following steps:
(B1) the not shot point wave field in the same time obtained in step (A2) is stored in wavefield data Buffer Pool;
(B2) the shot point wave field that will obtain from step (A2), along the continuation of time orientation forward to the maximum moment, obtains the wave field in shot point maximum moment; Forward continuation refers to the direction continuation wave field increased progressively along the time;
(B3) from the geological data of field acquisition, read the wave field in geophone station maximum moment, then from the maximum moment, by shot point wave field and geophone station wave field along time orientation backward extension, obtain shot point and geophone station not wave field in the same time; Backward extension refers to the direction continuation wave field successively decreased along the time;
(B4) the shot point wave field in each moment is multiplied with geophone station wave field obtains imaging results;
(B5) the corresponding single big gun of each shot point, is stored in the described imaging results in step (B4) corresponding for each single big gun separately on disk as single big gun migration imaging result;
(B6) each geological data from field acquisition comprises a lot of single big gun, and the described imaging results in the step (B4) utilizing each single big gun corresponding does the migration imaging result that the superposition of many big guns obtains the superposition of many big guns; Described many big gun superpositions refer to and the imaging value of multiple single big gun imaging results for same position place, underground are added up.
In described step (2), described angle information statistics comprises the following steps:
(C1) angle information statistics: the shot point wave field in described wavefield data Buffer Pool is stored according to time sequencing, and extracts amplitude and the propagation angle of each locus of each moment;
(C2) after completing steps (A2), using the shot point wave field that obtains in step (C1) the propagation angle in each locus peak swing moment as Angle Information data body store storage on disk;
(C3), in the migration imaging result that the described many big guns obtained in described step (B6) superpose, extract stratigraphic dip and be stored on disk as stratigraphic dip information data body.
Wherein, described step (B2) to (B5) and step (C1) to (C2) executed in parallel, perform step (C3) again after step (B6) is finished.
Described step (3) comprises the following steps:
(D1) the Angle Information data body obtained from read step disk (C2) and (C3) and stratigraphic dip information data body;
(D2) incident angle is calculated according to described Angle Information data body and stratigraphic dip information data body:
Incidence angle θ equals shot point wave field and reaches the propagation angle β in peak swing moment and the difference of stratigraphic dip α half at each locus place, and equation expression is: θ=β-0.5 α;
(D3) from single big gun migration imaging result that read step disk (B5) exports, then the road that single big gun migration imaging result outputs to the incident angle obtained corresponding to step (D2) is concentrated, obtain final angular-trace gather data volume.
Data Structures in described Data buffer comprises a real-coded GA, an integer data and a floating type array, and wherein real-coded GA storage current time value, integer data store the length of described floating-point array, the wave field of floating type storage of array current time.
Compared with prior art, the invention has the beneficial effects as follows:
(1) utilize data buffering pool technology in the present invention, achieve the parallelization embodiment of reverse-time migration, angle information statistics.Avoid in the past that the two can only the method for serial implementation, improve counting yield.
(2), because the angle information of shot point wave field adds up generation before imaging, under maintaining the angle calculation precision conditions before imaging, the extraction work carrying out angular-trace gather after reverse-time migration completes again is realized.Reverse-time migration and angular-trace gather in method one can be reduced like this and extract the memory cost simultaneously carried out, simultaneously can after imaging, angular-trace gather extracts in removing method two image artifacts.
(3) this patent is owing to only adding up the angle information in wave field ceiling capacity moment, so identical internal memory or disk space can be left in not angle calculation result in the same time, realize the recycling of storage space, significantly can reduce the demand of internal memory and disk like this.
Accompanying drawing explanation
Fig. 1 is the step block diagram of method one of the prior art.
Fig. 2 is the step block diagram of method two of the prior art.
Fig. 3 is that the present invention utilizes GPU reverse-time migration to extract the function structure chart designed in the method for angular-trace gather.
Fig. 4 is that the present invention utilizes GPU to extract the detailed step block diagram of reverse-time migration angular-trace gather method.
