CN106250101A - Migration before stack method for parallel processing based on MapReduce and device - Google Patents

Abstract

Providing a kind of migration before stack method for parallel processing based on MapReduce and device, the method includes: seismic channel set data sorted according to offset distance size, generates common offset road collection data；According to the offset distance packet parameters pre-set, the collection packet of common offset road is become offset distance group；Perform to map Map operation for each offset distance group, obtain imaging space；And merge each imaging space and obtain imaging results data.The method and device, by utilizing MapReduce parallel model, improve migration before stack Parallel Implementation mode, improve computational efficiency and the large-scale data adaptability of migration before stack, provide the seismic imaging data of efficiently and accurately for seismic prospecting.

Description

Migration before stack method for parallel processing based on MapReduce and device

Technical field

The invention belongs to Seismic Data Processing Technique field in field of seismic exploration, be specifically related to a kind of based on The migration before stack method for parallel processing of MapReduce and device.

Background technology

Migration before stack processes and has high requirements Large-scale parallel computing all the time, along with the expansion of exploration scale Greatly, in the face of the huge size data volume in thousands of square kilometres of exploratory areas, existing integration method pre-stack time migration face Face computational efficiency to reduce, parallel processing small scale, the problem high to the requirement of hardware.When how to improve prestack Between skew scale adaptability, improve extensive data process time computational efficiency and system suitability, improve The economic benefit that pre-stack time migration processes, is that pre-stack time migration process suffers from a problem that.

Summary of the invention

The purpose of the disclosure is to solve a difficult problem present in above-mentioned prior art, it is provided that a kind of based on The migration before stack method for parallel processing of MapReduce and device, improve the computational efficiency that migration before stack processes, It is allowed to be adapted to large-scale calculations.

On the one hand providing a kind of migration before stack method for parallel processing based on MapReduce, the method includes: Seismic channel set data are sorted according to offset distance size, generates common offset road collection data；According to pre-setting Offset distance packet parameters, the collection packet of common offset road is become offset distance group；For each offset distance group Perform to map Map operation, obtain imaging space；Merge each imaging space and obtain imaging results data.

On the other hand a kind of migration before stack parallel processing apparatus based on MapReduce, this device bag are provided Include: for seismic channel set data being sorted according to offset distance size, generate the parts of common offset road collection data； For according to the offset distance packet parameters pre-set, the collection packet of common offset road being become offset distance group Parts；For performing to map Map operation for each offset distance group, obtain the parts of imaging space；For Merge each imaging space and obtain the parts of imaging results data.

Each aspect of the present invention, by utilizing MapReduce parallel model, improves migration before stack Parallel Implementation Mode, improves computational efficiency and the large-scale data adaptability of migration before stack, provides efficiently for seismic prospecting Seismic imaging data accurately.

Accompanying drawing explanation

By combining accompanying drawing, disclosure illustrative embodiments is described in more detail, the disclosure above-mentioned And other purpose, feature and advantage will be apparent from, wherein, in disclosure illustrative embodiments In, identical reference number typically represents same parts.

Fig. 1 shows the schematic diagram of handling process based on MapReduce.

Fig. 2 shows that migration before stack based on MapReduce according to an embodiment of the invention is located parallel The flow chart of reason method

Fig. 3 shows the schematic diagram of an example of offset distance component prescription case.

Fig. 4 shows the processing procedure schematic diagram of a concrete example of the present invention.

Fig. 5 shows the treatment effect schematic diagram of the embodiment of the present invention.

Detailed description of the invention

It is more fully described the preferred implementation of the disclosure below with reference to accompanying drawings.Although accompanying drawing shows The preferred implementation of the disclosure, however, it is to be appreciated that may be realized in various forms the disclosure and should be by Embodiments set forth herein is limited.On the contrary, it is provided that these embodiments are to make the disclosure more saturating Thorough and complete, and the scope of the present disclosure intactly can be conveyed to those skilled in the art.

