The gravity anomaly inversion，which infers the density distribution of subsurface anomalies from surface gravity data，is an essential tool in geophysical exploration and is widely applied in fields such as oilfields， mineral deposits，geological structures，and underground works detection. Traditional gravity inversion methods face challenges of complex computation，low resolution，and dependence on prior information for inversion results. However，deep learning-based gravity anomaly inversion techniques show significant advantages，particularly in terms of improving inversion accuracy and reducing computation time，without the reliance on initial models or prior information. This paper reviews the development and limitations of traditional gravity anomaly forward and inversion methods and summarizes the current research on deep learning-based gravity inversion methods. Meanwhile，it introduces the improvements and innovations of different gravity inversion problems in four respects，including data preparation，network models，network optimization，and network validation. Additionally，the application effect of various gravity inversion methods on the measured data from Vinton Dome in Louisiana，the USA，and the San Nicolás ore deposit in Mexico. The multi-task framework CDUNet yields the most accurate inversion depth values on data of Vinton Dome，while the 3D U-Net++ network obtains clearer and more accurate inversion results on the data of the San Nicolás ore deposit than the U -Net network.

With the widespread application of node instruments in seismic exploration，their quality control （QC）problems become increasingly prominent，and their“blind acquisition”mode lacks real-time data transmission capability，making it impossible to monitor and effectively analyze the quality of single -shot data in real time. This results in delays in the synthetic analysis of seismic acquisition data and affects the timeliness of data QC and exploration decision -making responses. Therefore，based on the single -shot acquisition mode combining wired monitoring arrays and node instruments，this paper develops a real-time QC technology for single-shot data acquisition in node instrument“blind acquisition”for the characteristics of seismic data from monitoring arrays. Firstly，the single -shot energy is calculated based on the spherical spreading principle to achieve an accurate analysis of single-shot energy. Secondly，an environmental noise analysis technique based on the number of disturbed channels is employed，with the proportion of disturbed channels calculated to assess environmental noise. Finally，supporting real-time monitoring software is developed to enable real -time monitoring of single -shot excitation energy and noise. This technology is applied to the full-node 3D seismic acquisition project in the NCB area of the Sichuan Basin. The results show a high correlation between the energy of the monitoring array and the full -shot energy，as well as high consistency in the proportion of disturbed channels. This significantly improves seismic exploration efficiency and data quality，provides an effective method for real-time QC of blind acquisition seismic data，and holds significance in the commercial application of 5G real -time data transmission of nodes.

A gas field in western China is a shallow biogas reservoir. Due to the influence of gas -bearing reservoirs，the quality of P -wave seismic data is poor and cannot provide imaging，thus making the detailed description of the structure and reservoir quite challenging. Therefore，this paper tackles the problems by employing 3D S -wave seismic data to restore the structural morphology of the gas cloud area and characterize high -quality reservoirs. First，an applicability evaluation of the seismic data is conducted to determine the applicable range of different types of seismic data. Then，S-wave logging and S -wave VSP data are adopted for joint calibration of wellbore and surface seismic data，and based on this，fine structural interpretation is conducted. Finally，S-wave attributes and S-wave inversion are utilized to characterize the reservoir distribution，with development data combined to verify the distribution of high -quality reservoirs. The proposed 3D S -wave seismic interpretation workflow produces a high-precision structural map and accurate reservoir prediction results in the study area，effectively solving the structural and reservoir description problems in the gas cloud area. Finally，strong technical support is provided for the exploration and development planning of the gas cloud area.

With the advancement of geological theory and geophysical prospecting technology，shale reservoirs have emerged as the crucial target of oil and gas exploration and development. Research indicates that the development of lamination or stratification can conspicuously enhance the physical property conditions of the reservoir. Drilling outcomes also affirm that stratification characteristics such as the density of lamination tend to determine the development of oil and gas. Nevertheless，the study of lamination or stratification remains in its nascent stage. Therefore，it holds significant importance for the characterization of shale reservoirs to clarify the micro rock physical mechanism for shale stratification and search for inversion parameters directly characterizing the stratification. Based on the prevailing rock physics theory，the stratification inducing factors，such as lamination density，mineral shape，and random arrangement of clay blocks，are studied respectively. The corresponding rock physical models of each factor are summarized and the degree of influence is analyzed. On this basis，a shale rock physical model considering stratification is established and successfully applied to the shear wave prediction of actual logging data. To achieve the direct inversion of shale stratification characteristics，the stratification indicators are constructed by employing the concept of upper and lower limits in rock physics and utilized as the criterion to measure the stratification. At the same time，the elastic impedance form of three approximate equations of reflection coefficients of VTI media is derived. The relationship between the stratification indicators and the model parameters in the equation is retrieved by a machine learning algorithm，thereby enabling the stable inversion of the stratification indicators. The application of the actual data demonstrates that the inversion results are in excellent agreement with the actual drilling，which effectively delineates the distribution of the reservoir and is conducive to guiding the horizontal evaluation of the shale reservoir.