Fig. 5 is the detailed step block diagram that employing method one extracts angular-trace gather.
Fig. 6 is that the present invention calculates the schematic diagram of incident angle according to shot point wave field propagation angle and stratigraphic dip.
Fig. 7 is the schematic diagram that in the present invention, single big gun descends an imaging point incident angle accordingly.
Fig. 8 is sigseeb2a illustraton of model.
Fig. 9-1 utilizes the inventive method to carry out angular-trace gather to the illustraton of model of Fig. 8 to extract the design sketch obtained.
Fig. 9-2 is that the illustraton of model of Application way a pair Fig. 8 carries out angular-trace gather and extracts the design sketch obtained.
Fig. 9-3 is that the illustraton of model of Application way two couples of Fig. 8 carries out angular-trace gather and extracts the design sketch obtained.
Embodiment
Below in conjunction with accompanying drawing, the present invention is described in further detail:
The method that the present invention is directed to GPU reverse-time migration extraction angular-trace gather comprises three modules: reverse-time migration module (() in Fig. 4), angle information statistical module ((two) in Fig. 4) and angular-trace gather extraction module ((three) in Fig. 4), the segmentation of module is as Fig. 4, each step can realize with CPU and GPU, but efficiency can improve greatly on GPU, only have the needs utilizing GPU could meet commercial Application.Fig. 3 is the module map that CPU/GPU cooperated computing realizes the extraction of reverse-time migration angular-trace gather.Wherein reverse-time migration module and angle information statistical module can parallel practices, for angular-trace gather extraction module provides necessary input data; Shot point wave field forward continuation step in reverse-time migration module needs for Data buffer provides wavefield data, and angle information statistical module directly reads wavefield data from Data buffer.
Fig. 4 is the detail flowchart (() in Fig. 4 of modular concurrent process, (2), (3) be represent which module is concrete steps belong to), wherein white background represents reverse-time migration flow process (belonging to prior art), dark-background represents that the implementing procedure of angle information statistical module (comprises wavefield data Buffer Pool, angle information is added up, the propagation angle in peak swing moment calculates and calculates with the stratigraphic dip of output and imaging section and export), single big gun imaging results of reverse-time migration is utilized after these two flow processs complete, the peak swing Angle Information data body of locus and the stratigraphic dip information extraction angular-trace gather of imaging section.Idiographic flow is as follows:
(1) shot point wave field forward continuation
(2) executed in parallel reverse-time migration and angle information statistics
(3) angular-trace gather extracts.
In described step (1), the continuation of described shot point wave field forward comprises the following steps:
(A1) artificial wavelet focus is put in sp location;
(A2) utilize the continuation of finite difference method forward, not shot point wave field in the same time can be obtained.This is prior art, must perform on GPU, because execution just can reach satisfied counting yield on GPU, meets commercial Application requirement; Described finite difference method is one of algorithm of basic wave field extrapolation;
In described step (2), described reverse-time migration comprises the following steps:
(B1) the not shot point wave field in the same time obtained in (A2) step is stored in wavefield data Buffer Pool; In angle information statistical module, in order to ensure to implement with the parallelization of reverse-time migration module, present invention employs the intermediate link of wavefield data Buffer Pool, the not wave field information in the same time that shot point wave field is mainly just being drilled in process by this wavefield data Buffer Pool is temporarily stored on internal memory or disk, for the statistical work of angle information.The wave field extrapolation that wherein executed in parallel should comprise on GPU walks abreast, and the parallelization of angle information statistical module on CPU or on another GPU performs.The extraction work of angular-trace gather can be realized like this on the basis not affecting original reverse-time migration efficiency;
(B2) by the shot point wave field that obtains from step (A2) along time orientation forward continuation (be also called and just drill) to the maximum moment, obtain the wave field in shot point maximum moment;
(B3) from the geological data of field acquisition, read the wave field in geophone station maximum moment, then from the maximum moment, by shot point wave field and geophone station wave field backward extension, obtain shot point and geophone station not wave field in the same time (because the image-forming condition of reverse-time migration is along time reversal integration, so need imaging in back-propagation process, if the wave field in each moment is stored in forward continuation process, need a lot of disk spaces, the demand of tens bases to hundreds of TB can be reached during three-dimensional work area, backwards calculation once can avoid this to store, this is existing technology), backward extension just refers to the direction continuation wave field successively decreased along the time,
(B4) the shot point wave field in each moment is multiplied with geophone station wave field obtains imaging results (the application dependent imaging condition namely in Fig. 4); Wave field is its physical meaning of function of room and time is amplitude.
