Igneous rocks have shielding and absorbing effects on seismic waves，which results in poor seismic data quality of the underlying strata of igneous rocks. Igneous rocks with different lithologies have obvious differences in density and magnetic susceptibility. Thus，the lithology of igneous rocks can be identified by gravity and magnetic data. Combining gravity and magnetic data，this study conducts research on the method of gravity and magnetic data fusion based on a logistic function and applies it to the Qinggelidi Block to improve the reliability of igneous rock identification. Firstly，based on seismic constraints，3D inversion of gravity and magnetic data is performed to obtain the density and magnetic susceptibility distribution of igneous rocks in the study area. Then，an improved logistic function is used to achieve the fusion of gravity and magnetic inversion data，and the lithology of igneous rocks is imaged. Four groups of igneous rock units are identified，and their spatial distribution characteristics are interpreted. The interpretation results are consistent with the drilling information. The multi-parameter information of igneous rock gravity and magnetism is effectively fused by the logistic function，and the lithology identification results of igneous rocks are more comprehensive and accurate，which shows the effectiveness and reliability of the proposed method.

Identifying and analyzing the fault structure in seismic data can reveal the changes of underground structure and rock strata and provide an important basis for resource exploration and geological disaster prevention. However，the collected 3D seismic data contains a great deal of noise，and the proportion of fault bodies is very small. Thus，the results obtained by ant body identification methods have large errors and lack continuity and accuracy. Therefore，this paper proposes a fault identification method based on 3D residual attention network RAtte-UNet for deep learning. The method integrates the residual skip connection and attention mechanism and conducts model training. In the training process ，the mixed loss function is used to reduce the influence of the extreme imbalance between faults and non-faults on network training，so that the network has a good identification ability for small faults. Through fault identification of simulated 3D seismic data and real 3D seismic data，it is found that the evaluation indexes such as accuracy，recall，and precision have been improved. Compared with the ant body identification method and other fault identification methods，this method achieves better fault continuity in identification results. It can identify small faults and has a strong model generalization ability，which can be applied to actual seismic data.

To address the worldwide seismic exploration challenge of achieving clear and accurate structural imaging in dual-complexity areas with a low signal-to-noise ratio（SNR）in piedmont zones，this paper proposes a solution using the beam seismic technique. Through beam seismic acquisition based on“belt-patch dense point array，uniform and sufficient sampling，and vertical rolling”，high-precision uniform and sufficient sam-pling of effective and interference waves is performed in both vertical and horizontal directions. This overcomes the serious deficiency of lateral sampling in conventional 3D seismic surveys and achieves a transition from outdoor combined noise suppression to outdoor high-precision noise sampling（without noise suppression）and further indoor high-precision denoising. Fully exploiting the advantages of beam seismic acquisition data，this study forms a set of characteristic processing techniques represented by beam seismic shot-domain volume denoising，high-precision velocity modeling，and migration imaging. Multiple application examples in the Tarim Basin demonstrate that the beam seismic technique has achieved a qualitative leap in seismic data quality in dualcomplexity areas with a low SNR，significantly improving the ability to identify and interpret complex geological targets. It has become a replicable and referential seismic acquisition and processing technique，providing a new and effective technical means and empirical reference for solving the seismic imaging challenges in oil and gas exploration and development in dual-complexity areas in piedmont zones both in China and other countries.

