Amethodfordeterminingtheaspectratioofpatch：reanalysisof85%rule.ZHANGHua1,WANGMeisheng1,LIKanghu1,NINGHongxiao1,HANZhi-xiong1,andGUYong1.OilGeophysicalProspecting，2019,54(5):947-953.

On the premise of satisfying the needs of geological tasks，the paper proposes a method to determine the size and aspect ratio of patches for a more efficient and convenient 3D geometry design.With the center of the patch as the origin，and the spatial coordinates (x，y) as the independent variables，a rectangular coordinate system is constructed.With the center of the patch to the designed maximum offset as the radius，a circle is drawn.Receivers in the circle are considered effective，and those outside the circle are considered invalid.Taking the effective information and invalid information and their inter-relation information as dependent variables，the functional relations of the effective information and invalid information with the spatial coordinate position are calculated.The spacial coordinates determine the length and width of a 3D patch，and their proportion (the aspect ratio).At the same time，a template and a table for the horizontal and vertical parameter selection of the patch are generated.This method is convenient and flexible，and can be used for a reference in the 3D survey design.

Keywords：aspect ratio，85% rule，mathematical model，mathematical analysis，area integral，directional derivative，broadband，wide-azimuth，high-density (BWH)

1.Acquisition Technology Center，BGP Inc.，CNPC，Zhuozhou，Hebei 072751，China

Arandom-noisesuppressionapproachwithself-adaptivedata-driventightframeformicroseismicdata.TANGJie1,ZHANGWenzheng1,LIANGYuwei1,andGUYutian2.OilGeophysicalProspecting，2019,54(5):954-961.

The signal-to-noise ratio (SNR) of ground microseismic data is very low，which seriously affects the accuracy of first-break picking and the reliability of inversion results.This paper uses a noise estimation method based on a weak texture block to obtain a noise variance from noisy microseismic data，and then uses the data driven tight frame method to effectively suppress random noise and improve the SNR of the data.Model and real data tests show that the proposed approach can remove background patches caused by conventional me-thods，and the SNR after denoising is greatly improved.Therefore，compared with conventional methods，the proposed approach has obvious advantages.

Keywords：noise estimation，data-driven tight frame construction，seismic noise suppression，weak texture block，micorseismic

1.School of Geosciences，China University of Petroleum (East China)，Qingdao，Shandong 266580，China

2.Center for Oil & Gas Exploration Management，Shengli Oilfield Branch Co.，SINOPEC，Dongying，Shandong 257000，China

Matchingcorrelation-functionvelocityanalysisbasedonADCIG.LIJiang1,LIQingchun2，andTANGWen3.OilGeophysicalProspecting，2019,54(5):962-969.

Avoiding the multi-solution problem of wavefield propagation and being sensitive to errors of migration velocity，angle-domain common image gathers (ADCIG) are very suitable for migration velocity analysis.Residual depth equations suitable for inclined strata are derived based on kinematics properties of ADCIGs to satisfy the velocity analysis of complex structures.A matching correlation-function residual curvature spectrum algorithm is developed to obtain accurate residual curvature va-lues.Compared with stack and correlation methods，the proposed algorithm has better noise-resistance and a higher resolution，achieves more accurate velocity analysis.Model and real data tests prove the validity of the proposed algorithm.

Keywords：angle domain common image gather (ADCIG)，velocity analysis，residual curvature，correlation function

1.Xi’an Research Institute，China Coal Technical & Engineering Group Corp，Xi’an，Shaanxi 710077，China

2.School of Geological Engineering and Geoma-tics，Chang’an University，Xi’an，Shaanxi 710154，China

3.School of Earth Sciences and Engineering，Xi’an Shiyou University，Xi’an，Shaanxi 710065，China

Seismicrandomnoisesuppressionbasedonscale-adaptive3D-Shearlettransform.CHENGHao1,WANGDeli2，WANGEnde1,FUJianfei1,andHOUZhenlong1.OilGeophysicalProspecting，2019,54(5):970-978.

We use a scale-adaptive 3D-Shearlet transform to suppress random noise of multi-source seismic data.First multi-source seismic data is transformed into the 3D-Shearlet domain.In this domain，seismic data can be represented more sparsely.Since seismic signals distribute in a low scale and random noise distributes in the all scales，scale-adaptive factors suppress random noise based on hard thresholds.With the inverse 3D-Shearlet transform，the de-noising seismic data can be obtained.According to numerical and seismic data tests，the proposed approach can better suppress random noise than the 2D-Shearelet transform and the 3D-Shearlet transform；however it needs a larger computer memory，and may damage weak signals.