Channel identification is crucial for predicting fluvial facies reservoirs. However，when the P-wave impedance contrast between channel sandstones and surrounding rocks is minimal，it is difficult to use only poststack P -wave seismic data for channel identification. S -wave data can effectively enhance the reliability of predicting the spatial distribution of channels. However，the combined identification process of P-wave and S-wave involves challenges such as difficult parameter selection，high subjectivity，and extended working cycles， leading to inefficiencies and reduced reliability. This paper proposes an automatic channel identification methodbased on the joint P-wave and S-wave seismic data. First，to address the issue of insufficient sample data，it puts forward a method for automatically generating synthetic forward modeling samples of 3D channel geological models based on actual data interpretation and channel interpretation results，effectively expanding the sample data set. Subsequently，a new 3D automatic channel identification network structure is then designed， which effectively integrates P-wave and S-wave seismic data，enhancing the reliability of the identification results. Finally，the proposed method is applied to identify tight gas channel sandstones in a work area in southwestern China. Compared with traditional seismic attribute analysis and intelligent identification results relying on a single data type，the proposed method demonstrates higher efficiency and reliability，validating its applicability.

The traditional method of seismic data acquisition involves separate excitation between shots，leading to low construction efficiency. The simultaneous-source acquisition technique，which involves simulta-neous or delayed excitation of multiple shots，has significantly improved construction efficiency. To process multi-source seismic data conventionally，it is essential to separate the mixed data into single-source data. However，traditional data separation methods suffer from low efficiency. On the basis of the separation method using sparse inversion，this paper develops fast projection onto convex sets（POCS）algorithm through introducing an inertial parameter into the conventional POCS algorithm. Additionally，to address the problem that random time-delay encoding causes local continuity of aliasing noise and thus reduces separation accuracy，this paper proposes a method of periodic sinusoidal time-delay encoding. During the iterative process，the curvelet transform is used as a sparse basis，the soft thresholding function as the threshold model，and the exponential threshold formula as the threshold update formula. As a result，the blending interference from other sources becomes more discrete，and the aliasing noise is distributed uniformly within a certain delay time，which reduces the difficulty in simultaneous-source data separation. The results obtained from theoretical analysis and processing of measured data demonstrate that the two methods proposed in this paper not only greatly improve the convergence speed but also effectively enhance the accuracy of simultaneous -source data separation and have good anti-noise performance.

Conventional post-stack seismic data are usually affected by absorption attenuation，resulting ina lower peak frequency，narrower frequency band，low resolution，and poor inversion prediction effect. Therefore，an improved Q estimation and inverse Q -filtering method is proposed to improve the resolution of seismic data and obtain the fidelity and amplitude -preserving seismic data. Firstly，in Q estimation，given the problem that the extracted wavelet amplitude spectrum deviates from the actual situation due to the thin -layer tuning effect that affects the Q estimation accuracy，the complete ensemble empirical model decomposition with adaptive noise（CEEMDAN）method is introduced，which eliminates the interference of reflection coefficients by decomposing and reconstructing the log amplitude spectrum and thus obtains the wavelet amplitude spectrum that removes the tuning effect. Combined with the centroid frequency shift of the energy spectrum，higher-accuracy Q values are obtained. Then，in terms of inverse Q filtering，the amplitude compensation function of the time-varying gain stabilization factor method is optimized to overcome the density dependence and obtain more stable inverse Q -filtering results. The actual data processing results show that the proposed method can obtain fidelity and amplitude-preserving high -resolution post-stack seismic data，which lays a reliable data foundation for subsequent exploration and development of the study area.

The accuracy of surface structure investigation greatly influences the quality of seismic field acquisition. However，the commonly used surface structure investigation methods fail to accurately characterize the thin lithological change，and construction under special surface conditions is difficult. To solve this problem， the optical fiber sensor acquisition technology is applied to the investigation of surface structure. The advantages of a high spatial sampling rate and easy construction of optical fiber are used to solve the problems of traditional methods. The feasibility of optical fiber acquisition is verified by comparing the data collected by geophone. At the same time，the same point comparison test of surface velocity inversion and Q-value calculation is carried out. The results show that optical fiber data can identify more lithological change interfaces than geophone data. However，the signal -to -noise ratio of fiber optic data is low，with a narrow effective band range and unstable Q-value calculation results. This study shows that the optical fiber sensor acquisition technology has great potential in the application of surface investigation. Still，more perfect quality control means are needed to improve the quality of data acquisition.

Intelligent seismic velocity inversion is currently a hot and challenging topic in seismic exploration research. Nevertheless, the complex structure of deep learning networks demands significant computing power from hardware devices, which restricts the application of the model in scenarios with large data volumes and high timeliness requirements. To address these practical issues, in this paper, the U-Net is improved based on the concepts of feature engineering and model lightweighting, and the inversion networks U-Net vG for GPU and U-Net vC for CPU are proposed. Firstly, the characteristics of the velocity inversion network are analyzed to deduce the lightweighting principles of convolutional neural networks. Subsequently, lightweight processing is conducted on the multi-scale module, attention gate module, and feature extraction module to obtain a lightweight convolutional neural network for velocity modeling, which reduces the network volume while maintaining prediction accuracy. Data test results demonstrate that the training process of the proposed network has lower requirements for high-performance hardware resources, and that the network enables efficient velocity inversion, possesses higher seismic velocity inversion accuracy, and exhibits superior noise resistance. It provides a new idea for solving the computing power bottleneck problem in seismic data inversion.