(B5) the corresponding single big gun of each shot point, is stored in the described imaging results in step (B4) corresponding for each single big gun separately on disk as single big gun migration imaging result;
(B6) each geological data from field acquisition comprises a lot of single big gun, utilizes the described imaging results of each single big gun to make many big guns and superposes the migration imaging result obtaining the superposition of many big guns; Described many big gun superpositions refer to that the imaging value for underground same position adds up multiple single big gun imaging results.
In described step (2), described angle information statistics comprises the following steps:
(C1) angle information statistics: the wave field in described wavefield data Buffer Pool is stored according to time sequencing, angle information extraction module directly accesses the wave field in wavefield data Buffer Pool, and extract amplitude and the propagation angle (extracting method can be Poynting vector method or local wavelength-division solution, and this is prior art) of each locus of each moment, when in wavefield data Buffer Pool, the number in wave field moment is greater than 1, first then wave field amplitude does not more in the same time extract the propagation angle in peak swing moment, add up each locus to reach the propagation angle in peak swing moment (in calculating, the time is discretize, embody with equally spaced moment sequence, can wave field be temporarily stored in Buffer Pool in continuation process, in order to add up angle information, the angle information in each moment can be covered by the wave field in next moment after having added up, if when the speed of angle information statistics is slower than continuation speed, the wave field in multiple moment can be there is in Buffer Pool simultaneously, now need more existingly to be in Buffer Pool that wave field amplitude is not in the same time to select the maximum wave field of wherein amplitude to calculate angle information, add up rear Buffer Pool to empty, wait for that the wave field of later point enters Buffer Pool, carry out identical statistics again.)。Wave field is the function of room and time, and the result of statistics is also the function in space.
The implementation method of angle information statistics is also important protection content of the present invention, the Data Structures of wavefield data Buffer Pool comprises a real-coded GA, an integer data and a floating type array, wherein floating data stores current time value, (what the continuation of wave field forward referred to is exactly the direction continuation wave field increased progressively along the time for the wave field of floating-point storage of array current time, its result is not wave field in the same time, the wave field so comprised in wavefield data Buffer Pool retains to calculate angle, each moment calculates and once can be override by subsequent time, so needing to indicate calculated angle is that moment, namely described current time.), integer data stores the length of floating-point array.Not wave field value in the same time directly reads by angle information statistical module from wavefield data Buffer Pool, then calculates amplitude and the propagation angle of each locus of each moment.It is to be noted, first compare the angle information that then amplitude extract peak swing when the wave field moment is more in wavefield data Buffer Pool and (wave field in multiple moment in Buffer Pool, may be there is simultaneously, now need to compare amplitude), in order to follow-up angular-trace gather extraction work.The synchronism with the wave field extrapolation in reverse-time migration can be ensured like this, and without any need for communication, only need access the time in wavefield data Buffer Pool and wave field.