The present invention utilizes MapReduce parallel computational model to improve the Parallel Implementation side that existing migration before stack processes Formula, utilizes efficient packet scheme to realize Large-scale parallel computing.By the transformation of parallel mode, carry The computational efficiency that high migration before stack processes.

MapReduce is a kind of parallel processing computation model of Google company exploitation, due at magnanimity number According to process in there is good autgmentability and good fault-tolerance, be widely used in web page index, data Excavate, science is simulated and in cloud computing platform.

MapReduce provides one, on large-scale computer cluster, mass data is carried out distributed treatment Method, it, by two functions that have been abstracted into of complicated parallel computation process height, maps Map and reduction Reduce.Process by mass data collection is divided into small data set give different computers, thus Realize parallelization.For data, MapReduce is regarded as a series of ＜ key, and value ＞ is to (key Be worth to), data handling procedure is then simplified to Map map and two stages of Reduce reduction.? In MapReduce platform, often the input data set of an operation is divided into several independent data bases, By Map parallel processing, by one group of ＜ key, value ＞ to being mapped to one group of new ＜ key, value ＞ couple, As the input of Reduce, perform associative operation.Handling process schematic diagram such as Fig. 1 institute of MapReduce Show, be made up of input (piecemeal), Map task, Reduce task, output (result) four part.

MapReduce is the technology of the big Data processing being commonly used to the Internet, but inventor's understanding Arrive, the mass data processing advantage of MapReduce parallel model, it is possible to realize on a large scale for migration before stack process, Efficiently calculate and a kind of effective solution is provided.

Embodiment 1

Fig. 2 shows that migration before stack based on MapReduce according to an embodiment of the invention is located parallel The flow chart of reason method, the method includes:

Seismic channel set data are sorted by step 201 according to offset distance size, generate common offset road collection data；

Step 202, according to the offset distance packet parameters pre-set, becomes the collection packet of common offset road partially Move away from group；

Step 203, performs to map Map operation for each offset distance group, obtains imaging space；And

Step 204, merges each imaging space and obtains imaging results data.

The present embodiment, by utilizing MapReduce parallel model to improve migration before stack Parallel Implementation mode, carries The computational efficiency of high migration before stack and large-scale data adaptability, provide the ground of efficiently and accurately for seismic prospecting Shake imaging data.

Generate common offset road collection data

In one example, first to seismic channel set data (such as CMP seismic channel set data) according to Offset distance size is ranked up, such as, can by seismic channel set data according to offset distance from small to large (or from Big to little) order arrangement, after obtained arrangement, data are referred to herein as common offset road collection data.

It is grouped into offset distance group

In one example, the collection packet of common offset road become offset distance group comprise the steps that according to setting in advance Common offset road collection data are once grouped and obtain primary offset distance group by the offset distance packet parameters put；With And in the case of the number of channels of primary offset distance group is less than threshold value, using primary offset distance group as described skew Away from group, in the case of the number of channels of primary offset distance group exceedes threshold value, carry out for described primary offset distance group Secondary is grouped, and obtains secondary offset distance group, as described offset distance group.

1, once it is grouped

In one example, can be according to the offset distance packet parameters pre-set, by common offset road collection data Once being grouped and obtain primary offset distance group, the most each primary offset distance group is once grouped corresponding to one piece After common offset road collection data (also referred to as data block).

In one example, offset distance packet parameters can be pre-set as required by user.Such as, use Family can carry out examination process by different packet parameters, the packet ginseng that Selection effect is good from examination result Number is as offset distance packet parameters.

The example of a kind of packet mode given below, wherein, with maximum offset value F_max, smallest offset is away from value F_min, offset distance interval D_fAs offset distance packet parameters.It will be understood by those skilled in the art that offset distance is grouped Parameter is not limited in this example listed these, as long as it can as the foundation of offset distance component prescription case i.e. Can.