To address the shortcomings of traditional lithology identification methods in missing logging curves handling，accuracy，and model interpretability，this paper proposes an interpretable lithology identification method based on a MSCNN-GRU neural network to complement logging curves，along with Optuna hyper parameter optimization for the XGBoost model. Firstly，to tackle the issue of lost and distorted logging curves in specific layer segments，the paper introduces a curve reconstruction method based on the combination of a multiscale convolutional neural network（MSCNN）and a gated recurrent unit（GRU）neural network，which provides an accurate data basis for subsequent lithology identification. Secondly，the wavelet packet adaptive thresholding method is employed for denoising and normalizing the data，so as to mitigate the impact of noise on lithology identification. Next，the Optuna framework is utilized to determine the hyperparameters of the XGBoost algorithm，and an efficient lithology identification model is established. Finally，the Shapley additive explanations（SHAP）interpretability method is used for the attribution analysis of the XGBoost model，which reveals the contribution of different features to lithology identification and enhances the interpretability of the model. Experimental results demonstrate that the proposed Optuna-XGBoost model achieves a comprehensive lithology identification accuracy of 79. 91%，outperforming traditional methods such as support vector machine （SVM），naive Bayes，and random forest，by margins of 24. 89%，12. 45%，and 6. 33%，respectively. The proposed lithology identification method based on the SHAP interpretability of the Optuna-XGBoost model has enhanced accuracy and interpretability，aligning well with the practical requirements of lithology identification in production scenarios.

High-quality three-dimensional seismic exploration data in the field forms the basis for subsequent data processing and interpretation. However，traditional manual or semi-automatic methods for assessing the quality of seismic records can no longer meet the efficiency requirements of high-density three-dimensional seismic exploration，nor can they locate the sources of seismic noise. Utilizing deep learning techniques supplemented by cosine similarity algorithms，t his paper proposes a classification method of seismic data quality based on deep learning. First，t he study automatically classifies the quality of seismic records into six categories：normal traces，strong seismic source interference traces，i ndustrial electrical interference traces，i nstrument problems（poor coupling between the geophone and the earth，blank traces），weak interference traces，and co-channel and anti-channel traces. A well-trained convolutional neural network achieves a speed of less than 3 seconds for quality assessment of single-shot seismic records（over 8000 traces），with an accuracy of 86% compared to manually classified results，and the evaluation results are objective. The results of model training show that this approach not only facilitates the rapid identification of different types of noise or instrument problems in seismic records，t hereby improving the efficiency and quality of seismic construction，but also provides an important decision-making basis for comprehensive assessment of seismic data quality by grades and zones，suitable for quality monitoring in mass seismic acquisition fields.

Kriging interpolation is a modeling method that can be combined with empirical knowledge，in which the accuracy of variogram determines the effect of the interpolation，thus affecting the construction of low-frequency model of seismic inversion based on Kriging interpolation. It is difficult for the traditional Kriging interpolation method to use multiple different theoretical models of variogram at the same time to improve the accuracy of low-frequency model construction，and only using a single theoretical model to solve the variogram leads to the uncertainty of theoretical model selection，the smoothing effect with a low fitting value of the variogram， and the hole effect caused by a long well distance. To solve the above problems，a neural network CNN-GRU model is introduced to adaptively fit the complex semi-variance relationship between the vector and the corresponding well and further realize the effective fusion of the spherical model，the Gaussian model，the exponential model，and the hole effect model，so as to address the uncertainty，the smoothing effect，and the hole effect of the variogram. The model takes into account the correlation between wells and conveniently realizes the point-by-point variation analysis with a convenient processing process，which can well match the randomness of the parameters selected by the variogram. The actual data show that the Kriging method based on the CNN-GRU model can establish a high-precision low-frequency model of seismic inversion，and it has a better effect than the traditional method.

The fast sweeping method is a traveltime calculation approach based on solving eikonal equations， which can quickly compute the traveltime of grid nodes by sweeping in four directions across the grid. Depen-ding on the number of nodes used in the local computational unit，there are two types of grids：the five-point triangular grid and the nine-point triangular grid. There are differences in the accuracy of traveltime calculations using the fast sweeping method with different grid types. To verify the extent of these differences，this paper compares the traveltime calculations of the fast sweeping method under the two grid types. Firstly，the relationship between travel time and slowness is established for different grid units. The eikonal equation is discretized， and a quadratic equation of the upwind difference scheme is derived. Herein，numerically solving the equation yields possible traveltime. Then，the study discusses the steps for traveltime calculation using the fast-swee-ping method in the two types of grids and provides the treatment method of grid boundaries and the causa-lity condition for traveltime calculation with nine-point triangular grids. Finally，tests are conducted using different velocity models（uniform velocity model，constant-gradient velocity model，and Sigsbee2a velocity model）， and the results indicate that nine-point triangular grids can effectively improve the accuracy of traveltime calculation using the fast sweeping method.

The impedance inversion method based on sparse constraint is widely used in oil and gas exploration. However，due to the limited sparse information extracted by traditional sparse constraints and the neglect of multitrace correlation，the inversion results contain artifacts and exhibit poor lateral continuity. To address this issue， this paper proposes a seismic impedance inversion method based on sparse constraint of logarithmic total variation. The proposed method utilizes a logarithmic penalty factor to extract sparse information. Compared with traditional sparse constraint，the logarithmic penalty factor can not only extract more sparse information but also ensure convexity of the objective function，reducing artifacts and enhancing inversion accuracy. Additionally，the horizontal logarithmic penalty factor is adopted to introduce multi-trace similarity information in inversion，enhancing lateral continuity of the inversion results. To improve the inversion stability，this study combines the initial model constraint with sparse constraint of logarithmic total variation to form the objective function. The alternating direction method of multipliers is used to solve this objective function. Finally，the proposed method and conventional methods are tested by the Marmousi2 model and field data. The results show that the proposed method can remove artifacts and improve lateral continuity and accuracy，compared to the conventional methods.

It is important for oil and gas exploration to accurately identifying buried subsurface salt bodies. However， traditional semantic segmentation algorithms still have problems in salt body recognition such as low recognition accuracy，poor edge recognition effect，and low recognition efficiency. This paper proposes a salt body re-cognition method based on the MC-Res2Unet network. The overall network architecture is improved on the basis of U-Net. First，the Res2net network is used as an encoder to extract the salt body feature information. Then，the convolutional block attention module（CBAM）is introduced after the convolution in the decoding layer to redistribute spatial and channel information of the salt body and suppress unimportant information. Finally，the multi-scale feature fusion module is utilized to fuse the spatial and semantic information for improving the salt body recognition accuracy. The MC-Res2Unet model proposed in this paper is validated on the TGS salt body dataset，achieving pixel accuracy of 96. 6% and an intersection over union to be 86. 8%. It has a better recognition effect on subsurface salt bodies than traditional semantic segmentation methods such as deeplabv3+ and DANet.

The widespread application of reflection seismic exploration benefits from not only the multiple cove-rage technique but also sparse and regular acquisition methods. However，due to the ubiquity of random media and their symbiosis with layered strata，single geometry，outdoor geophone array，and common midpoint （CMP）acquisition have shown many drawbacks. Based on the concept of full-wave-field seismic exploration and considering that the backscattering of seismic waves needs to achieve irregular spatial sampling through single-node，high-density，random source and receiver layout，this paper proposes the common-midpoint discretization（CMPD）seismic acquisition technology. First，it briefly reviews the signal sampling and spatial sampling theorems as well as the theories of random sampling and compressed sensing. Then，the application scope and implementation method of CMPD are clarified，such as local random layout of sources and receivers for regular arrangement，trace-density-controlled layout for irregular arrangement，and stakeless layout for timelapse seismic data acquisition. Finally，the functions and characteristics of CMPD are analyzed，such as redu-cing the collected footprints，suppressing the regular noise，increasing the scattered wave energy，facilitating multi-faceted processing，and realizing polymorphic seismic wave imaging. The experimental results show that the CMPD technology adopts random and irregular acquisition to improve the quality of the backscattering of seismic waves and has certain reference significance for promoting multi-geometry acquisition，time-lapse seismic data acquisition，multi-faceted processing，and wave field separation processing.

Nowadays，the seismic impedance inversion methods based on deep learning usually realize inversion through low-dimensional time series modeling，ignoring the topological structure information of the spatial structure of seismic impedance and thus resulting in low inversion accuracy. To solve this problem，this paper proposes a seismic impedance inversion method based on spatio-temporal modeling with spatio-temporal graph convolutional networks（STGCN）. Considering the topological structure and mutual correlation of seismic data，the method uses Mahalanobis distance to weight the spatial proximity of seismic data so that an adjacency matrix can be established. Furthermore，the Chebyshev polynomial is used to enlarge the spatial receptive field and reduce the number of parameters. The spatial structure features of seismic data are extracted efficiently， and the temporal correlation is captured by a gated recurrent unit. Finally，a spatio-temporal graph convolution unit is constructed for the mapping of seismic data and wave impedance in both time and space based on STGCN. The results of the model test and actual data inversion show that the proposed method improves inversion accuracy，has certain adaptability to noise，and can well reflect the lateral changes of strata.

Heterogeneous bodies such as faults，salt domes，karst caves，and fractures，as important targets in current structural and lithologic exploration，usually act as extremely complex diffractions on seismic records. Therefore，high-resolution diffraction imaging is of great significance for oil and gas exploration and development. This study proposes a diffraction imaging separation and enhancement workflow based on beam decomposition and spherical angle decomposition，which can be applied after imaging. This method is compatible with different migration algorithms and avoids the substantial computational load associated with pre-migration diffraction separation due to massive data volumes. Firstly，a sequential computation of sub-beams at varying scales，from large to small，is applied to the migrated data，followed by multiple iterations to achieve the preliminary separation of the diffraction wavefield. Subsequently，spherical angle decomposition is used to form gathers within specified stratigraphic dip ranges，which are referred to as dip image partition（DIP）gathers. Finally，the amplitude envelope attributes and similarity of the DIP gathers are calculated and applied to the preliminarily separated diffraction wavefield，which results in a high-resolution diffraction imaging volume. This workflow is applied for the first time to the imaging of the Mesozoic buried hill interiors of Bohai Bay，significantly improving the resolution for identifying small-scale geological discontinuities within the buried hill structures.

Geoelectric field measurement plays a crucial role in geophysical exploration techniques such as magnetotelluric and spontaneous potential（SP） exploration，and a solid non-polarized electrode（NPE） with long-term high stability is the basis of achieving high-precision geoelectric field observation. In practical applications，the performance of such electrodes significantly influences the accuracy of observations and exploration results. Therefore，how to maintain long-term high stability poses a challenge. Based on the reaction principle of electrodes，this paper enhances the stability of the NPE by enhancing the water retention and homogeneity of electrolyte through the addition of a trace of nano-fumed silica to prepare a gel. Then，improvements have been made to the size and position of the small channels within the internal cavity of the electrode to further enhance its stability and lifespan. Additionally，a removable maintenance cylinder is introduced to cater to harsh environmental conditions. Through repeated tests，the long-term high-stability NPE series，solid NPE，has been developed. The internal resistance，polarization potential difference，stability，self-noise level，temperature，and frequency response of the electrode are tested. The test results show that the range of any electrode couple is less than 0. 1 mV. The NPE series has a range drift of less than ±0. 05 mV in one day and less than ±0. 1 mV in one month. The internal resistance is less than 200 Ω，and the temperature coefficient is less than 25 μV/℃ . The main performance indexes have reached the international levels. This research provides an innovative domestic sensor solution for high-precision geoelectric field measurements.

The frequency-domain controlled source electromagnetic method（CSEM）offers distinct advantages compared to the traditional magnetotelluric（MT）method，such as robust resistance to interference and higher observed field magnitudes. It has found widespread application in fields like mineral resource exploration and engineering surveys. The precise definition of apparent resistivity，when appropriately applied，can offer a more intuitive representation of comprehensive subsurface electrical property variations. Research into the definition of apparent resistivity has been a fundamental topic in electromagnetic theory. Based on the formula of the electric field generated by a horizontal electric dipole on a uniform Earth surface，this study derived formulas for the frequency and spatial gradients of the horizontal electric field component E_x. The analytical solutions were utilized to compute the gradient responses of field magnitudes on a uniform Earth surface. Furthermore，the study calculated one-dimensional electromagnetic response with the weighted extrapolation method and employed a cubic spline interpolation algorithm to calculate difference values of field magnitudes. The accuracy of the formulas for apparent resistivity of frequency and spatial gradients was verified by the comparison between analytical solutions and difference values of the electric field. The field magnitude gradient in homogeneous half-space was used to be equivalent to the field magnitude gradient in a layered medium，and the dichotomy for numerical iteration was adopted to solve the nonlinear equation to obtain the apparent resistivity of electric field gradient. The calculation results demonstrate that the apparent resistivity calculation formula proposed in this study can effectively depict changes in the geoelectric structure. Substituting field magnitude gradient with difference values to calculate apparent resistivity can meet the precision requirements. Similar to wide-field apparent resistivity，the apparent resistivity of spatial gradient of the horizontal electric field converges in both the near and far regions to values approximating resistivities of the bottom and top layers，respectively，consistent with theoretical expectations. However，the sounding curve of the apparent resistivity of the frequency gradient does not converge and is thus unable to accurately represent changes in subsurface electrical properties，which is rendered incapable of sounding.

Shale reservoirs have both high-angle natural fractures and horizontal shale bedding fractures，which has significant influences on the expansion of artificial fractures under hydraulic fracturing and fracturing construction. To clarify the interaction between natural fractures and artificial fractures，this study first identified the development characteristics of natural fractures through core and field outcrop observations，electrical ima-ging logging identification，and seismic attribute prediction and then comprehensively analyzed the influences of natural fractures on the expansion law of artificial fractures and fracturing construction by relying on the downhole microseismic monitoring technology and considering the characteristics of changes in fracturing construction parameters. The results of the case analysis show that there are differences in the influence of natural fractures with different orientations on the hydraulic fracturing effect. High-angle natural fractures are prone to fracturing construction abnormalities such as repeated stimulation between fracturing sections，casing deformation， and casing damage. The influence of natural fractures on the expansion law of artificial fractures changes with different orientations，with EW natural fractures having a promoting effect，NS natural fractures having a bloc-king effect，and NE/NW natural fractures having a turning effect. The existence of horizontal shale bedding fractures causes artificial fractures to first rupture and extend laterally and then expand upward and downward. The overall performance of artificial fractures is that they do not expand high or extend far，and the construction pressure fluctuates in the high range，which is easy to cause sand blockage. The research results can provide a basis for the subsequent design of horizontal well trajectories and the optimization and adjustment of hydraulic fracturing parameters.

High-efficiency blended acquisition technology can greatly improve field acquisition efficiency and reduce seismic exploration costs while obtaining the same data quality. Deblending is one of the two core techniques to realize blended acquisition，which determines the final quality of the seismic data obtained. The sparse inversion deblending technique has been widely used and recognized in the industry. This paper improves the traditional sparse inversion deblending technique and proposes variable grid sparse inversion. The method optimizes the iterative convergence process by gradually changing the grid scale of the data during the iterative inversion process，so as to make it more in line with the nature of inversion in terms of extracting the signals with the strong event followed by the weak event and from the low to the high frequency，thus obtaining a better separation effect. In addition，to address the adverse effect of the lack of sources on the separation accuracy of sparse inversion，which is common in the field acquisition and results in the decline in quality of deblending，an empty-source compensation technique is proposed to improve the processing effect. The processing results of simulation and actual blended data show that this technique can achieve higher signal separation fidelity than the traditional sparse inversion deblending technique.

Pre-stack amplitude variation with offset（AVO）inversion is a crucial method for obtaining petrophysical properties. Traditional pre-stack AVO inversion methods often rely on the approximate reflection coefficient equation，which usually lose accuracy in specific geological environments or under large incident angles. In response，this paper introduces a pre-stack nonlinear inversion method for joint PP-PS wave based on the exact Zoeppritz equation. This method combines a multi-objective global optimization algorithm with the PP-PS joint inversion and can simultaneously optimize the two objective functions of PP and PS，thereby achieving fully nonlinear parameter inversion. To overcome the challenge of assigning weight coefficients for PS data in traditional PP-PS joint inversion methods，this paper establishes a multi-objective function for PP-PS joint inversion within the Bayesian framework and employs a multi-objective intelligent optimization algorithm called strength Pareto evolutionary algorithm 2（SPEA2）to solve the constructed multi-objective function for inversion. Tests are conducted on single-well synthetic seismic records，synthetic seismic records based on the Marmousi model，and seismic records of an actual field. The results show the method proposed in this paper，based on the exact Zoeppritz equation，can accurately estimate the elastic parameters of strata. It has better inversion results than traditional AVO inversion methods when dealing with seismic data for complex strata and large incident angles.

Recently，gas was obtained from Well Dongba 1 in the Longwangmiao Formation of Penglai Gas Field in the northern paleo-uplift of central Sichuan，which indicates that there is potential for new discoveries in the Longwangmiao Formation of Penglai Gas Field. Therefore，it is of great significance to clarify the distribution scale of reservoirs for predicting the exploration potential of the Longwangmiao Formation in Penglai Gas Field. However，the Longwangmiao Formation in the study area also faces challenges such thin strata，rapid lithological changes in overlying strata，and difficulty in identifying seismic response of reservoirs. Therefore， based on fully considering the influencing factors of reservoir identification in the Longwangmiao Formation， this paper conducts high-resolution data processing，establishes seismic response models for different zones with the help of zoned forward modeling analysis，and forms qualitative and quantitative prediction techniques for reservoirs. Finally，a comprehensive evaluation is made on the distribution characteristics of reservoirs in the Longwangmiao Formation of Penglai Gas Field. The research results show the followings：① Based on real drilling data and forward modeling analysis，it is clear that the thinning of strata in the Longwangmiao Formation will lead to unclear seismic response characteristics of its reservoirs，while high-fidelity high-resolution data processing can effectively improve the reservoir identification ability for thin strata. ② The change of overlying strata in the Longwangmiao Formation has a certain influence on the seismic response characteristics of reservoirs，and different zones have different seismic response characteristics. Four types of seismic response models are established through zoning. ③ Through petrophysical analysis，it is believed that the facies-controlled quantitative characterization method for reservoirs can effectively reduce the influence of complex lithology on reservoir prediction and improve prediction accuracy. The high-resolution data processing technology and the zoned reservoir prediction method can effectively improve the reservoir identification ability and accuracy in the Longwangmiao Formation. A total of 4200 km² of reservoir area in the Longwangmiao Formation of Penglai Gas Field is predicted，which clearly indicates that favorable reservoir areas in the Longwangmiao Formation of Penglai Gas Field have developed in a large area.

The compressed sensing（CS）seismic exploration technique can increase the trace density and im-prove the image quality at the same cost. The key challenge of the technique is the missing seismic data reconstruction. The reconstruction method based on sparse domain constrained inversion has the advantage of high reconstruction accuracy. However，the low computational efficiency limits its application in the industry. To solve the issue，this paper proposes a fast iterative thresholding 3D seismic data reconstruction method based on curvelet transform by optimizing the conventional reconstruction method based on sparse domain constrained inversion and applied it to industrial data processing. 3D curvelet transform has an excellent sparse signal characterization ability，and the computation redundancy is high. In actual processing，seismic data segmentation is adopted，followed by parallel computation of 3D curvelet transform coefficient，which improves computational efficiency. In addition，a fast iterative thresholding algorithm is designed to improve the convergence speed and the reconstruction accuracy in comparison with the conventional method. Finally，amploying for the CS acquisition data，the proposed method is validated to be effective in data reconstruction and is applicable to data reconstruction in industrial production.

The propagation of seismic waves in viscoacoustic media induces the attenuation of amplitude and phase dispersion，thereby influencing the resolution and accuracy of seismic imaging profiles. The Q-compensated reverse time migration method（QRTM）considering the compensation effect of attenuation in viscous acoustic media is an effective approach to improve the imaging accuracy and resolution in strongly attenuative areas，such as desert surfaces and chimneys. Efficient numerical solutions are achieved based on the viscoacoustic wave equation solution algorithm of the generalized standard linear solid model. However，a discrepancy between the phase velocity and the theoretical solution leads to position shifts in the structural imaging，compromising the imaging focus and accuracy. To address this issue，this paper proposes a method to correct the phase velocity in QRTM. This method introduces an optimization term into the equation and constructs an inversion solution of the correction coefficient to solve the objective function. Based on the solved correction coefficient， the phase velocity is corrected，thereby the attenuation compensation of amplitude is realized under the condition of maintaining the dispersion relationship of the phase velocity unchanged. This ensures accurate positio-ning of the imaging results and enhances imaging accuracy. The results of the theoretical model and actual data tests demonstrate that using the QRTM to correct phase velocity can compensate for the attenuation of amplitude and correct phase dispersion. This method is suitable for eliminating the impact of the attenuation effect on mid-deep imaging during ultra-deep exploration，thus improving the profile quality.

Time-frequency analysis is an important method that reveals the change regularity of signal spectrum over time. It is of great significance in the fields of seismic signal processing and feature analysis. To improve time-frequency analysis accuracy，this paper proposes a time-frequency analysis method based on local frequency by using the W-transform. The method uses local frequency to obtain the smooth dominant frequency of the signal，which is then applied to the W-transform to obtain the time-frequency spectrum of the signal. By smoothing the dominant frequency using the local frequency approach，this method effectively reduces the im-pact of noise，yielding a more accurate dominant frequency and improving the reliability of time-frequency analysis. In comparison to the traditional S-transform，this W-transform based on local frequencies partially overcomes dispersion at high frequencies and achieves higher time resolution at low frequencies. Compared with the W-transform，this method makes the spectral energy of the signal better concentrated in the energy center，having better focusing capability. The introduction of local frequencies enhances the W-transform，contributing to increased accuracy in time-frequency analysis. Finally，the effectiveness and superiority of the method are validated through numerical experiments.

With the continuous breakthrough of exploration and development technology，the Qiangtang Basin has been called an important potential area of oil and gas resources in China. The Qiangtang Basin is located in the northern part of the Qinghai-Xizang Plateau. Complex geological structural factors，strong wind interfe-rence，and unique perennial distribution of plateau permafrost and alpine and anoxic environments have brought many challenges to seismic exploration. The non-uniform distribution of high-speed permafrost can shield seismic wave energy，distort seismic wave travel time，and reduce the signal-to-noise ratio and consistency of seismic data. How to correctly understand the propagation mechanism of seismic waves in permafrost has become a key issue in seismic exploration in a plateau environment. Firstly，in response to the problem of permafrost in the Qiangtang Basin，this study measures the rock physical elastic parameters of permafrost in the field by using rock physical measuring equipment special for plateaus and establishes a complex structure model with permafrost by combining the field understanding and the actual two-dimensional seismic reflection profile. Secondly， the seismic wave field characteristics of permafrost are studied by finite difference elastic wave forward mode-ling. It is found that permafrost makes the surface wave more developed，which influences seismic wave travel time. When the direct wave field has a“cap”feature，permafrost shields the seismic reflection amplitude，resulting in a weaker seismic reflection amplitude. Finally，by comparing the reverse time migration profiles of the model with permafrost and the model without permafrost，it is found that the seismic reflection imaging energy below permafrost is weaker. The research results of seismic propagation characteristics and mechanism of permafrost provide theoretical support for permafrost identification，field seismic data acquisition，permafrost energy compensation，and migration imaging in Qiangtang Basin.