(C2) completing steps (A2) afterwards (namely after whole moment continuation), using the shot point wave field obtained in step (C1) each locus peak swing moment propagation angle as Angle Information data body store (because can not ensure that in C1 implementation all moment comprise, must all moment continuation be complete just can obtain this Angle Information data body.) be stored on disk; A part for the input data that described Angle Information data body extracts as follow-up angular-trace gather;
(C3), in the migration imaging result that the many big guns obtained in described step (B6) superpose, extract stratigraphic dip and be stored on disk as stratigraphic dip information data body; A part for the input data that described stratigraphic dip information data body extracts as follow-up angular-trace gather;
Wherein, step (B2) to (B5) can with (C1) to (C2) executed in parallel, (B6) performs C3 after being finished again;
In described step (3), described angular-trace gather extracts and comprises the following steps:
(D1) the Angle Information data body obtained from read step disk (C2) and (C3) and stratigraphic dip information data body;
(D2) incident angle is calculated according to described Angle Information data body and stratigraphic dip information data body: what angle information statistical module of the present invention finally preserved the is shot point wave field stratigraphic dip after each locus place reaches the wave field propagation angle in peak swing moment and reverse-time migration many big guns stacking image.Utilize this two angle informations, calculate that the imaging value that the incident angle information of seismic event extracts in imaging results generates angular-trace gather.Concrete projectional technique is (in Fig. 6, S represents shot point, R represents geophone station) as shown in Figure 6, incidence angle θ equals shot point wave field and reaches the wave field propagation angle β in peak swing moment and the difference of stratigraphic dip (being also called reflection line-ups inclination angle) α half at each locus place, and equation expression is: θ=β-0.5 α.
(D3) from single big gun migration imaging result that read step disk (B5) exports, then the road that single big gun migration imaging result outputs to the incident angle obtained corresponding to step (D2) is concentrated, obtain final angular-trace gather data volume.It is self-assembling formation that Fig. 7 describes the corresponding different incident angle of different single big gun.
In angular-trace gather abstraction module, each imaging point needs the information of different incidence angles, and the present invention is mapped as different single emplacements different incidence angles and puts, (X as shown in Figure 7
srepresent sp location coordinate), the corresponding incident angle of imaging results of single big gun data, the imaging results (result that the single big gun migration result namely in step (B5) stores separately) of single big gun data is preserved respectively, as the input data of follow-up angular-trace gather extraction module in implementation process.
Described step (A2) completes on GPU, and other step completes on GPU or CPU.The present invention mainly emphasizes that the division methods of this flow process is conducive to parallel and raises the efficiency, and the finite difference formulations amount of (A2) step very large needs uses GPU, but angle extraction can be taked GPU or CPU completes for distinct methods.
Be described effect of the present invention using international SEG master pattern sigseeb2a (as shown in Figure 8) as instance data of the present invention below, the earthquake work area information of this model is as shown in table 1.
| Big gun number | 500 big guns |
| Sampling interval | 8ms |
| Sampling length | 12s |
| Single big gun number of channels | 1067 roads |
| Migration aperture | 35 kilometers, maximum diameter of hole (diameter) |
| Deviation frequency | Maximum frequency 105HZ |
| Sizing grid | Grid 2133 × 1200 |
| Bin size | Bin 37.5 × 25inch |
Table 1
Its hardware environment used is as shown in table 2.
| Host number | 1 |
| Tesla S1070 quantity | 1 |
| Intel CPU overall performance | 4 cores × 2.0GHZ=8GHZ |
| GPU overall performance | 4×240×1.44GHZ＝1382GHZ |
| Host memory | 24GB |
| GPU video memory | 16GB |
| Hard disk | 2TB |
| Network environment | 1000M b/s |
| Operating system | LIN UX CentOS 4.8 |
Table 2
The angle intervals exported is 1 degree, and export 180 angles, utilize the inventive method and method one, method two to realize angular-trace gather respectively and extract, the time used is as shown in table 3.
| The inventive method | Method one | Method two | |
| Computing time | 2.5 hour | 5 hours | 4.8 hour |
| Required memory space | 78MB | 13.7GB | 3.8GB |
| Disk space required | 6.5GB | 576GB | 160GB |
Table 3
As can be seen from Table 3, the calculating that the inventive method is used and storage resources minimum.