In this example, the once packet scheme of common offset road collection data can be expressed by following formula:

$F_{N\max }=[\frac{|F_{\max }-F_{\min }|}{D_f}+1]---\left(1\right)$

$F_N=[\frac{|F_i-F_{\min }|}{D_f}+1.5]---\left(2\right)$

Formula (1) represents total packet number, F_NmaxRepresent packet count, F_maxRepresent maximum offset value, F_minRepresent Smallest offset is away from value, D_fRepresent the offset distance interval between primary offset distance group.

Formula (2) represents the group number of each primary offset distance group.F_iFor forming the institute of a primary offset distance group There are the offset distance of seismic channel set data, F_NGroup number for primary offset distance group.It is to say, offset distance is F_i All seismic channel set data, composition group number is F_NPrimary offset distance group.The offset distance component group of this example Scheme can keep the basic structure that original place is managed to greatest extent, it is ensured that calculates the consistent of effect before and after packet Property.

It will be understood by those skilled in the art that packet scheme is not limited to above example, those skilled in the art can To select other packet schemes as required, as long as be capable of common offset road collection data are grouped Purpose.

Fig. 3 shows a concrete example of the offset distance packet scheme according to the present embodiment.Art technology Personnel should be understood that this example is only for the purposes of understanding, including concrete numerical value in interior any details the most not It is intended to limit the invention.

Fig. 3 gives with smallest offset away from value 100, maximum offset value 900, offset distance class interval 200 As a example by grouping process, in Fig. 3, chequered with black and white block represents the earthquake number after arranging from big to small by offset distance According to, the data after CIG represent the offset distance value in the primary offset distance group correspondence imaging space after packet, at this In example, take the offset distance intermediate value of primary offset distance group as the offset distance after this primary offset distance composition picture Value.

In this example, offset distance divides positive and negative, and the division of offset bank uses absolute value to process.

In the case of the number of channels of primary offset distance group is less than threshold value, can be using primary offset distance group as The offset distance group obtained eventually.

2, secondary packet

In another example, if the number of channels of the primary offset distance group after being once grouped exceedes threshold value (such as 5000 Other threshold values that road or other users are arranged according to actual needs), primary offset distance group can be carried out secondary and divide Group processes, say, that can be using the primary offset distance group after once packet as packet object, to each Primary offset distance group is grouped again according to number of channels, obtains secondary offset distance group (the most sub-offset distance group) As the offset distance group finally given.The block number of each primary offset distance group secondary packet can depend on that this is first Total number of channels that level offset distance group comprises.It is as a example by 5000 roads by threshold value, when the primary total number of channels of offset distance group is less than During 5000 road, do not carry out secondary packet；When the primary total number of channels of offset distance group more than 5000 roads less than 10000 roads Time, each primary offset distance group can be grouped into two pieces, such as one piece 5000 road, another part is remaining part Number of channels；When the primary total number of channels of offset distance group is big more than 10000, during less than 15000 road, three pieces can be grouped into, Two pieces is 5000 roads, and the 3rd piece is remaining number of channels, the like, inclined to each primary more than 5000 Shifting group carries out secondary packet.Secondary offset distance group in same primary offset distance group has identical offset distance group Number.

Perform Map operation

After completing the packet transaction of above-mentioned offset distance group, can perform to map Map operation for each offset distance group, Obtain imaging space.A Map is i.e. utilized to process, an offset distance group be once grouped In the case of, this offset distance group is primary offset distance group, in the case of carrying out secondary packet, this offset distance group It it is secondary offset distance group.Each imaging space corresponds to a primary offset distance group, say, that at the beginning of one No matter being grouped into how many secondary offset distance group in level offset distance group again, they both correspond to an identical imaging Space.The size of the imaging space so obtained depends on the scale of the imaging data of a primary offset distance group, Compared with this processing mode imaging space whole with overall calculation, memory usage reduces.

Map calculates three processes that are generally divided into: data are read in, skew kernel function calculates, result of calculation output. Data are read in various ways, can once read in a track data, read in next track data after having calculated again； Can also once read in many track datas.In order to improve the computational efficiency of kernel function, data read in mode can be adopted Take the processing mode once reading in a full block of data.