Keywords：3D-Shearlet transform，sparsity，random noise，scale-adaptive factor，signal-to-noise ratio (SNR)

1.Key Laboratory of Safe Mining of Deep Metal Mines，Ministry of Education，Northeastern University，Shenyang，Liaoning 110819，China

2.College of Geo-Exploration Sciences and Technology，Jilin University，Changchun，Jilin 130026，China

3Dseismicdatareconstructionbasedonnon-convexLpnorm.LIUQun1,FULihua1,andZHANGWanjuan1.OilGeophysicalProspecting，2019,54(5):979-987,996.

The multichannel singular spectrum analysis (MSSA) is a classical method to deal with 3D seismic data reconstruction.With the random singular value decomposition，the rank of block Hankel matrix constructed by frequency slices of seismic data is reduced by MSSA to reconstruct seismic data，but the solutions are not always optimal ones.Lpnorm is a non-convex function between the L0norm and the L1norm，and it is closer to the rank function than the convex nuclear norm.So a non-convex Lpnorm Hankel reconstruction method is proposed for 3D seismic data rank-reduction reconstruction.Since this is a non-convex optimization problem，singular values are constrained by setting weights for an iterative process，which ensures a low-rank of reconstructed data.Numerical experiment results demonstrate that the proposed me-thod obtains the higher signal-to-noise ratio reconstructed data compared with MSSA and the orthogonal matrix pursuit Hankel reconstruction.

Keywords：3D seismic data，Lpnorm，low rank approximation，iterative weighting，Hankel matrix

1.School of Mathematics and Physics，China University of Geosciences(Wuhan)，Wuhan，Hubei 430074，China

Near-surfaceQvalueestimationandquantitativeamplitudecompensation.SUQin1,ZENGHuahui1,TIANYancan1,XUXingrong1,XIAOMingtu1,andWUJie1.OilGeophysicalProspecting，2019,54(5):988-996.

Most strata compensation methods only consider the middle and the deep absorption rather than the strong lateral velocity and thickness variations at the near-surface，so the results are not satisfactory.To solve the problem，we propose a deterministic quantitative compensation method based on near-surface model to realize quantitativeQprediction and space-time variation compensation.Based on the principle of surface consistency，the conception of surface relative attenuation coefficient is introduced.A relative attenuation coefficient is calculated with statistics iterations in the common offset domain，and a near-surface relativeQis obtained.Then the near-surface relativeQis calibrated with the measured absoluteQand a near-surfaceQmodel is built.Finally，a stable near-surfaceQspace-time variation compensation is realized.Model tests verify the high accuracy of the proposed method，and 3D seismic data applications in a few oil fields prove the validity of the proposed method.

Keywords：relative attenuation coefficient，iteration，near-surface relativeQvalue，near-surfaceQcalibration，space-time variation compensation

1.Northwest Branch，Research Institute of Petroleum Exploration & Development，PetroChina，Lanzhou，Gansu 730020，China

ACurveletthresholditerationmethodbasedonenergyratioforsurface-wavesuppression.LIJiwei1,LIUXiaobing2，ZHOUJunhua1,JIANShikai3，andZENGZhen1.OilGeophysicalProspecting，2019,54(5):997-1004.

The surface-wave suppression with the Curvelet transform is affected by the surface-wave and effective wave overlapping in the Curvelet domain.In the practice，the Curvelet transform cannot completely separate surface waves from effective waves.Therefore this paper proposes a Curvelet threshold iteration method based on energy ratio for surface-wave suppression.Using their characteristic differences in the frequency，velocity，and direction，effective waves and surface waves in different bands are decomposed in the Curvelet domain for many times.In different scales and angles，multiple iterations with different thresholds are carried out based on the distribution characteristics of surface waves and effective waves.According to our applications，the proposed method achieves better surface-wave separation and suppression than conventional methods.

Keywords：Curvelet transform，surface wave，noise attenuation，multi-scale，multi-direction

1.Southwest Branch，GRI，BGP Inc.，CNPC，Chengdu，Sichuan 610000，China

2.Exploration and Development Research Institute，Southwest Oil & Gas Field Company，PetroChina，Chengdu，Sichuan 610000，China

3.School of Geosciences，China University of Petroleum (East China)，Qingdao，Shandong 266580，China

SeismicdatareconstructionwiththediscretecosinetransformandShearletdictionariesunderthemorphologicalcomponentanalysisframework.ZHANGKai1,2,ZHANGYikui1,LIZhenchun1,2,TIANXin1,OUYANGYi1,andCHENJunyi1.OilGeophysicalProspecting，2019,54(5):1005-1013，1056.