Method one is compared with the inventive method, and the rise of computing time is due to reverse-time migration part in method one and angular-trace gather Extraction parts serial computing, often does a step wave field extrapolation and all will wait for that completing of angular-trace gather extraction work just can carry out next step continuation.The rise of method one disk space required and memory headroom is that the Output rusults of each big gun is all one and is angular-trace gather owing to needing for angular-trace gather opening space reverse-time migration while, and it is equal with output angle number that it increases multiple.The angular-trace gather that Application way one calculates and come to the same thing (see accompanying drawing 9-1 and 9-2) that the inventive method calculates.
Method two is compared with the inventive method, and the rise of computing time is that the increase of its calculated amount is the use of translation image-forming condition because method two needs the imaging value of multiple offset distance in reverse-time migration process.Disk space needed for method two and the rise of memory headroom are owing to needing in reverse-time migration process to export common offset road collection, it is relevant to exported offset distance number that it increases multiple, the migration result of each single big gun is an offset gather, and its rise multiple is the number of offset distance.The present embodiment adopts output 50 offset distances to calculate.The illusion that the present invention extracts angular-trace gather will significantly reduce, Fig. 9-3 is angular-trace gather extraction effects that Application way two obtains, Fig. 9-1 is the angular-trace gather extraction effect utilizing the inventive method to obtain, comparison diagram 9-1 and Fig. 9-3 is visible, the angular-trace gather that the present invention extracts all is evened up substantially at the road collection at reflecting interface place, there is illusion hardly.
The innovative point of this patent is as follows:
(1) innovation of the extraction scheme of angular-trace gather: the present invention is the advantage in conjunction with the angular-trace gather extracting method of (method two) after (method one) before imaging in prior art and imaging, under inheriting the prerequisite of the computational accuracy of angle Information Statistics before imaging, the method that after achieving imaging, angular-trace gather extracts.
(2) introduce the parallel computation that Data buffer achieves angle information statistics part and reverse-time migration part, improve counting yield.
(3) the independent storage of stratigraphic dip data volume in the independent storage of angle-data body in the independent storage of single big gun migration result in reverse-time migration module, angle information statistical module, angle information statistical module, ensure that the feasibility of carrying out angular-trace gather extraction work after reverse-time migration completes again, and avoid reverse-time migration in the past and angular-trace gather to extract the memory cost simultaneously carried out.
(4) this patent is owing to only adding up the angle information in wave field ceiling capacity moment, so be placed on identical internal memory or disk space in not angle calculation result in the same time, realize the recycling of storage space, significantly can reduce the demand of internal memory and disk like this.
Technique scheme is one embodiment of the present invention, for those skilled in the art, on the basis that the invention discloses application process and principle, be easy to make various types of improvement or distortion, and the method be not limited only to described by the above-mentioned embodiment of the present invention, therefore previously described mode is just preferred, and does not have restrictive meaning.
Claims (4)
1. utilize GPU to carry out a method for reverse-time migration angular-trace gather extraction, it is characterized in that: said method comprising the steps of:
(1) shot point wave field forward continuation;
(2) executed in parallel reverse-time migration and angle information statistics;
(3) angular-trace gather extracts;
Wherein, step (1) comprises the following steps:
(A1) artificial wavelet focus is put in sp location;
(A2) utilize the wave field of finite difference method to the seismic event that the artificial wavelet epicenter excitation in step (A1) goes out to carry out forward continuation along time orientation, obtain not shot point wave field in the same time;
Step (A2) completes on GPU;
Wherein, in step (2), described reverse-time migration comprises the following steps:
(B1) the not shot point wave field in the same time obtained in step (A2) is stored in wavefield data Buffer Pool;
(B2) the shot point wave field that will obtain from step (A2), along the continuation of time orientation forward to the maximum moment, obtains the wave field in shot point maximum moment; Forward continuation refers to the direction continuation wave field increased progressively along the time;
(B3) from the geological data of field acquisition, read the wave field in geophone station maximum moment, then from the maximum moment, by shot point wave field and geophone station wave field along time orientation backward extension, obtain shot point and geophone station not wave field in the same time; Backward extension refers to the direction continuation wave field successively decreased along the time;
(B4) the shot point wave field in each moment is multiplied with geophone station wave field obtains imaging results;
(B5) the corresponding single big gun of each shot point, is stored in the described imaging results in step (B4) corresponding for each single big gun separately on disk as single big gun migration imaging result;
(B6) each geological data from field acquisition comprises a lot of single big gun, and the described imaging results in the step (B4) utilizing each single big gun corresponding does the migration imaging result that the superposition of many big guns obtains the superposition of many big guns; Described many big gun superpositions refer to and the imaging value of multiple single big gun imaging results for same position place, underground are added up;
Wherein, in step (2), described angle information statistics comprises the following steps:
(C1) angle information statistics: the shot point wave field in described wavefield data Buffer Pool is stored according to time sequencing, and extracts amplitude and the propagation angle of each locus of each moment;
(C2) after completing steps (A2), using the shot point wave field that obtains in step (C1) the propagation angle in each locus peak swing moment as Angle Information data body store storage on disk;
(C3), in the migration imaging result that the described many big guns obtained in step (B6) superpose, extract stratigraphic dip and be stored on disk as stratigraphic dip information data body.