Reduction operation

In one example, in the case of carrying out secondary packet, the result that can operate each Map is carried out Reduction Reduce operates, so that the result belonging to same primary offset distance group is carried out reduction process.Reduction principle For: the result of the Map operation for same primary offset distance group is carried out reduction process, same to generate Imaging space.Complete the imaging data file number after reduction process consistent with the packet count of primary offset distance group. Afterwards, can merge each Reduce operation result as final process result, i.e. obtain complete imaging road Collection data.

Imaging space piecemeal processes

In one example, it is also possible to carry out piecemeal process according to the hardware resource imaging space to obtaining.

As it was noted above, a corresponding imaging space of primary offset distance group, the process of this embodiment obtains Complete imaging results data be the summation after imaging space corresponding to all primary offset distance groups merges.So And, in the case of each Map only calculates the imaging space of a primary offset distance group, memory usage is still The space being improved, in the face of big work area, when mass data calculates, internal memory yet suffers from the situation of deficiency.

In order to solve mass data processing problem, imaging space piecemeal can be taked to process.This piecemeal processes can To be the internal processes for single map, namely map is calculating a corresponding primary offset distance During the imaging space organized, if processing inadequate resource, then this imaging space can be carried out piecemeal process, Through repeatedly calculating, complete the process of an imaging space.

The piecemeal of imaging space refers to the process resource status according to the equipment of calculating, to each primary offset distance group Corresponding imaging space carries out piecemeal, to process the imaging space after piecemeal respectively.The size of imaging space is According to processing what requirement determined, the parameter set determine, unrelated with the size of offset distance group, with offset distance Group packet count is the most unrelated.In one example, for imaging space too big in the case of, calculate the place of equipment Reason resource (the such as storage such as internal memory resource) state may not meet the requirement simultaneously processing (such as depositing), Therefore can carry out imaging space piecemeal process, the size of piecemeal stores resource with internal memory etc. or other process money Be correlated with in source, unrelated with the size of offset distance group.

Below for ease of understanding, provide the concrete example that an imaging space piecemeal processes.In this example, After offset distance group packet transaction, when processing large-scale data, first obtain in calculating equipment available Process resource information, such as amount of ram, contrast imaging space and memory size, within imaging space exceedes During storage, imaging space is carried out piecemeal process, process a part of imaging space every time.Can be according to internal memory Size and the size of imaging space carry out piecemeal, such as in save as 1GB, imaging space is 4GB, internal memory one The secondary imaging space that can only process four point, therefore imaging space is divided into four pieces, processes one piece every time, passes through The process that four times have processed an imaging space, after having processed, result exports hard disk, internal memory Vacate, process for next block imaging space, through recycled for multiple times internal memory, it is achieved mass data is once Property disposal ability.

For ease of understanding, Fig. 4 gives the secondary packet of the offset distance group comprised in above-mentioned example, imaging sky Between the processing procedure schematic diagram of one concrete example of the present invention of piecemeal and reduction operation.Those skilled in the art Should be understood that the purpose of this schematic diagram is only that the scheme helping to understand above-mentioned example, wherein, offset distance group Secondary packet, imaging space packet and reduction operation are not to realize the step that the purpose of the disclosure is had to carry out Suddenly.

Embodiment 2

According to another embodiment of the present invention, it is provided that a kind of migration before stack based on MapReduce is located parallel Reason device, this device includes: for seismic channel set data being sorted according to offset distance size, generates skew altogether Parts away from road collection data；For according to the offset distance packet parameters pre-set, by common offset road collection number According to the parts being grouped into offset distance group；For performing to map Map operation for each offset distance group, become The parts of image space；And obtain the parts of imaging results data for merging each imaging space.

In one example, the collection packet of common offset road become offset distance group may include that according in advance Common offset road collection data are once grouped and obtain primary offset distance group by the offset distance packet parameters arranged； And in the case of the number of channels of primary offset distance group is less than threshold value, using primary offset distance group as described partially Move away from group, in the case of the number of channels of primary offset distance group exceedes threshold value, enter for described primary offset distance group Row secondary is grouped, and obtains secondary offset distance group, as described offset distance group.