Compared to a single transformation，the morphological component analysis (MCA) is a more efficient method of sparse representation of signals.In this paper，we propose a seismic data reconstruction method with the discrete cosine transform (DCT) and Shearlet dictionaries under the MCA framework.Firstly，we select the DCT dictionary and the Shearlet dictionary to represent local singular components and smooth linear components of seismic data respectively.Then we reconstruct all the components by the block coordinate relaxation (BCR) algorithm with an exponential threshold model and exponential threshold function.Finally，we merge the components to get a reconstructed result.Experiments on synthetic and real data show that the proposed method can effectively reconstruct missing seismic data and the accuracy of reconstruction is higher than the single Shearlet dictionary，the Curvelet + the DCT dictionaries，the Shearlet + the Curvelet dictionaries.

Keywords：morphological component analysis (MCA)，Shearlet transform，discrete cosine transform (DCT)，seismic data reconstruction，compressed sensing

1.School of Geosciences，China University of Petroleum (East China)，Qingdao，Shandong 266580，China

2.Laboratory for Marine Mineral Resources，Qingdao National Laboratory for Marine Science and Technology，Qingdao，Shandong 266071，China

Efficientalgorithmsformodedecouplingofelasticwavefieldintransverselyisotropicmediawithaverticalsymmetryaxis.RUANLun1，2andCHENGJiu-bing1，2.OilGeophysicalProspecting，2019,54(5):1014-1023.

For the elastic wave mode decoupling in anisotropic media，conventional approaches with divergence and curl operations lead significant errors.The accurate mode decomposition algorithm，however，is prohibitively expensive because the polarization projection operators are model- and direction-dependent.Based on previous work，we focus on developing efficient algorithms for mode decoupling in heterogeneous medium with vertically transverse isotropy (VTI).First，for the pseudo-divergence or pseudo-curl operators constructed by the deviation of propagation and polarization direction of the dominant energy，we improve the accuracy of qP/qS separation by taking into account the variations of the deviation angle with the propagation direction.This improvement is very important when different wave modes overlap in the same spatial zone at the same time.Then，for the accurate but prohibitively expensive algorithm of mode decoupling through polarization projection，based on the existing low-rank approximation of the projection operator，we achieve a balance between the accuracy and computational cost using the strategies of model partitioning and graph processing unit (GPU) acceleration.The example on the SEG Hess VTI model demonstrates the features and validities of the improved pseudo-divergence/pseudo-rotation algorithm and the GPU-based low-rank approximate polarization projection algorithm.

Keywords：anisotropic medium，elastic wave，mode decoupling，polarization projection，pseudo-divergence/pseudo-curl，low-rank decomposition

1.State Key Laboratory of Marine Geology，Tongji University，Shanghai 200092，China

2.School of Ocean and Earth Sciences，Tongji University，Shanghai 200092，China

Multi-axialcomplex-frequencyshiftingnearlyperfectlymatchedlayerforseismicforwardmodelinginelasticmedia.LUOYuqin1andLIUCai1.OilGeophysicalProspecting，2019,54(5):1024-1033.

The perfectly matched layer (PML) is a very efficient method and has become widely used in the seismic forward modeling.The development from split methods to various non-split methods greatly improves the computational efficiency of PML and reduces the required memory.The nearly perfectly matched layer (NPML) belongs to non-split PML，and it transforms directly the wavefield.Therefore，it does not change the form of the governing equations and is easy to be implemented.However，the NPML suffers from large spurious reflections at grazing incidence and evanescent wave.So，the convolution perfect matching layer (CPML) is still usually used.We implement the complex frequency shifted-NPML (CFS-NPML) to improve the capacity of absorbing grazing-incidence wave.We introduce different damping profiles that are proportional to each other in the orthogonal direction to make CFS-NPML stable called multi-axial CFS-NPML (MCFS-NPML).Meanwhile we design a new attenuation function which can improve the NPML absorption further.Eventually，the MCFS-NPML is able to absorb any incident wave more stably and efficiently.

Keywords：perfectly matched layer (PML)，seismic forward modeling，complex frequency shifting (CFS)，multi-axes，absorption boundary condition

1.College of Geo-exploration Sciences and Technology，Jilin University，Changchun，Jilin 130026，China

Optimizeddifferencecoefficientofstaggeredcompactfinitedifferenceschemeand2Dacousticwaveequationnumericalsimulation.WANGYong1,WANGPeng1,CAIWenjie1,andGUIZhixian1.OilGeophysicalProspecting，2019,54(5):1034-1045.

In this paper，we use the least square and Lagrange multiplier to perform difference coefficient optimization for the staggered compact finite difference scheme of first-order derivatives based on the idea of dispersion relationship preservation.Then，we analyze the optimized-scheme simulation accuracy，dispersion relation，and acoustic wave equation stability.The following understandings are obtained：A.For the same difference accuracy，the optimized staggered compact difference (OSCD) scheme uses two more nodes to calculate the first-order derivative than staggered compact difference (SCD) schemes；B.OSCD has a smaller truncation error and lower simulation dispersion than staggered difference (SD) schemes and SCD，so OSCD has higher calculation accuracy，and it is more suitable for coarse grid computing；C.For the same difference accuracy，the stability condition of OSCD for acoustic wave equation is slightly stricter than SD and SCD，and the applicable time grid size is slightly smaller.Wavefield numerical simulations of acoustic wave equation on uniform，horizontal-layered，and Marmousi models demonstrate that the proposed method is suitable for complex media and it has higher accuracy and efficiency.

Keywords：staggered compact finite difference，dispersion relation preservation，numerical simulation，numerical dispersion，stability condition

1.Key Laboratory of Exploration Technologies for Oil and Gas Resources (Yangtze University)，Ministry of Education，Wuhan，Hubei 430100，China

AjointvelocitymodelbuildingmethodwithGaussianbeamtomographyandfullwaveforminversionforlandseismicdata.LIUDingjin1,HUGuanghui1,CAIJiexiong1,NIYao1,andHEBinghong1.OilGeophy-sicalProspecting，2019,54(5):1046-1056.

The full waveform inversion (FWI) based on prestack seismic wavefield fitting is facing a few challenges in the application，for instances，low signal-to-noise ratio (SNR) land seismic data，missing low frequency information，complicated near-surface，and huge memory needed for massive 3D seismic data processing.In terms of low-precision land seismic migrations，we propose in this paper a velocity model building method by Gaussian Beam tomography，which not only makes up medium wave-number components unachievable in conventional methods based on ray tracing tomography but also provides FWI with a high precise macroscopic velocity model that helps to avoid effectively cycle skips.Besides，in order to reduce the dependency of inversion on low frequencies，we replace the L2 norm with a cross-correlation function based on phase matching，which figures out validly the problem of intensive computing of FWI.Furthermore，with a self-adjoint forward operator from pseudo-conservative wave equations，the gradient computing capacity based on an adjoint-state method is enhanced dramatically，which improves the inversion adaptation and realizes a high-resolution velocity model building for land seismic data.

Keywords：full waveform inversion (FWI)，Gaussian-beam tomography，velocity model building，land seismic data

1.Geophysical Research Institute，SINOPEC，Nanjing，Jiangsu 211103，China

VTImediumanisotropicparametertomographyinversionbasedonangledomaincommonimaginggathers.WANGFeiyi1,ZHANGKai1,LIZhenchun1,ZHAOShuo1,LIDaoshan2，andNINGBin2.OilGeophysicalProspecting，2019,54(5):1057-1066.

Current anisotropic tomography methods mainly use the information of different offset ranges or information of angle domain common imaging gathers (ADCIGs) to perform the anisotropic parameter inversion，but lack the definition of partition in ADCIGs.Firstly，according to the travel time first-order partial derivative to Thomsen parameter and the travel time second-order partial derivative to Thomsen parameter and phase angle，the range of small，medium and large ADCIGs is clearly defined.Then the basic velocity field is established by an isotropic one-way wave migration and velocity tomography.The ADCIGs of VTI media inverse-time migration is extracted by Poynting vector method.Finally，we can develop a method to invert theVP0with the small angle range gathers information，δwith the middle range，andεwith the large range.Trial results of a simple concave model illustrate the correctness of the proposed method，and trial results of a complex structural model verify the applicability of the proposed method.

Keywords：anisotropic，reverse time migration (RTM)，angle domain common imaging gather (ADCIG)，residual curvature，tomography

1.School of Geosciences，China University of Petroleum (East China)，Qingdao，Shandong 266580，China

2.BGP Inc.，CNPC，Zhuozhou，Hebei 072750，China

Angle-domainreversetimemigrationwithGaussianbeamsforTImedia.XIAOJian’en1,LIZhenchun1,ZHANGKai1,andLIUQiang1.OilGeophysicalProspecting，2019,54(5):1067-1074.

The reverse time migration with Gaussian beams combines the high efficiency and flexibility of Gaussian beam migration and the high precision of reverse time migration，which can be used for target-oriented imaging.In this paper，an anisotropic ray tracing algorithm based on phase velocity is introduced into the reverse time migration with Gaussian beams，and combined with the propagation angle information of Gaussian beam calculation，a more efficient angle-domain Gaussian beams reverse time migration method for TI media is realized.According to our model trial，compared with the conventional anisotropic algorithm based on elastic parameters，the proposed method has higher computational efficiency and extracted angle-domain common imaging gathers (ADCIGs) can not only provide support for subsequent migration velocity analysis，but also be used for stack imaging to suppress imaging noise and improve image quality.

Keywords：TI media，reverse time migration with Gaussian beams，Green function，angle-domain common imaging gather (ADCIG)

1.School of Geosciences，China University of Petroleum (East China)，Qingdao，Shandong 266580，China

Least-squaresreversetimemigrationonundulatedsurfacebasedonT-distribution.LIQingyang1.OilGeophysicalProspecting，2019,54(5):1075-1083.

The hydrocarbon seismic exploration in areas with complex structure，complex lithology，and complex surface conditions has many difficulties，among which the severely undulating surface is one of the bottlenecks.The least-squares reverse time migration (LSRTM) has higher imaging resolution，amplitude fidelity，and balance than conventional migration methods.However，the LSRTM based on rectangular grid cannot deal with complex surface and overcome severe undulating terrains such as piedmont zones.In addition，seismic data in these areas often contain a lot of noise.Compared with Huber norm and mixed norm，the T distribution has a better performance in incomplete data，and it has no redundant parameters.So it is simple and practical.The results of Huber norm and mixed norm methods depend heavily on the parameter selection，which requires a lot of attempts.Therefore，a fully-staggered grid is introduced into the body-fitted grid，and the T distribution is extended to LSRTM with undulating surfaces.The linear Born forward equation of body-fitted is further deduced.On this basis，the LSRTM algorithm for undulating surface based on body-fitted fully staggered grid is proposed，which can better solve influences of undulating surfaces.Model tests verify the effectiveness and complex-model suitability of the proposed algorithm.

Keywords：irregular topography，fully-staggered grid，least-squares reverse time migration (LSRTM)，T distribution，object function

1.Geophysical Exploration Research Institute，Zhongyuan Oilfield Branch Co.，SINOPEC，Puyang，Henan 457001，China

Multi-sourceelasticfullwaveforminversionbasedonP-waveandS-waveseparation.HUANGShaohua1,RENZhiming1,LIZhenchun1,GUBingluo1,andLIHongmei2.OilGeophysicalProspecting，2019,54(5):1084-1093，1105.

The elastic wave full waveform inversion (EFWI) can obtain high-accuracy P-wave and S-wave velocity，and density，but the computational cost is rather high.The P-wave and S-wave coupling causes crosstalk noise and decrease inversion accuracy.Multi-source coding can greatly improve EFWI calculation efficiency，but increasing the wavefield complexity，leading to more serious nonlinear problems.In this paper，a multi-source elastic wave full waveform inversion method based on P-wave and S-wave separation is proposed.The method uses dynamic random coding to suppress multi-source crosstalk noise，and uses wavefield separation to mitigate crosstalk effects caused by the P-wave and S-wave coupling.The proposed method is tested on the Marmousi model with consistent P-wave and S-wave velocity structures and on the Marmousi-II model with uncorrelated P-wave and S-wave velocity structures.The results show that the inversion method can effectively suppress crosstalk noise and improve computational efficiency.

Keywords：elastic wave full waveform inversion (EFWI)，P-wave and S-wave separation，multi-source，converted wave

1.School of Geosciences，China University of Petroleum(East China)，Qingdao，Shandong 266580，China

2.Geophysical Research Institute，Shengli Oilfield Branch Co.，SINOPEC，Dongying，Shandong 257022，China

Time-frequencycharacteristicsanalysisoforderlythickness-gradedthininter-bedsbasedonaffinesmoothedpseudo-Wignerdistribution.LIXueying1,2,TIANYajun1，3，CHENGYun1,2,andNIEWeidong1，2.OilGeophysicalProspecting，2019,54(5):1094-1105.

The orderly thickness-graded thin inter-bed is typical one in many thin inter-bed types.Thus an accurate description of its time-frequency characteristics is of great significance for its identification.We first establish a geological model of orderly thickness-graded thin inter-beds with different total thicknesses，different numbers of inter-beds，and different thickness variation.Then we obtain synthetic data with the wave equation forward modeling，and the time-frequency spectrum with the affine smoothed pseudo-Wigner distribution.On this basis，we analyze characteristics and variation regulations of time-domain waveform，time-frequency spectrum，and instantaneous frequency spectrum of different thin inter-beds.According to our research，for typical orderly thickness-graded thin inter-beds，the frequency of thin beds increases while the spectrum of thin beds is wide；and the frequency of thick beds decreases while the spectrum of thick beds is narrow.And the time frequency variation regulation of thin bed is its main controlling mechanism.Besides，time-frequency characteristics are different with thickness changes：when a single-bed thickness is less than 1/8 wavelength，the peak frequency decreases and the spectrum becomes narrower with its thickness decrease；when a single-bed thickness is greater than 1/2 wavelength，the first band is compressed and the bandwidth is broadened with its thickness increase.In addition，for a single-bed thickness between the above two，it has similar spectral characteristics as the single thin bed.Thus,single-bed thickness interval,numbers of inter-beds and thickness variation are their main controlling factors.In this paper，we propose also the concept of transitional-graded thin inter-beds and equivalent thin beds，so we can analyze thin beds，equivalent thin inter-beds，and orderly thickness-graded thin inter-beds under a unified characteristic frame.

Keywords：orderly thickness-graded thin inter-beds，affine smoothed pseudo-Wigner distribution，transitional graded thin inter-beds，equivalent thin bed，wave equation forward modeling

1.College of Earth Science，Northeast Petroleum University，Daqing，Heilongjiang 163318，China

2.Heilongjiang Key Laboratory of Hydrocarbon Reservoir Forming Mechanism and Resource Evaluation，Daqing，Heilongjiang 163318，China

3.School of Electronic and Information Engineering，Xi’an Jiaotong University，Xi’an，Shaanxi 710049，China

IdentificationofOrdoviciansubtlecavereservoirsinTaheonseismicdata.LIUKunyan1andXUJie1.OilGeophysicalProspecting，2019,54(5):1106-1114.

Drilling results shows that Tahe Ordovician formation developed two types of concealed cave reservoirs：one is shielded by T74strong reflections，the other is shielded by strong bead-like reflections.Some reservoirs show great potentials with high productiion.Because of their buried depth (more than 5300m)，the seismic dominant frequency is rather low (about 28Hz)，the seismic resolution is about 30m (1/4wavelength).The size of some caves is small，seismic responses are quite weak，and seismic characteristics of these caves are blurry.Conventional methods are not suitable to identify this kind of caves.To sovle the problem，we develop a new approach with the help of drilling and logging data.For caves shielded by T74reflections，the principal component analysis (PCA) is used to remove adjacent strong noise and highlight reflections of caves.For caves shielded by bead-like reflections，the geo-statistics inversion based on low frequency model optimization is use to remove bead-like reflections secondary lobes.The proposed approach is succsessfullly applied to identify Ordovician subtle cave reservoirs in Tahe，8m-thick caves are clearly described.The coincidencerate is up to 83% between inversion results and drilling and logging data，which proves high reliability of the proposed approach.

Keywords：cave，strong T74reflection shielding，strong bead-like reflection shielding，principal component analysis (PCA)，geo-statistics inversion，Tahe

1.Petroleum Exploration and Production Research Institute，SINOPEC，Bejing 100083，China

Coalstructurequantitativepredictionwithsensitive-attributeandparameter-inversionfusion.FENGXiaoying1,YANGYanhui1,2,ZUOYinqing1,2,DINGRuixia1,HANSheng1,andQINChen3.OilGeophysicalProspecting，2019,54(5):1115-1122.

Coal structures are closely related to the gas production of coal reservoirs，so the coal structure quantitative prediction is essential.We analyze seismic sensitive attributes and inversion sensitive parameters on a 3D seismic dataset in Coal 3#，Mabi，Qinshui Basin.It is found that the most sensitive attribute is the texture，and the most sensitive is resistivity of logging data.The texture attribute change is relatively smooth in the horizontal direction，and makes faults clear，but its vertical resolution is low.On the other hand，the vertical resolution for inversed resistivity is high，and the inversed data in well positions is consistent with logging curves，but its horizontal change is not smooth，and fault identification is impossible.With the texture-attribute and inversed-resistivity fusion，the coal structure quantitative prediction is achieved，which is proved by the late well drilling.Our practice and experience demonstrate the validity of the proposed approach and may provide a useful reference for similar conditions.

Keywords：coal structure，texture attribute，reservoir parameter inversion，validity，fusion

1.Exploration and Development Research Institute，Huabei Oilfield Company，PetroChina，Renqiu，Hebei 062552，China

2.Exploitation Pilot Base for Coalbed Methane，CNPC，Renqiu，Hebei 062552，China

3.No.2 Oil Production，Daqing Oilfield Company，PetroChina，Daqing，Heilongjiang 163414，China

PrestackAVOcharacteristicsandidentificationofflatspotinYinggehaiBasin.DENGYong1,PANGuangchao1,LIMing1,ZHANGMingwei1,andLIUAiqun1.OilGeophysicalProspecting，2019,54(5):1123-1130.

Combined with drilling data of high-temperature and overpressure in Yinggehai Basin，reflection characteristics of flat spots on poststack seismic data are analyzed systematically in order to identify hydrocarbon reservoirs under complex conditions.On this basis，prestack AVO features of flat spots are also analyzed.It is found that flat points have positive reflection coefficients at the normal incidence，and the reflection coefficient increases gradually with the increase of incident angle.AVO features of flat points are different from conventional four classes of Castagna AVO scheme.These phenomena are confirmed by theoretical deduction and seismic forward modeling.A new strategy of flat spot identification is proposed.For poststack data，the flat point identification is based flat point shapes and reflection energy；for prestack data，the flat point analysis should focus on AVO characteristics and AVO types of sand top & bottom.The proposed strategy is used in the study area and the flat spot identification is greatly improved.