2. the GPU of utilization according to claim 1 carries out the method for reverse-time migration angular-trace gather extraction, it is characterized in that: step (B2) to (B5) and step (C1) to (C2) executed in parallel, perform step (C3) again after step (B6) is finished.
3. the GPU of utilization according to claim 2 carries out the method for reverse-time migration angular-trace gather extraction, it is characterized in that: step (3) comprising:
(D1) the Angle Information data body obtained from read step disk (C2) and (C3) and stratigraphic dip information data body;
(D2) incident angle is calculated according to described Angle Information data body and stratigraphic dip information data body:
Incidence angle θ equals shot point wave field and reaches the propagation angle β in peak swing moment and the difference of stratigraphic dip α half at each locus place, and equation expression is: θ=β-0.5 α;
(D3) from single big gun migration imaging result that read step disk (B5) exports, then the road that single big gun migration imaging result outputs to the incident angle obtained corresponding to step (D2) is concentrated, obtain final angular-trace gather data volume.
4. the GPU of utilization according to claim 1 carries out the method for reverse-time migration angular-trace gather extraction, it is characterized in that: the Data Structures in described Data buffer comprises a real-coded GA, an integer data and a floating type array, wherein real-coded GA storage current time value, integer data store the length of described floating-point array, the wave field of floating type storage of array current time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210184592.8A CN103472481B (en) | 2012-06-06 | 2012-06-06 | A kind of GPU of utilization carries out the method that angular-trace gather is extracted in reverse-time migration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210184592.8A CN103472481B (en) | 2012-06-06 | 2012-06-06 | A kind of GPU of utilization carries out the method that angular-trace gather is extracted in reverse-time migration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103472481A CN103472481A (en) | 2013-12-25 |
| CN103472481B true CN103472481B (en) | 2016-03-23 |
Family
ID=49797398
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210184592.8A Active CN103472481B (en) | 2012-06-06 | 2012-06-06 | A kind of GPU of utilization carries out the method that angular-trace gather is extracted in reverse-time migration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103472481B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105891881A (en) * | 2014-05-15 | 2016-08-24 | 蔡晓刚 | CCP (Common Converted Point) trace extraction method of horizontal interface VTI medium |
| CN103969684B (en) * | 2014-05-27 | 2017-01-25 | 中国海洋石油总公司 | Method and device for dipole hypocenter seismic imaging |
| CN104199088B (en) * | 2014-08-15 | 2017-02-08 | 中国石油天然气集团公司 | Incident angle gather extraction method and system |
| CN108072895B (en) * | 2016-11-09 | 2020-09-15 | 中国石油化工股份有限公司 | Anisotropic prestack reverse time migration optimization method based on GPU |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102269820A (en) * | 2010-06-01 | 2011-12-07 | 潜能恒信能源技术股份有限公司 | Three-dimensional seismic pre-stack reverse-time migration imaging method based on GPU (graphics processing unit) staggered grid with small memory capacity |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100565773B1 (en) * | 2005-01-27 | 2006-03-28 | 윤광진 | Seismic imaging method with reverse-time migration using poynting vector |