In one example, this device may also include that in the case of carrying out secondary packet, to each The result of Map operation carries out the operation of reduction Reduce, to carry out the result belonging to same primary offset distance group The parts of reduction process.

In one example, according to following formula, common offset road collection data once can be grouped:

$F_{N\max }=[\frac{|F_{\max }-F_{\min }|}{D_f}+1]$

$F_N=[\frac{|F_i-F_{\min }|}{D_f}+1.5]$

Wherein, F_NmaxRepresent packet count, F_maxRepresent maximum offset value, F_minRepresent that smallest offset is away from value, D_f Represent the offset distance interval between primary offset distance group, F_iFor forming all seismic channels of a primary offset distance group The offset distance of collection data, F_NGroup number for primary offset distance group.

In one example, this device may also include that for the process resource status according to the equipment of calculating, right Imaging space carries out piecemeal, to process the parts of the imaging space after piecemeal respectively.

Application example

For ease of understanding scheme and the effect thereof of the embodiment of the present invention, a concrete application example given below. It will be understood by those skilled in the art that this example is merely illustrative, its any detail is not intended to limit The present invention.

In this application example, utilize seismic exploration data that the scheme of the embodiment of the present invention is tested.Survey Examination data volume 2.1TB, test node 67,20 physical computing cores of each node.Fig. 5 is prestack time Migration result profile, which illustrates the treatment effect of the embodiment of the present invention.Table 1 is at different parallel schema Reason time and memory usage compare.Test result shows, the embodiment of the present invention significantly reduces the interior of program Deposit utilization rate, improve computational efficiency, it is achieved that the massive processing power of mass data, improve earthquake The economic benefit that survey data processes.

Tupe	The process time	Single node memory requirements
			MPI parallel processing	92 hours	48.5GB
MapReduce parallel processing	86 hours	8.4GB

Table 1

The disclosure refer to " imaging road collection ", " imaging space ", " imaging data " three concepts.Such as nothing Being particularly limited to, the calculated imaging data of each map is commonly referred to as imaging space, is made up of imaging space Result data be commonly referred to as imaging road collection (such as by after imaging space is combined according to imaging road collection Put in order and after carrying out data rearrangement, obtain imaging road collection), no matter imaging road collection or imaging space or The intermediate object program (result after such as stipulations) obtained in processing procedure can be referred to as imaging data.

The disclosure can be system, method and/or computer program.Computer program can include Computer-readable recording medium, containing for making processor realize the computer of various aspects of the disclosure Readable program instructions.

Computer-readable recording medium can be to keep and to store the instruction used by instruction execution equipment Tangible device.Computer-readable recording medium such as may be-but not limited to-storage device electric, Magnetic storage apparatus, light storage device, electromagnetism storage device, semiconductor memory apparatus or above-mentioned any conjunction Suitable combination.The more specifically example (non exhaustive list) of computer-readable recording medium includes: portable Formula computer disks, hard disk, random access memory (RAM), read only memory (ROM), erasable type can be compiled Journey read only memory (EPROM or flash memory), static RAM (SRAM), Portable compressed dish are only Read memorizer (CD-ROM), digital versatile disc (DVD), memory stick, floppy disk, mechanical coding equipment, example As stored the punch card or groove internal projection structure and the combination of above-mentioned any appropriate having instruction on it. Computer-readable recording medium used herein above is not construed as instantaneous signal itself, such as radio wave or The electromagnetic wave of other Free propagations of person, the electromagnetic wave propagated by waveguide or other transmission mediums are (such as, logical Cross the light pulse of fiber optic cables) or by the signal of telecommunication of wire transfer.

Computer-readable program instructions as described herein can download to each from computer-readable recording medium Calculating/processing equipment, or by under network, such as the Internet, LAN, wide area network and/or wireless network It is downloaded to outer computer or External memory equipment.Network can include copper transmission cable, fiber-optic transfer, wireless Transmission, router, fire wall, switch, gateway computer and/or Edge Server.Each calculating/place Adapter or network interface in reason equipment receive computer-readable program instructions from network, and forward This computer-readable program instructions, for the computer-readable storage medium being stored in each calculating/processing equipment In matter.