In the process of seismic data interpretation，it is necessary to highlight the physical properties of the target geological body for better characterizing the features of the target body. Traditional amplitude gain control（AGC）only adjusts the display effect of the data in the perspective of amplitude attributes，which cannot meet the more refined display requirements of seismic data interpretation. This paper proposes a frequency gain control（FGC）method，whose fundamental principle is as follows. An actual average amplitude spectrum is calculated for seismic data within a given time window，and at the same time，a desired amplitude spectrum with a controllable dominant frequency is constructed. The two amplitude spectra are matched to calculate the gain factor，and it is applied to the seismic data to achieve the purpose of adaptive FGC. On the premise of keeping the data frequency band range unchanged ，this method can obtain a better visual effect by adjusting the energy relationship of different frequency components in the data：it can highlight the low-frequency components of the seismic data for the overall display of geological structural trends and also the high-frequency com-ponents of the seismic data for the detail description of local geological structures. The application effect of actual data indicates that the method proposed in this paper has certain practical value.

Refraction and the first arrival wave traveltime tomography are two mainly used methods for near-surface model building and static correction. Refraction is appropriate to be applied in the area where the refraction layer is relatively stable，but cannot be directly applied to the case of prestack depth migration because the provided model is a layered one. The first arrival wave traveltime tomography can demonstrate the vertical and horizontal variation trends of the near-surface relatively well. However，due to factors such as insufficient actual data sampling and dramatically rugged topography，there exist problems of low accuracy and local illogicality of the inversion model. Therefore，this paper proposes a first arrival wave traveltime tomography method with refraction model constraint. As for the specific practice of the method，firstly the delay time and refraction velocity are calculated by the refraction method，and a geological refraction model is established based on information such as surface survey，first arrival，and near-surface structure. Then the refraction model is discretized in the principle of vertical traveltime equivalence，according to the vertical and horizontal variation rules of the actual surface velocity. Lastly，constrained tomography is applied to obtain the final near-surface velocity model. The theoretical model verifies the effectiveness of the proposed method. The actual data shows that the near-surface velocity model inverted by the proposed method is more reasonable and the application effect is better than that of the unconstrained tomography.

Ultra-deep fracture-cavity carbonate oil and gas reservoirs have become an important area for exploration and development in the Tarim Basin，showing promising prospects. The identification of effective ultra-deep fault-controlled fracture-cavity reservoirs in the Fuman area of the Tarim Basin is challenging，which affects the evaluation of reservoir effectiveness and the division of reservoir units. This paper conducts research on identification of fracture-cavity reservoirs with a seismic wave phase reconstruction method. First，the raw seismic data are decomposed in the phase domain. Then，the paper conducts one-dimensional and two-dimensional forward modeling of geological models. The phase component sensitive to the carbonate reservoir is selected. Finally，the component data volume with sensitive phase angle is reconstructed to generate a new phase seismic data volume. The research results show that seismic wave phase reconstruction technology can effectively remove the shielding effects caused by strong seismic reflections due to other factors in the strata，eliminating invalid signals and highlighting weak reflection signals from the reservoir. This technology can effectively delineate the favorable distribution range of fracture-cavity reservoirs，providing valuable information on favorable areas for reservoir evaluation and development planning. The use of seismic wave phase decomposition and reconstruction technology for identifying fracture-cavity carbonate reservoirs can serve as a useful refe-rence for evaluating similar reservoir characteristics in other areas.

The mechanism of land internal multiples is complicated，and internal multiple attenuation is a big challenge. Additionally，unequal source and receiver sampling in pre-stack seismic data of land does not satisfy the theoretical assumption of internal multiple prediction，which is an important problem encountered in land exploration. In response，this paper proposes a pre-stack internal multiple prediction method based on a four-dimensional data index tree real-time interpolation. First，to address the problem that the acquisition mode is not in line with the theoretical assumption，the paper comprehensively considers the center point position，azimuth， and offset of each seismic data to construct a four-dimensional data index tree to manage pre-stack seismic data. Then，data interpolation is carried out to obtain seismic data meeting the theoretical requirement in real time for internal multiple prediction，which avoids regularization of pre-stack seismic data. The model data and actual data both show that this method is effective in internal multiple prediction. After pre-stack internal multiple attenuation，the phenomenon of reverse velocity in the velocity spectrum is significantly improved for actual 3D data，and breakpoints and fractures become clearer on the imaging profile.