| US8275550B2 (en) * | 2009-04-03 | 2012-09-25 | Chevron U.S.A. Inc. | Method for target-oriented reverse time migration for prestack depth imaging |
-
2012
- 2012-06-06 CN CN201210184592.8A patent/CN103472481B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102269820A (en) * | 2010-06-01 | 2011-12-07 | 潜能恒信能源技术股份有限公司 | Three-dimensional seismic pre-stack reverse-time migration imaging method based on GPU (graphics processing unit) staggered grid with small memory capacity |
Non-Patent Citations (3)
| Title |
|---|
| 起伏地表叠前逆时偏移理论及GPU加速技术;刘红伟 等;《地球物理学报》;20110731;第54卷(第7期);第1887页第2栏第1段，及图3 * |
| 逆时偏移成像技术研究进展;丁亮 等;《地球物理学进展》;20110630;第26卷(第3期);第1085-1095页 * |
| 通过逆时偏移计算角道集研究;YuZhang 等;《勘探前沿》;20101130;第1364页第1栏第1段倒数第1-4行，第1369页第2栏第6行-第1370第1栏第3行 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103472481A (en) | 2013-12-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Abdelkhalek et al. | Fast seismic modeling and reverse time migration on a GPU cluster | |
| CN103336959B (en) | A kind of vehicle checking method accelerated based on GPU multi-core parallel concurrent | |
| CN102565854B (en) | Mass data GPU (graphics processing unit) wave equation reverse time migration imaging method | |
| CN102628956B (en) | Equipment for correlated processing of vibroseis data in GPU/CPU coordinated mode and method thereof | |
| CN103472481B (en) | A kind of GPU of utilization carries out the method that angular-trace gather is extracted in reverse-time migration | |
| CN101853522A (en) | Multidimensional space information quality detecting and analyzing model and modeling method thereof | |
| CN105005072B (en) | The PML border dimensionally seismic wave propagating mode utilizing CUDA intends method | |
| CN102609982B (en) | Topology discovery method of space geological data based on unstructured mode | |
| CN104570081A (en) | Pre-stack reverse time migration seismic data processing method and system by integral method | |
| CN102692644A (en) | Depth domain common-image gather generation method | |
| CN103279974A (en) | High-accuracy high-resolution satellite imaging simulation engine and implementation method | |
| CN103675908A (en) | Wave-equation reverse-time migration imaging method for mass-data graphic processing unit | |
| CN109212589A (en) | It is a kind of to cooperate with parallel earthquake-capturing observation system design method based on GPU/CPU | |
| CN105403913A (en) | Pre-stack depth migration method and device | |
| Carr et al. | Scalable contour tree computation by data parallel peak pruning | |
| CN111638551A (en) | Seismic first-motion wave travel time chromatography method and device | |
| CN105894439A (en) | Ocean eddy and Argo buoy intersection data rapid extraction algorithm based on CUDA | |
| CN104570080A (en) | Multi-GPU-card cooperative and quick calculation method for pre-stack reverse time migration of mass data | |
| CN109657197A (en) | A kind of pre-stack depth migration calculation method and system | |
| CN103675895A (en) | Method for utilizing GPU (Graphic Processing Unit) to increase computing efficiency of wave field continuation | |
| CN102722653B (en) | Ray tracking and accelerating algorithm based on MapReduce | |
| CN106662665B (en) | The interpolation and convolution of rearrangement for the processing of faster staggered-mesh | |
| CN102609614B (en) | Excavation analysis method aiming at complex ore body | |
| Shih et al. | Out-of-core visualization of time-varying hybrid-grid volume data | |
| CN108072895B (en) | Anisotropic prestack reverse time migration optimization method based on GPU |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |