英文摘要

ABSTRACTS

Detectionofmicroseismiceventsbasedonconvolutionalneuralnetwork.WANG Weibo1,XU Xilong1,SHENGLi1,andGAO Ming1.OilGeophysicalProspecting,2020,55（5）:939-949.

Conventional detection methods for microseismic events are almost based on calculating the features of signals.The accuracy of events detection depends on the parameters of algorithms，so it is greatly affected by the changes in the features of signals and signal-to-noise ratio.This paper proposes a method for microseismic event detection based on a convolutional neural network（CNN）.To train and test the CNN，a sample set is constructed on the microseismic data monitored by multiple stations in an oil well that was hydraulically fractured.The data consist of effective event signals and ineffective background noises and their classifications.Then the CNN is trained and tested by the sample data set，and an optimal CNN model is obtained with best accuracy of event detection.To test the performance of the CNN model，synthesized microseismic signals with different signalto-noise ratios，and actual microseismic signals from several oil and gas wells are fed into the CNN model.The processing results demonstrate that the CNN model can automatically and effectively detect microseismic events.It has good abilities for noise suppressing and generalization.

Keywords：microseismic events，event detection，convolutional neural network，model training，realtime data processing

1.College of Control Science and Engineering，China University of Petroleum（East China），Qingdao，Shandong 266580，China

Automaticfaultrecognition withresidualnetwork andtransferlearning.ZHANG Zheng1,YAN Zhe1,and GU Hanming1.Oil Geophysical Prospecting,2020,55（5）:950-956.

The application of machine learning algorithms in the field of geophysics has been expanded and deepened.In fault recognition on seismic data，the main approach is training a shallow convolutional neural network to achieve fault recognition using actual or synthetical fault samples.Actual fault samples require manual marking，which is very time-consuming.Synthetic fault samples are easy to obtain，but the effect of the trained network model is inadequate when applied to actual seismic data.For this reason，this paper combines a deep residual network with transfer learning to fault recognition.First train synthetical fault samples by constructing a deep residual network with better performance，then use a small number of actual fault samples for transfer learning.This way the generalization ability of the network can be enhanced，and the recognition results can be optimized.After transfer learning，the network can more effectively improve the recognition accuracy of actual faults than ever before.Actual seismic data have proved the feasibility and effectiveness of the method.

Keywords：seismic data interpretation，fault recognition，deep residual network，transfer learning，network structure optimization

1.Institute of Geophysics and Geomatics，China University of Geosciences，Wuhan，Hubei 430074，China

Iterativeschemeinspirednetworkfornon-stationary randomdenoising.ZHANG Wenzheng1,TANGJie1,LIUYingchang1,MENGTao1,andCHEN Xueguo2.OilGeophysicalProspecting,2020,55（5）:957-964,972.

Conventional filtering methods often magnify the influence of noise，which in return impedes the improvement of resolution and“smooths＂discontinuous information in seismic data.We introduce a non-stationary random noise filtering method based on an iterative scheme-inspired network（IIN）which has a simple and tight structure and can be used to smooth non-stationary random noises.The L1 norm is used to optimize the objective function of the alternating directional multiplier algorithm which the IIN is derived from.A new auxiliary variable is added to transform the extreme value of the objective function into an augmented Lagrange form，and using the L-BFGS algorithm to distinguish and train all the network parameters.Finally an optimal denoising model is obtained.Applications to model and real data show that：①the trained denoising model can effectively suppress noises while maintaining the characteristics of events according to the features of useful signals；and the simple and tight iterative network can speed up the rate of convergence and rapidly finish denoising and achieve expected results using a smaller database and shorter training time；②the method proposed has a good adaptability and can suppress non-stationary random noises in conventional seismic data.

Keywords：deep learning，iterative scheme inspired network，non-stationary random noises，denoising model

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

2.Research Institute of Exploration＆Production，SINOPEC Shengli Oilfield，Dongying，Shandong 257015，China

A new deep-towed,multi-channel high-resolution seismicsystemanditspreliminaryapplicationinthe South China Sea.WEI Zhengrong1,2,3,PEI Yanliang2,3,4,andLIU Baohua4,5.OilGeophysicalProspecting,2020,55（5）:965-972.

The proprietary deep-towed，multi-channel high-resolution seismic system（Kuiyang-ST2000）consisting of a plasma electric spark source with the advantages of high sound source level，wide frequency band，and multichannel receiver array（48 channel，interval 3.125m）is capable of operating at 2000m in ocean.During operating，the source and the receiving cable are towed to the offshore floor（less than 100m）.This shortens the distance between the system and the detected target，reduces the absorption and attenuation of seawater during the propagation of seismic signals，and can overcome multiples，bubbles and ocean noises caused by towing streamers on sea surface.The first test on the ST2000 system was completed in the E sea area of the South China Sea（the regional sea floor is about 1500m deep，and the system working depth is about 1400m）.The final imaging profile after fine processing of test data has verified the superiority of the Kuiyang-ST2000 system，which overcomes the shortcomings of insufficient resolution of conventional marine multi-channel seismic systems.