Keywords：Yinggehai Basin，high temperature and overpressure，AVO characteristics，gas-water interface identification

1.Zhanjiang Branch，CNOOC，Zhanjiang，Guangdong 524057，China

Low-gradefaultidentificationincomplexfault-blockzonesbasedonwellandseismicdata.BAIQinglin1,YANGShaochun1,LUZhiyong2,ZHANGYanzeng3，CHEHongwei3，andFENGJianwei1.OilGeophysicalProspecting，2019,54(5):1131-1140.

It is very difficult to identify low-grade faults due to small fault displacements and short extended length，which affects the hydrocarbon exploration and production in complex fault-block zones for the late stage of oilfield development.We propose an approach to identify low-grade faults with above 3m fault displacement in the fault block Yong 3，Dongying Sag with the joint use of logging and seismic data.This is a high-density well pattern zone.First，the similarity of same-microfacies logging curves between neighboring wells is used to obtain the development index of low-grade faults，and low-grade fault breakpoints encountered by drilling are quantitatively identified.Then drilling information is used to guide the seismic data interpretation after an accurate well-seismic calibration.Plane distribution characteristics of low-grade faults are depicted based on ant traces under the guide of these fault breakpoints.The approach takes advantages of high vertical resolution in logging data and seismic lateral information.The low-grade fault interpretation is realized from a point-line-surface (well-seismic line-horizon attributes) aspect.The low-grade fault identification in the study area with the proposed approach is verified by the dynamic production of the oilfield，which provides a very helpful reference for the oilfield future development.

Keywords：low-grade fault，joint use of well and seismic data，fault block Yong 3，development index of low-grade fault，wavelet reconstruction，quantitative identification，ant traces

1.School of Geosciences，China University of Petroleum (East China)，Qingdao，Shandong 266580，China

2.Xianhe Oil Plant，Shengli Oilfield Branch Co.，SINOPEC，Dongying，Shandong 257068，China

3.Dongxin Oil Plant，Shengli Oilfield Branch Co.，SINOPEC，Dongying，Shandong 257001，China

Astudyofkeyfactorsofhydrocarbonaccumulationinafaultedbasinbasedonpetroleum-accumulationseismology.ZHOUJiaxiong1,HUGaowei1,DENGYong1,LIHui1,XUTao1,andZHOUGang1.OilGeophysicalProspecting，2019，54(5)：1141-1150.

In the exploration of faulted basins lacking wells，source rocks and their hydrocarbon generation potential should be determined first.However，due to the lack of drilling data in the Sag Wenchang B，conventional geochemical reservoir formation evaluation methods such as organic geochemistry，organic petrology，and reservoir geochemistry do not work.Therefore，we propose a petroleum accumulation seismic analysis method for the western part of the Pearl River Mouth Basin，which is hydrocarbon source-rock identification for under-explorated areas with very little drilling data.Combining theoretical research and exploration practice，this paper summarizes a new idea of petroleum accumulation seismology.Based on modern hydrocarbon reservoir formation theory，key factors of hydrocarbon accumulation in the study area are analyzed on seismic data.The process and modes of hydrocarbon reservoir formation are revealed，and the distribution of hydrocarbon reservoirs is accurately predicted.According to our research results，high-quality source rocks of middle-deep lacustrines in the Formation Wenchang have high abundance of organic matter (type Ⅰ to Ⅱ1).This great potential zone covers 110 km2with NE-trending long strips parallel to the South Pearl River Fault，which becomes a new exploration zone in the Pearl River Basin.

Keywords：petroleum-accumulation seismology，key factors of hydrocarbon accumulation，Pearl River Mouth Basin，middle and deep layers，source-rock identification

1.Zhanjiang Branch，CNOOC，Zhanjiang，Guangdong 524057，China

Newunderstandingsofnear-shoresubaqueousfansedimentarystylesanditsseismicresponses.LEILei1,HANHongwei1,andYUJingqiang1.OilGeophysicalProspecting，2019，54(5)：1151-1158.

The trap identification in Dongying Depression is rather difficult because of complex seismic response characteristics of near-shore subaqueous fan.To solve this problem，we analyze sedimentary styles of glutenite in near-shore subaqueous fan based on a flume simulation experiment，and define seismic response characteristics with 3D forward modeling.It is found that multi-source near-shore subaqueous fans can be divided into four sub-facies such as root-fan，middle-fan，edge-fan，and inter-fan.Microfacies changes of the four sub-facies result in different types of structures and lithologic traps.Different traps correspond to different seismic response characteristics.Therefore，a seismic response-characteristics plate based on facies changes is established for the identification of different traps.In addition，it is found that the consequent deposition along a bedrock surface in the time-domain seismic data is caused by pitfall of high-velo-city of glutenite.The true occurrence of fan bodies is structurally inclined.This viewpoint can reasonably explain the oil-water relationship of reservoirs and provide a reference for the exploration of similar near-shore subaqueous fan reservoirs.