Can be assembly instruction, instruction set architecture for performing the computer program instructions of disclosure operation (ISA) instruction, machine instruction, machine-dependent instructions, microcode, firmware instructions, condition setup data, Or the source code write with the combination in any of one or more programming languages or object code, described programming language Speech includes OO programming language such as Smalltalk, C++ etc., and the process type programming of routine Language such as " C " language or similar programming language.Computer-readable program instructions can fully with Perform on the computer of family, perform the most on the user computer, perform as independent software kit, Part part on the user computer performs or on the remote computer completely in remote computer or service Perform on device.In the situation relating to remote computer, remote computer can be by the network of any kind It is connected to subscriber computer including LAN (LAN) or wide area network (WAN), or, it may be connected to outer Portion's computer (such as utilizes ISP to pass through Internet connection).In certain embodiments, Personalized customization electronic circuit is carried out by the status information utilizing computer-readable program instructions, the most able to programme Logic circuit, field programmable gate array (FPGA) or programmable logic array (PLA), this electronic circuit Computer-readable program instructions can be performed, thus realize various aspects of the disclosure.

Referring herein to the method according to disclosure embodiment, device (system) and the stream of computer program Journey figure and/or block diagram describe various aspects of the disclosure.Should be appreciated that each of flow chart and/or block diagram The combination of each square frame in square frame and flow chart and/or block diagram, can be real by computer-readable program instructions Existing.

These computer-readable program instructions can be supplied to general purpose computer, special-purpose computer or other can compile The processor of journey data processing equipment, thus produce a kind of machine so that computer is being passed through in these instructions Or other programmable data processing means processor perform time, create in flowchart and/or block diagram The device of the function/action of regulation in one or more square frames.Can also be these computer-readable program instructions Storage in a computer-readable storage medium, these instruction make computer, programmable data processing means and/ Or other equipment work in a specific way, thus, storage has the computer-readable medium of instruction then to include one Manufacture, it includes the function/action of regulation in the one or more square frames in flowchart and/or block diagram The instruction of various aspects.

Can also computer-readable program instructions be loaded into computer, other programmable data processing means, Or on miscellaneous equipment so that on computer, other programmable data processing means or miscellaneous equipment, perform one Series of operative steps, to produce computer implemented process, so that at computer, other number able to programme According to the one or more sides in the instruction flowchart performed in processing means or miscellaneous equipment and/or block diagram Function/the action of regulation in frame.

Flow chart and block diagram in accompanying drawing show the system of multiple embodiments, method and meter according to the disclosure Architectural framework in the cards, function and the operation of calculation machine program product.In this, flow chart or block diagram In each square frame can represent a module, program segment or a part for instruction, described module, program segment Or a part for instruction comprises the executable instruction of one or more logic function for realizing regulation.Having In a little realizations as replacement, the function marked in square frame can also be to be different from marked in accompanying drawing suitable Sequence occurs.Such as, two continuous print square frames can essentially perform substantially in parallel, and they sometimes can also Performing in the opposite order, this is depending on involved function.It is also noted that block diagram and/or flow chart In each square frame and the combination of square frame in block diagram and/or flow chart, can be by the function performing regulation Or the special hardware based system of action realizes, or can be with specialized hardware and computer instruction Combination realizes.

Being described above the presently disclosed embodiments, described above is exemplary, and non-exclusive, And it is also not necessarily limited to disclosed each embodiment.In the scope and spirit without departing from illustrated each embodiment In the case of, many modifications and changes will be apparent from for those skilled in the art. The selection of term used herein, it is intended to explain that the principle of each embodiment, reality are applied or to market best In the technological improvement of technology, or make other those of ordinary skill of the art be understood that to disclose herein Each embodiment.

Claims (10)

1. a migration before stack method for parallel processing based on MapReduce, the method includes:

Seismic channel set data are sorted according to offset distance size, generates common offset road collection data；

According to the offset distance packet parameters pre-set, the collection packet of common offset road is become offset distance group；

Perform to map Map operation for each offset distance group, obtain imaging space；And

Merge each imaging space and obtain imaging results data.

Migration before stack method for parallel processing based on MapReduce the most according to claim 1, wherein, Offset distance group is become to include the collection packet of common offset road:

According to the offset distance packet parameters pre-set common offset road collection data are once grouped at the beginning of obtaining Level offset distance group；And

In the case of the number of channels of primary offset distance group is less than threshold value, using primary offset distance group as described partially Move away from group, in the case of the number of channels of primary offset distance group exceedes threshold value, enter for described primary offset distance group Row secondary is grouped, and obtains secondary offset distance group, as described offset distance group.

Migration before stack method for parallel processing based on MapReduce the most according to claim 2, also wraps Include:

In the case of carrying out secondary packet, the result operating each Map carries out the operation of reduction Reduce, So that the result belonging to same primary offset distance group is carried out reduction process.

Migration before stack method for parallel processing based on MapReduce the most according to claim 2, wherein, According to following formula, common offset road collection data are once grouped:

$F_{N\max }=[\frac{|F_{\max }-F_{\min }|}{D_f}+1]$

$F_N=[\frac{|F_i-F_{\min }|}{D_f}+1.5]$

Wherein, F_NmaxRepresent packet count, F_maxRepresent maximum offset value, F_minRepresent that smallest offset is away from value, D_f Represent the offset distance interval between primary offset distance group, F_iFor forming all seismic channels of a primary offset distance group The offset distance of collection data, F_NGroup number for primary offset distance group.

5. locate parallel according to the migration before stack based on MapReduce described in any one in claim 1-4 Reason method, also includes:

According to the process resource status of the equipment of calculating, imaging space is carried out piecemeal, and after processing piecemeal respectively Imaging space.

6. a migration before stack parallel processing apparatus based on MapReduce, this device includes:

For seismic channel set data being sorted according to offset distance size, generate the parts of common offset road collection data；

For according to the offset distance packet parameters pre-set, the collection packet of common offset road being become offset distance The parts of group；

For performing to map Map operation for each offset distance group, obtain the parts of imaging space；And

The parts of imaging results data are obtained for merging each imaging space.

Migration before stack parallel processing apparatus based on MapReduce the most according to claim 6, wherein, Offset distance group is become to include the collection packet of common offset road:

According to the offset distance packet parameters pre-set common offset road collection data are once grouped at the beginning of obtaining Level offset distance group；And

In the case of the number of channels of primary offset distance group is less than threshold value, using primary offset distance group as described partially Move away from group, in the case of the number of channels of primary offset distance group exceedes threshold value, enter for described primary offset distance group Row secondary is grouped, and obtains secondary offset distance group, as described offset distance group.

Migration before stack parallel processing apparatus based on MapReduce the most according to claim 7, also wraps Include:

For in the case of carrying out secondary packet, the result operating each Map carries out reduction Reduce behaviour Make, the result belonging to same primary offset distance group to be carried out the parts of reduction process.

Migration before stack parallel processing apparatus based on MapReduce the most according to claim 7, wherein, According to following formula, common offset road collection data are once grouped:

$F_{N\max }=[\frac{|F_{\max }-F_{\min }|}{D_f}+1]$

$F_N=[\frac{|F_i-F_{\min }|}{D_f}+1.5]$

Wherein, F_NmaxRepresent packet count, F_maxRepresent maximum offset value, F_minRepresent that smallest offset is away from value, D_f Represent the offset distance interval between primary offset distance group, F_iFor forming all seismic channels of a primary offset distance group The offset distance of collection data, F_NGroup number for primary offset distance group.

10. parallel according to the migration before stack based on MapReduce described in any one in claim 5-9 Processing means, also includes:

For the process resource status according to the equipment of calculating, imaging space is carried out piecemeal, to process respectively point The parts of the imaging space after block.