Keywords：deep-towed seismic system，plasma spark source，multi-channel cable，high resolution seismic，vertical resolution，horizontal resolution

1.College of Geological Engineering and Geomatics，Chang'an University，Xi'an，Shaanxi 710046，China

2.First Institute of Oceanography，MNR，Qingdao，Shandong 266061，China

3.Key Laboratory of Marine Geology and Metallogeny，MNR，Qingdao，Shandong 266061，China

4.Laboratory for Marine Geology，Qingdao National Laboratory for Marine Science and Technology，Qingdao，Shandong 266061，China

5.National Deep See Center，Qingdao，Shandong 266273，China

Selectingspatialsamplesbasedonseismicmigration ofdipformation.CAIXiwei1,HEBaoqing1,ZHANG Yang1,WANG Yan1,andTANG Donglei1.OilGeophysicalProspecting,2020,55（5）:973-978.

The size of spatial sample is an important parameter for seismic data acquisition and processing.At present，the formula for calculating spatial sampling sizes only considers the aliasing-free samplesof reflected wave，or diffracted wave in the migration aperture，but it is not suitable for the imaging requirements of dip formations.In this paper，the time-distance equation of reflected wave is established based on the essence of seismic migration imaging and the distribution characteristics of diffracted wave field，then the spatial sampling formula for migration imaging of dip formations is derived based on the principle of seismic wave inphase superposition，and finally，the spatial sampling of seismic imaging with different dips and different frequencies is analyzed.The result provides a reliable basis of spatial sampling selection for seismic data acquisition design and migration imaging.

Keywords：spatial sampling，dip formation，diffracted wave field，time-distance equation of reflection wave，in-phase stacking

1.Acquisition Technique Center，BGP，CNPC，Zhuozhou，Hebei 072751，China

Automaticpick-upofseismicP-wavefirstarrivals viafuzzy C-means method constrained by supervirtualinterferometry.TAN Jiawei1,LIJing1,LI Feida2,and ZENG Zhaofa1.OilGeophysicalProspecting,2020,55（5）:979-990.

In seismic data processing，such as static correction and seismic tomography，accurate and quick pick-up of first arrivals is the basic premise of velocity structure imaging and comprehensive interpretation of seismic data.It is difficult to finish big data through manual picking，and manual method may cause man-made errors.For seismic data with low signal-to-noise ratio（SNR），conventional automatic picking methods，such as correlation method，energy ratio method（STA／LTA），and fractal method，should keep adjusting parameters to achieve desired accuracy，therefore resulting in poor stability.This paper proposes a method to automatically pick up first arrivals based on fuzzy Cmeans and super-virtual interferometry.Fuzzy Cmeans（FCM）clustering analysis is an unsupervised machine learning method.It only depends on data themselves for classification，so it can be more flexibly and conveniently applied for picking actual seismic first arrivals.For data with low SNR，to get accurate and stable first arrivals，first super-virtual interferometry（SVI）is used to enhance the energy of weak first arrival signals，such as far offsets，and improve the SNR of the seismic data.Tests on theoretical model data and actual land seismic data show that the method provides a stable and efficient technical means for automatically picking up first arrivals from big seismic data.

Keywords：fuzzy C-means，cluster analysis，first arrival picking，super-virtual interferometry，P-wave tomography

1.College of Geo-Exploration Science and Technology，Jilin University，Changchun，Jilin 130021，China

2.Geophysical Exploration Institute of Jilin Province，Changchun，Jilin 130062，China

Aseparation method ofup-goingand down-going wavefieldsfordual-sensorseismicdata.GAO Shaowu1,QIAN Zhongping1,SUN Pengyuan1,FANG Yunfeng1,LI Jianfeng1,and MA Guangkai1.Oil GeophysicalProspecting,2020,55（5）:991-996.

The separation of the up-going and down-going wavefields is a key step in the data processing of dual-sensor seismic data.Aiming at the disadvantage that the up-going wavefield separated by conventional methods contains the down-going wavefield，and the down-going wavefield contains the down-going wavefield，that is，the up-going and the down-going wavefields can not be separated completely，this paper introduces three energy matching factors and four wavefield separation contribution factors.Three energy matching factors are the energy matching factor of dual-sensor data，the up-going wavefield energy matching factor of the down-going wavefield，and the down-going wavefield energy matching factor of the up-going wavefield.Four wavefield separation contribution factors are the up-going wavefield contribution factor of the hydrophone data，down-going wavefield contribution factor of the hydrophone data，the upgoing wavefield contribution factor of geophone data，and down-going wavefield contribution factor of the geophone data.The up-going and down-going wavefield separation equations are established.The optimal energy matching factors of the up-going and down-going wavefields of dual-sensor seismic data are computed and determined by an iteration algorithm.Then the wave field separation contribution factors are directly calculated to realize the complete separation of the up-going from down-going wavefields of dual-sensor seismic data.The separated up-going wavefield data not only eliminates the ghost multiple interferences，but also widens the effective frequency band.The signal-tonoise ratio and resolution of the seismic data are improved too，and the high-fidelity up-going and down-going wavefields are provided for subsequent joint deconvolution and migration imaging processing.The effectiveness and practicability of this method are illustrated by synthetic and real data examples.

Keywords：dual-sensor seismic data，up-going and down-going wavefields，wavefield separation，energy matching factor，separation contribution factor，signal-to-noise ratio

1.Geophysical Research ＆ Development Center，BGP，CNPC，Zhuozhou，Hebei 072750，China

Fasttwo-stepinterpolationalgorithmforseismicdata.MA Zechuan1,LIYong1,2,CHEN Lixin1,CHEN Jie1,WANGPengfei1,andLIXuemei1.OilGeophysicalProspecting,2020,55（5）:997-1004.

In order to improve interpolation efficiency and choose an optimal interpolation scheme，based on the analysis formula of convex set projection（POCS）algorithm and iterative threshold（IST）algorithm，fast iterative shrinking threshold（FIST）algorithm and fast convex set projection（FPOCS）algorithm are developed.The basic idea is that the interpolation results from the previous step and the interpolation results from the first two steps are linearly combined with the linear operator to get the iterative contraction operator，and the interpolation algorithm is used for interpolation.Then a new quality control criterion is introduced to improve the computational efficiency and accuracy.IST，POCS，FIST and FPOCS algorithms are used to interpolate the incomplete seismic data of the four-layer seismic model and Marmousi model established by Seismic Lab，and the best threshold strategy is selected and finally verified by actual seismic data.The results show that the signal-tonoise ratio from an exponentially declining threshold is higher than those from a constant threshold，a linearly declining threshold and a data-driven threshold.Combined with termination criterion，when the maximum iterations are 35 to 50，a better interpolation effect can be obtained.

Keywords：fast iterative shrinking threshold algorithm，fast convex set projection algorithm，threshold strategy，termination criterion，seismic data interpolation

1.School of Geophysics，Chengdu University of Technology，Chengdu，Sichuan 610059，China

2.State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation（Chengdu University of Technology），Chengdu，Sichuan 610059，China

Combination ofadaptivesubtraction and Curvelet transform to suppresssurface waves.LI Jiwei1,ZANG Dianguang1,DIAO Yongbo1,ZHANG Lin1,JIANShikai2,andSHIYong3.OilGeophysicalProspecting,2020,55（5）:1005-1015.

Surface wave is one of the primary noises in raw seismic data.It is so strong that it seriously affects the quality of seismic data.In order to improve the quality of seismic data without damaging effective signals，this paper proposes a method for suppressing surface waves based on the combination of adaptive subtraction filtering and Curvelet transform.First，adaptive subtraction filtering has the fidelity and amplitude-preserving characteristics，which is used to suppress most surface waves while maximally protecting effective signals.Then Curvelet transform has the multi-direction characteristics，which is used to suppress residual surface waves.After the processing above，surface waves are effectively suppressed，the signal-to-noise ratio is significantly improved，the amplitude time slice is smoother，and the low-frequency componentis better protected in the cross-frequency scanning record.Analysis and research show that the method proposed in this paper can effectively suppress surface waves while protecting effective signals.It is a relatively fidelity and amplitude-preserving method for surface wave suppression.

Keywords：surface waves，adaptive subtraction，Curvelet transform，fidelity and amplitude-preserving，noise attenuation

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

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

3.Southwest Geophysical Prospecting Branch，BGP，CNPC，Chengdu，Sichuan 610000，China

Characteristicsofseismic wavefieldinfrequencyspacedomaininstrongattenuation media.ZHANG Yi1,WANG Yun1,CHENBenchi2,and WANG Xiangchun1.Oil Geophysical Prospecting,2020,55（5）:1016-1028,1046.

As a viscoelastic medium model，the strong attenuation model applies Biot's basic idea to modify the viscous dissipation coefficient，and describes the absorption and attenuation characteristics of the medium through the equation of motion.Compared with some conventional viscoeastic medium models and viscoelastic-porous medium models，the strong attenuation model can more easily and accurately describe the strong attenuation properties of media such as heavy oil reservoirs and loose sediments near the surface.In this paper，we implement a method of 25-point frequency-space domain finite difference to simulate the wave field characteristics of a strong attenuation model medium，and study the attenuation mechanism of seismic waves caused by three physical factors including porosity，fluid viscosity and viscoelasticity of medium.From the numerical simulation results，we can find that porosity，fluid viscosity and viscoelasticity are all important factors on inducing strong attenuation and high-velocity dispersion of seismic waves，and their influence on attenuation of S-wave is more obvious than that of P-wave.Of which，viscoelasticity is the key factor on affecting the attenuation of high-frequency component of seismic waves，and both porosity and fluid viscosity can cause the attenuation of energy in the effective frequency range of seismic waves，especially porosity has the strongest attenuation effect.We also study the influence of the shallow medium with strong attenuation on the deep wave field，and find that the strong attenuation model is more practical in describing the lowvelocity medium by comparing the wave field simulation results under the complete elasticity theory and general viscoelasticity.These results provide a reference to the study of strong attenuation media involving different attenuation mechanisms，and also lay a foundation for the establishment of the comprehensive strong attenuation compensation theory.

Keywords：strong attenuation model，frequencyspace domain，finite difference algorithm，porosity，fluid viscosity，viscoelasticity

1.School of Geophysics and Information Technology，China University of Geosciences（Beijing），Beijing 100083，China

2.Sinopec Science＆Technology Department，Beijing 100728，China

Full waveform inversion oftime-domain acoustic wavebasedonfirst-orderapproximateinstantaneous frequency.XU Yipeng1,RENZhiming1,LIZhenchun1,LIU Chang1,HE Zilin1,and CHEN Jinmao2.Oil GeophysicalProspecting,2020,55（5）:1029-1038.

A deep velocity model with high accuracy can be obtained by full waveform inversion（FWI）in theory.However，it is difficult to find underground conditions in actual exploration.This is because that，on the one hand，full waveform inversion is strongly dependent on the initial velocity model；and on the other hand，deep signals are weak，and their contribution to the objective function is small.This paper proposes time-domain full waveform inversion based on first-order approximate instantaneous frequency.First，according to the instantaneous phase formula，the objective function and the adjoint source term formula of the first-order approximate instantaneous frequency are derived；Then the initial velocity model for full waveform inversion is established by making full use of the advantage of the instantaneous frequency to highlight low frequency information and deep weak signals.Finally，a gradient attenuation factor is added to further improve the effect of deep inversion and prevent the interference between deep and shallow layers.The feasibility and validity of the method have been proved by model and real data.

Keywords：initial model，instantaneous frequency，first-order approximation，full waveform inversion（FWI）

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

2.Shengli Branch of Sinopec Geophysical Corporation，Dongying，Shandong 257000，China

Fourierfinite-differencereversetimemigrationusing GPU.DUAN Xinbiao1,2.Oil GeophysicalProspecting,2020,55（5）:1039-1046.

Reverse time migration（RTM）is a pre-stack depth migration technique commonly used for imaging complex structures.In conventional RTM，high-order finite difference and post-imaging filtering algorithms are usually used，but spatial dispersion will be caused by large computing girds and effective low-frequency information will be lost when using post-imaging filtering to remove lowfrequency noises.A GPU-based Fourier finitedifference RTM method is developed.It uses the Fourier finite-difference algorithm to calculate wavefield propagation，and explicit wavefield decomposition and cross-correlation imaging are directly performed in time-wavenumber domain during wavefield extrapolation.In addition，the GPUbased algorithm can significantly improve the calculation speed of Fourier finite difference RTM.The application in field data shows that the method can protect and utilize the high and low frequency information effectively，so the imaging resolution can be improved apparently.

Keywords：reverse time migration（RTM），Fourier finite-difference，analytical wavefield decomposition，imaging bandwidth

1.Wave Phenomena and Intelligent Inversion Imaging Group（WPI），Tongji University，Shanghai 200092，China

2.Sinopec Geophysical Research Institute，Nanjing，Jiangsu 211103，China

Anelastic-wavereverse-timemigrationmethodbased onoptimalfinite-differenceoperatorsusinganew plane wavesolution.ZOU Qiang1,2,HUANG Jianping1,YONG Peng1,and LI Zhenchun1.OilGeophysicalProspecting,2020,55（5）:1047-1059.

Elastic-wave revere time migration based on the theory of two-way wave equation is not limited by the dip angle of formation，so it has a strong adaptability to complex models.However，high computing cost and a large memory limit its application for processing big seismic data.This paper deduces an elastic-wave reverse-time migration imaging method using equivalent staggered-grid finitedifference based on time-space domain optimization.To improve imaging accuracy，we reduce the dispersion error by optimizing difference operators，which means that a low-order difference operator can achieve the imaging accuracy of a high-order difference operator，so as to reduce the computing cost.To reduce memory usage，the second-order displacement equation is adopted to avoid intermediate variables.We also adopt an efficient boundary storage strategy，which uses the wavefield value near the boundary to control the reverse-time migration wavefield storage，which effectively reduces the memory use.Model tests show that the new method with a 6th-order spatial difference has a similar imaging accuracy to the conventional staggered-grid scheme with the 10th-order spatial difference，and the new method with a 10th-order corresponds to the conventional method with a 18th-order difference.Compared with conventional elastic wavefield storage strategy，effective boundary storage strategy can save a lot of memory usage on the premise of increasing less computing cost，and this strategy is more beneficial to shorter difference operators.

Keywords：elastic-wave reverse-time migration，plane wave optimization，computational cost，storage usage，efficient boundary storage

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

2.PetroChina Tarim Oilfield Company，Korla，Xinjiang 841000，China

Exactreflectioncoefficientsoforthorhombicanisotropic media.ZHANG Xueying1,SUN Pengyuan2,MA Xuejun3,LU Jun4,and LI Mengqi4.OilGeophysicalProspecting,2020,55（5）:1060-1072.

Most of continental sedimentary formations in China are characterized by thin interbeds.Under the influence of tectonic movement，in sedimentary formations，vertical or nearly vertical high-dip fractures are developed，which result in typical orthorhombic anisotropy.This kind of orthorhombic anisotropy can be regarded as the superposition of VTI and HTI.Therefore，the study on AVO responses of orthorhombic anisotropic media is significant for fine characterization of fractured reservoirs.To the orthorhombic anisotropy induced by a set of upright fractures developed in a VTI background，Tsvankin anisotropy parameters are used to construct the stiffness coefficient matrix.Then，the method for calculating exact reflection and transmission coefficients is presented based on the Christoffel equations and the corresponding boundary conditions.Model test shows that：①if P-wave is incident from a low-impedance isotropic medium to a high-impedance orthorhombic anisotropic medium，the increase of the VTI anisotropy coefficient in the background media will lead to the increase of the PP-wave reflection coefficient and the decrease of the PS1-and PS2-wave reflection coefficients；and the increase of fracture weakness will lead to the decrease of the PP-wave reflection coefficient and the increase of the PS1-and PS2-wave reflection coefficients；②if Pwave is incident from a high-impedance isotropic medium to a low-impedance orthorhombic anisotropic medium，the reflection coefficients of PP-，PS1-and PS2-waves present the opposite characteristics；③the azimuthal anisotropy of PP-wave reflection coefficients are much smaller than those of PS1-and PS2-waves.By comparing with the results of Rüger approximate reflection coefficient formulas，the accuracy of our calculation method is verified.

Keywords：reflection coefficient，orthorhombic anisotropy，stiffness coefficient matrix，AVO response，shear wave splitting

1.School of Energy Resources，China University of Geosciences（Beijing），Beijing 100083，China

2.Geophysical Research and Development Center，BGP，CNPC，Zhuozhou，Hebei 072751，China

3.Petroleum Exploration and Production Research Institute，Sinopec Northwest Oilfield Company，Urumqi，Xinjiang 830011，China

4.School of Geophysics and Information Technology，China University of Geosciences（Beijing），Beijing 100083，China

Generalizedimpedanceblockyinversionbasedonanalyticsolutiontowaveequation.LIYuanqiang1,2,3,HUOZhizhou4,LIJingye1,2,3,CHEN Xiaohong1,2,3,ZHANGJian1,2,3,andGENG Weiheng1,2,3.OilGeophysicalProspecting,2020,55（5）:1073-1083.

Since the pre-processing of pre-stack gathers is based on the assumption of acoustic medain in many cases，the the gathers tend to be with more acoustic AVO features.In addition，density inversion is unstable.This paper proposes a generalized impedance blocky inversion based on analytic solution to acoustic wave equation.The generalized acoustic impedance is inverted by a partially stacked profile，which varies with the angle of incidence；and on this basis，more accurate velocity and stable density are extracted.For the conventional impedance inversion method，transmission loss and interlayer multiples are neglected.Based on the recursive formula of derivation，the one-dimensional acoustic wave equation is solved analytically to obtain the full-wavefield responses at different incident angles，and the Fréchet derivatives are analytically derived for gradient-descent inversion algorithm.Most of the inversion methods are based on smoothing constraints，which fundamentally lead to unfocused boundaries for inversion results.In order to improve the resolution of the inversion results，blocky constraints can be introduced based on the Bayesian inference framework to obtain stable and high resolution inversion results.According to the above theory，we first uses model data to analyze the influence of the incompleteness of the forward method on seismic responses，further verify the validity of the inversion method，and extract the accurate velocity and density.Then the ability to characterize the boundary for blocky constraintis tested by adding noises.Both model and actual data prove that the inversion results from the new method have higher resolution，the boundary is clearer，and the extracted velocity and density profiles are stable and accurate.

Keywords：analytical solution，nonlinear，generalized impedance，blocky constraint，wave equation inversion

1.College of Geophysics，China University of Petroleum（Beijing），Beijing 102249，China

2.State Key Laboratory of Petroleum Resources and Prospecting，Beijing 102249，China

3.National Engineering Laboratory for Offshore Oil Exploration，Beijing 102249，China

4.Sinopec Petroleum Exploration and Production Research Institute，Beijing 100083，China

Ajointinversionmethodusingamplitudeandvelocityanisotropy.ZHOUXiaoyue1,GANLideng1,YANG Hao1,WANG Hao2,andJIANG Xiaoyu1.OilGeophysicalProspecting,2020,55（5）:1084-1091.

The fracture prediction method based on amplitude anisotropy with high resolution is sensitive to the degree of medium anisotropy and easy to realize，but it is poor in dealing with noises and has high requirements on the quality of seismic data.For this reason，a joint inversion method using prestack amplitude and velocity anisotropy is proposed.For each time window，the anisotropic gradient obtained from velocity inversion at the bottom of the time window is taken as a constraint of amplitude inversion at the top of the time window，and then an inversion volume can be obtained by moving the time window.Applications to model and real data show that：①compared with prestack and poststack amplitude anisotropy inversion，the inversion accuracy of the proposed method is the highest；②in the GS1 well block，the result of prestack velocity anisotropy inversion coincides with the fracture curve of the well，which can be used as a constraint of amplitude inversion，but the resolution of velocity anisotropy inversion is lower than that of amplitude anisotropy inversion.The joint inversion of prestack amplitude with velocity anisotropy can improve the resolution of velocity inversion.