Keywords：near-shore subaqueous fan，sedimentary styles，3D forward modeling，seismic response cha-racteristics，high velocity pitfall

1.Geophysical Research Institute，Shengli Oilfield Branch Co.，SINOPEC，Dongying，Shandong 257022，China

Logfaciesrecognitionbasedonconvolutionalneuralnetwork.HEXu1,LIZhongwei1,LIUXin1,andZHANGTao2.OilGeophysicalProspecting，2019,54(5):1159-1165.

Natural gamma data of sandy braided river delta sedimentary environments in the Gasfield F，the East China Sea is selected as the training data to construct a deep convolutional neural network，which is used for log facies identification for the first time.Four kinds of gamma ray (GR) curve shapes are selected as characteristics，and their values are converted into the image form.Several processing steps are carried out for these images such as normalization，noise addition，rotation，and grayscale turning，and then this image data set is enhanced and expanded.In this way，training and test data sets are established.After that，a convolutional neural network is trained and used to establish the log facies identification model.During the training process，dropout，local response normalization，and L2 regularization are added to limit the complexity of the model and improve the generalization ability of the model.To automatically identify superposed deposition units of different grades in logging information，different scales of wavelet basis function and extreme value segmentation are used to classify different-scale deposition units of logging data.The comparison with other algorithms demonstrates that the proposed method achieves better log facies identification.

Keywords：convolutional neural network，log facies identification，multi-scale，wavelet base

1.School of Computer and Communication Engineering，China University of Petroleum (East China)，Qingdao，Shandong 266580，China

2.Geological College，Shandong University of Science and Technology，Qingdao，Shandong 266590，China

Anintegrationofinterpolation，edgepadding，anddownwardcontinuationforgravitydatabasedontheprojectionontoconvexsets.ZENGXiaoniu1,LIXihai1,LIUJihao1,andHOUWeijun1.OilGeophysicalProspecting，2019,54(5):1166-1173.

Gravity exploration areas are usually irregular and there is some data vacancy.Missing data must be interpolated before the downward continuation in the wave-number domain.Meanwhile，in order to improve the processing precision，the data needs to be edge-padded to the length satisfied with the fast Fourier transform algorithm.For conventional downward continuations of gravity data，only the downward continuation is considered，or the three steps of interpolation，expansion and downward continuation are executed independently.In this paper，the three ill-posed problems are considered uniformly，and an integration method based on the projection onto convex sets theory is proposed.The integration method uses Gerchberg-Saxton iteration method to interpolate and expand data until a predetermined iteration number is reached，and then the interpolated and expanded results are downward continued.A comparison of the proposed method with some classical interpolation，border padding，and downward continuation methods is carried out on model and real gravity data.The results show that the proposed method is simple，easy to be accomplished，and has higher interpolation，edge-padding and downward continuation accuracy.

Keywords：gravity，projection onto convex sets，interpolation，edge expansion，downward continuation

1.Rocket Force University of Engineering，Xi’an，Shaanxi 710025，China

3DDCresistivityinversionbasedoninversion-orientedadaptivemeshes.ZHOUYong1,ZHANGZhi-yong1,ZHANGDazhou2，XIEShangping1,MAXin1,andYUPengzhou1.OilGeophysicalProspecting，2019,54(5):1174-1180.

We use the unstructured adaptive grid algorithm based on model sensitivity information to generate high-quality inversion-oriented meshes，and adopt them to 3D DC resistivity inversion.The model sensitivity is a measure of the overall response of underground model change to real datasets.So，the optimized mesh generated based on the model sensitivity information has higher quality than the ordinary mesh.This new mesh can reduce the dependence of inversion on model regularization constraints and improve the quality of inversion results.We use the minimum-structure inversion object function based on a tetrahedral element mesh.The Gauss-Newton method optimizes the inversion object function.The Gauss-Newton equation is solved by the stable double conjugate gradient method in order to increase the stability of 3D DC resistivity inversion.The calculating results of both synthetic and field data prove the validity of the proposed approach.

Keywords：direct current(DC) resistivity method，unstructured mesh,model sensitivity，Gauss-Newton method，regularization inversion

1.School of Geophysics and Measurement-control Technology，East China University of Technology，Nanchang，Jiangxi 330013，China

2.School of Geosciences and Info-Physics，Central South University，Changsha，Hunan 410083，China