Keywords：anisotropy，fracture prediction，prestack amplitude，velocity，joint inversion

1.Research Institute of Petroleum Exploration＆Development，CNPC，Beijing 100083，China

2.Southern Sichuan Gas District of PetroChina Southwest Oil＆ Gasfield Company，Luzhou，Sichuan 646000，China

Geologicalmodelingofbraidedriverreservoirbased ongenesisandevolution:acasestudyonblockMin Orinoco Heavy Oil Belt, Venezuela. CHEN Shizhen1,2,LinChengyan1,2,REN Lihua1,2,ZHANG Xianguo1,2,andHUANG Wensong3.OilGeophysical Prospecting,2020,55（5）:1092-1101.

The Orinoco Heavy Oil Belt in Venezuela is the largest heavy oil accumulation area in the world.It is rich in geological reserves and has a huge exploitation potential.However，because the primary oil pay zone in the area is almost fluvial deposits of sandy braided river，it is characterized by laterally fast facies change and vertically complex superimposition of sand bodies，resulting in serious heterogeneity inside the reservoir，and restricting the following production and the implementation of the development plan.Taking the lower Oficina formation in block M as a case，according to the cores，well logging and seismic data，four main genetic units were identified and summarized including compound bars，braided channels，abandoned channels and residual floodplain mudstone.Under the guidance of a sedimentary conceptual model，the strata slicing method was used to extract and analyze the evolution process of the genetic units over time.This information extracted was transformed into a probability volume to control the geological modeling process.Taking the identified results of horizontal and vertical wells as hard data，a final geological model controlled by genesis and evolution information was established under the constraint of the probability volume.The drill-in rate by horizontal wells has been increased to more than 95%after the model was applied in field operation，indicating a good application effect.

Keywords：seismic sedimentology，Orinoco Heavy Oil Belt，braided river，geological modeling，genesis and evolutional information

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

2.Reservoir Geology Key Laboratory of Shandong Province，Qingdao，Shandong 266580，China

3.Research Institute of Petroleum Exploration Development，PetroChina，Beijing 100083，China

A methodofdecompactioncorrectionbasedonlong trendofintervaltransittime.LAIShenghua1,CAO Jianhua2,and ZHANG Cuiping3.Oil Geophysical Prospecting,2020,55（5）:1102-1109.

The sedimentary environment of the reservoir in Block F in the Yanchang gas field is a delta front.Due to the formation compaction，the reservoir has features of low porosity and low permeability.The contrast of interval transit time and density between the reservoir and the surrounding rock is small，which seriously affects the reliability and precision of seismic reservoir prediction.In order to reduce or partially eliminate the influence of compaction on interval transit time and increase the interval transit time contrast（velocity）between the reservoir and the surrounding rock，a method of decompaction correction with a long trend of interval transit time is proposed in the paper.Firstly，the interval transit time are transformed after timefrequency analysis，and divided into different components in the frequency field.Taking the low-frequency component of the standard well as criteria，low-frequency long-trend correction to the interval transit time is carried out for all wells in Block F.Then new interval transit time data are formed by integrating the original high-frequency component and the corrected low-frequency part.Geophysical responses are reanalyzed using the new interval transit time.Now the reservoir is obvious with lower interval transit time，and could be distinguished clearly from the surroundings.Finally，neural network seismic velocity inversion is carried out using the corrected new interval transit time，and a high-resolution acoustic velocity inversion profile is obtained.The reservoir thickness has been accurately predicted.

Keywords：compaction，time-frequency analysis，long trend of interval transit time，decompaction correction，reservoir prediction

1.Xi'an Shiyou University，Xi'an，Shanxi 710065，China

2.Tianjin University of Science and Technology，Tianjin 300222，China

3.No.6 Oil Production Plant of PetroChina，Changqing Oilfield Company，Dingbian，Shaanxi 718600，China

Deepenedapplicationofhorizon-flatteningtechnique inseismicinterpretation.GUO Wen1,LIU Yongtao2,3,4,ZHAOJunfeng2,3,WANG Xiuzhen4,ZHAO Hongge2,3,andDINGFufeng5.OilGeophysicalProspecting,2020,55（5）:1110-1120.

From the perspective of seismic interpretation，horizon-flattening is a paleo-surface reconstruction technique.Due to the limitation of the horizon-flatten technique，it is impossible to recover the thickness of inclined strata，the thickness of compacted strata and the horizontal expansion of the strata.According to the principle of the horizon-flattening technique，first it was compared with other recovery techniques，and then by adding geological constraints or improving the original techniques，the impact of recovering error on results was minimized to let it be better applied for seismic interpretation.The following conclusions have been drawn：①If the angle between the top and bottom boundaries of a wedge sequence is less than 15°and the thickness of the interval is less than 1000m，thin reservoirs can be approximately predicted with horizontal slices instead of stratal slices；②When the compaction constrast above and under an unconformity is relatively small，the thickness of the denuded layer can be recovered by horizon-flattening，and then restored to the paleo-geomorphology by the modified residual thickness method；③In a hydrocarbon generation area with increased temperature，no compaction correction is needed to restore the paleo-structure in the main accumulation stage，therefore，while improving the recovery accuracy，it also enhances the hydrocarbon geological significance of the paleo-structure；④In a regions with less intense tectonic deformation，the tectonic evolution can be analyzed through seismic horizonflattening instead of balanced cross-section.These conclusions can further promote the application of seismic horizon-flattening technique in oil and gas exploration.

Keywords：seismic horizon-flattening，stratal slice，horizontal slice，paleo-geomorphology，paleo-structure，tectonic evolution，balanced cross-section，Ordos Basin

1.Research Institute of Petroleum Exploration＆Development，CNPC，Beijing 100083，China

2.State Key Laboratory of Continental Dynamics，Northwest University，Xi'an，Shaanxi 710069，China

3.Department of Geology，Northwest University，Xi'an，Shaanxi 710069，China

4.Changqing Branch of Geophysical Research Institute，BGP，CNPC，Xi'an，Shaanxi 710021，China

5.Exploration Department of PetroChina Changqing Oilfield Company，Xi'an，Shaanxi 710018，China

Structurecharacteristicsandgenetic mechanism of extensionalfaultsin Chedubaaccretionary wedgeslopebasin,Myanmar.YANGSongling1andXIEJielai2.OilGeophysicalProspecting,2020,55（5）:1121-1130.

In recent years，as a kind of rare oil／gas-bearing basin，accretionary wedge-slope basins have been paid more and more attention to.At present，exploration and research on this kind of basin is relatively less，so its complex structural features，especially the genesis of the complex fault system，have been debating.There are a lot of seismic and geological data about the Cheduba accretionary wedge-slope basin，which provide a basis for studying the tectonic characteristics and fault genesis of the basin.Based on the regional geology and under the guidance of the idea of“vertical（tectonic）layers and lateral（segments）blocks＂，firstly the fault type is defined in different stages；then the genetic mechanism of the extensional faults that are nearly parallel to the orientation of the accretionary wedge structural belt is discussed.This solves the contradiction between geological understanding and stress.On this basis，a new induced extension model of normal fault development is proposed，and three group of faults with different properties are identified in the Cheduba accretionary wedge-slope basin：S-N compressional faults，NE-SW extensional-transregional faults and NW-SE extensional faults.Controlled by oblique subduction of the oceanic crust of the Indian plate to the West Myanmar microplate，the compressive stress field perpendicular to the trench formed S-N compressional faults，and resulted in large-scale thrust activities，the shear stress parallel to the trench formed regulatory NE-SW extensional or transregional faults under the regional right-lateral strike-slip effect and the extensional stress field induced by different horizontal motion components during the activities of the accretionary wedge compression thrust fault zone in the front and back edge of the slope basin resulted in NW-SE extensional faults.The findings of this study are references to other accretionary wedge-slope basins.

Keywords：accretionary wedge，slope basin，structure style，extension fault，induced extensional effect

1.Research Institute of Exploration ＆ Development Technology，CNOOC International Ltd，Beijing 100028，China

2.SPG Secretariat of CPS，BGP，CNPC，Zhuozhou，Hebei 072751，China

Time-frequency electromagnetic（TFEM） method:Data acquisition system and its application.HE Zhanxiang1,2,3,CHEN Zhongchang1,3,REN Wenjing4,PANG Hengchang4,TIAN Zhiquan4,andSHEN Yibin1.OilGeophysicalProspecting,2020,55（5）:1131-1138.

A high-power electromagnetic instrument system is the foundation of deep mineral resource exploration.It is of great significance to develop high-power and high-precision electromagnetic instruments. The time-frequency electromagnetic（TFEM）exploration instrument system based on deep oil and gas exploration demand can realize electromagnetic integration exploration in the time domain and frequency domain simultaneously.The simulation results show that the technical indexes of the system can meet the requirements of integrated exploration of time domain and frequency domain，electric method and magnetic method，and can also realize data acquisition in near and far regions.This time-frequency electromagnetic instrument system independently developed by BGP includes high-power constant-current electromagnetic transmitter and distributed／nodal multi-type electromagnetic signal receiver.The dynamic range of the receiver is greater than 120dB，and the frequency range of the received signal is 0.001-500Hz.The data acquisition experiment proves that its technical index reaches the international level of similar instruments，which can provide strong support for China＇s deep resource exploration.

Keywords：Time-frequency electromagnetic instrument，comparative test，industrialized application

1.SUSTech Academy for Advanced Interdisciplinary Studies，Shenzhen，Guangdong 518055，China

2.Southern Marine Science and Engineering Guangdong Laboratory，Shenzhen，Guangdong 511458，China

3.Shenzhen Key Laboratory of Deep Offshore Oil and Gas Exploration Technology，Department of Earth and Spaces Science，SUSTech，Shenzhen，Guangdong 518055，China

4.GME ＆ Geochemical Surveys of BGP，CNPC，Zhuozhou，Hebei 072751，China

PredictionofmagneticbodytopbasedonBPneural network.ZHAO Wenju1,LIU Yunxiang1,TAO Deqiang1,ZHAO Li1,and HU Wentao1.OilGeophysicalProspecting,2020,55（5）:1139-1148.

Prediction of magnetic body top based on magnetic anomaly is one of the major objectives of magnetic exploration.It is of great significance for the investigation of magnetic minerals，igneous rocks and magnetic basement.By extracting the magnetic anomalies from a large number of theoretical cuboid models and the maximum values of the first，second，or third derivatives in the vertical direction，three extremum ratios were obtained and a sequence dataset was formed.The constructed BP neural network was trained by the extremum ratio sequences and the depth of the model top to build training samples，and the trained BP neural network was stored for predicting the magnetic body top.Modeling results show that the prediction errors of more than 91%of the training samples are less than 10%，and the prediction of the samples not participating in the training also matches well with the results.The method has been applied for the prediction of the depth of the igneous rock in the YX area，China.The predicted depth highly agrees with the depth of the igneous rocks from drilling data，indicating the effectiveness of the method.

Keywords：extremum of magnetic anomaly，vertical derivative，extremum ratio，depth of magnetic source，BP neural network，depth of igneous rocks 1.GME ＆ Geochemical Surveys，BGP，CNPC，Zhuozhou，Hebei 072751，China

Three-dimensionalhigh-efficiencyandhigh-precision numericalsimulationofgravityandmagneticpotentialfieldsofcomplexbody.ZHOUYinming1,2,3,DAI Shikun1,2,LI Kun1,2,LING Jiaxuan1,2,HU Xiaoying3,andXIONGBin4.OilGeophysicalProspecting,2020,55（5）:1149-1159,1168.

The forward modeling of gravity and magnetic potential fields is the basis of inversion and interpretation.For applicable forward algorithms，it is difficult to consider both calculation accuracy and calculation efficiency in complex conditions.A three-dimensional numerical simulation method for gravity and magnetic potential fields is proposed.It transforms the three-dimensional integration of gravity and magnetic potential fields to one-dimensional integration with independent wave number through two-dimensional Fourier transform in the horizontal direction.The one-dimensional integral can be discretized vertically into the sum of the integrals of multiple elements，and shape function interpolation is conducted within the element.Both the calculation accuracy and efficiency are high.This method makes full use of the high accuracy of the shape function integral and the high efficiency of the Fourier transform.Finally，a prism model is designed，and the analytical solution to the model is compared with the numerical solution to the method，indicating that the theory of the method is correct and the accuracy is high.A complex model with continuous vertical variation is designed，and the accuracy of traditional prism uniform subdivision is compared with the quadratic interpolation of shape function method，proving that the method has a high applicability to the complex model.

Keywords：gravity and magnetic potential fields，three-dimensional numerical simulation，Fourier transform，shape function method

1.School of Geosciences and Info-physciences，Central South University，Changsha，Hunan 410083，China

2.Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring，Ministry of Education，Changsha，Hunan 410083，China

3.GME ＆ Geochemical Surveys of BGP，CNPC，Zhuozhou，Hebei 072751，China

4.College of Earth Sciences，Guilin University of Technology，Guilin，Guangxi 541006，China

Physicalpropertiesandmodelingofdensity,susceptibilityandresistivityofPre-CambriansysteminSichuanBasin.XIANGKui1,YANLiangjun1,HUHua1,YANG Hui2,and WEN Baihong2.Oil Geophysical Prospecting,2020,55（5）:1160-1168.

The discovery of the large Anyue gas field in the Sichuan Basin has proved that oil and gas resources in the Pre-Cambrian strata are rich and promising in China.Then the Proterozoic system in the Sichuan Basin is becoming a key field for deep oil and gas exploration.Non-seismic exploration is an important means for investigating ultra-deep oil and gas reservoirs.However，gravity，magnetic and electromagnetic data of the Pre-Cambrian in the Sichuan Basin are insufficient，so it is difficult to model the basin upon the gravity，magnetic and geoelectrical structures，limiting the application of gravity，magnetic and electromagnetic exploration.Based on the Pre-Cambrian outcrops in the Sichuan basin，the density，susceptibility，complex resistivity and polarization were analyzed on the samples of the sedimentary rocks，metamorphic rocks and igneous rock，models of gravity，magnetic，and geoelectrical properties of the ancient Pre-Cambrian strata were built.The important parameters of the Pre-Cambrian hydrocarbon source rock based on gravity，magnetic and electric data provide the physical basis for the processing and interpretation of non-seismic data for oil and gas exploration in Sichuan Basin.

Keywords：Sichuan basin，Pre-Cambrian，density，susceptibility，complex resistivity，physical property model building

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

2.PetroChina Research Institute of Petroleum Exploration＆Development，Beijing 100083，China