Journal of Applied Geophysics最新文献

{"title":"Horizontal fracture prediction in shale gas reservoirs based on a generalization-enhanced framework integrating rock physics-driven data augmentation and CNN","authors":"Xiaodong Zhang,&nbsp;Zhiqi Guo,&nbsp;Cai Liu","doi":"10.1016/j.jappgeo.2025.106075","DOIUrl":"10.1016/j.jappgeo.2025.106075","url":null,"abstract":"<div><div>Fracture detection is essential for characterizing shale gas reservoirs. Although amplitude variation with azimuth methods are widely applied to predict vertical fractures, identifying horizontal fractures remains challenging due to their complex seismic responses, which differ from azimuthal anisotropy signatures. Quantitative seismic interpretation that integrates rock physics modeling with deep learning provides a promising framework for horizontal fracture prediction. However, the representativeness of available data poses a key limitation in areas with sparse borehole control, constraining the generalization capability of predictive models. A generalization-enhanced framework that combines rock physics-driven data augmentation with convolutional neural networks (CNN) is proposed to address this limitation. A shale-specific rock physics model for horizontal fractures is first established, followed by a model-based inversion scheme to estimate horizontal fracture density from well logs. The estimated fracture densities are then statistically expanded as random variables to generate augmented datasets that simulate spatial variability beyond borehole control. Corresponding elastic properties are computed using the rock physics model, forming physics-constrained datasets for CNN training. Cross-validation results demonstrate that the proposed data augmentation strategy reduces the root-mean-square error (RMSE) of horizontal fracture density estimation by approximately 14 %. Field application further confirms that the augmented model improves consistency with log-derived fracture densities and mitigates spurious anomalies compared with the non-augmented approach. The proposed framework thus provides a physics-guided and data-augmented methodology for robust prediction of horizontal fracture density, offering enhanced fracture characterization in shale gas reservoirs.</div></div>","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"245 ","pages":"Article 106075"},"PeriodicalIF":2.1,"publicationDate":"2025-12-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840194","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Three-dimensional anisotropic modelling of magnetotelluric data to determine the boundary between cap rock and reservoir formation: A case study of the Sarab field, Iran","authors":"Mohammad Filbandi Kashkouli ,&nbsp;Matthew J. Comeau ,&nbsp;Milad Farshad ,&nbsp;Abolghasem Kamkar-Rouhani","doi":"10.1016/j.jappgeo.2025.106077","DOIUrl":"10.1016/j.jappgeo.2025.106077","url":null,"abstract":"<div><div>Reservoirs of interest for resource exploration, including geothermal and hydrocarbon reservoirs, commonly have an impermeable cap, which traps fluids below. Identifying this boundary is important for resource development. The cap rock for hydrocarbon reservoirs in southwest Iran contains evaporites and thus some geophysical exploration methods, specifically seismic reflection, have faced problems recovering subsurface information in this environment. As an alternative, we generate an electrical resistivity model from magnetotelluric (MT) data. Furthermore, we consider three-dimensional triaxial electrical anisotropy, which is rarely done. The study objectives are to a) define and map the boundary between the cap rock and the principal reservoir, b) characterize geological and tectonic formations in the area, and c) analyze the tectonic factors influencing the evolution of the region. A total of 359 MT measurements were acquired across the Sarab field in an array consisting of five profiles separated by &gt;2000 m with a measurement spacing of &gt;200 m. Transient electromagnetic (TEM) measurements were co-located with the MT measurements at 181 locations and used to correct for static shifts. Isotropic and anisotropic inversions of the MT data were performed, using all impedance tensor elements. The anisotropic electrical resistivity model exhibits both a significantly better alignment with the depths of geological formations known from drilling data and a better fit to the data. Therefore, the boundary between the primary cap rock and principal reservoir, the Gachsaran and Asmari formations, is defined and mapped across the survey area. In addition, major tectonic and fault-related features in the region are identified.</div></div>","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"246 ","pages":"Article 106077"},"PeriodicalIF":2.1,"publicationDate":"2025-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145928697","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Retrieving reflections using passive seismic modeling: A case study based on passive seismic operation in Dehdasht Area, SW Zagros, Iran","authors":"Fatemeh Alsadat Tayeb Hosseini ,&nbsp;Zaher-Hossein Shomali ,&nbsp;Javad Jamali ,&nbsp;Mohammad Reza Hatami","doi":"10.1016/j.jappgeo.2025.106076","DOIUrl":"10.1016/j.jappgeo.2025.106076","url":null,"abstract":"<div><div>Passive seismic studies have made significant advancements in subsurface structure modeling. However, retrieving body-wave reflections remains challenging. This study aims to overcome this challenge through transient noise modeling and field-data application in the Dehdasht Area in southwestern Iran. In the synthetic modeling, noise sources were randomly distributed in both time and space to emulate the stochastic nature of passive seismic noise. Reflections were extracted using cross-correlation and cross-coherency techniques from both the synthetic model and the real Dehdasht data, and common virtual shot gathers were generated after suppressing surface waves to enhance reflection visibility. Even with a limited number of stations (13) and large inter-station spacing (2 km), The results demonstrate that passive seismic interferometry can successfully retrieve deep reflection waves (up to ∼7 s two-way travel time), largely independent of data acquisition geometry. Moreover, reflection hyperbolas in the virtual shot gathers were clearer when using the cross-coherency method compared to cross-correlation. A comparative analysis between common virtual shot gathers and corresponding active-source shots confirmed the consistency of retrieved reflections, highlighting the potential of passive seismic interferometry as a complementary tool to active source methods, particularly in areas with complex geological structures and high wave attenuation observed in active-source data and those identified in passive seismic recordings.Keywords: cross-coherency, cross-correlation, modeling, reflection, passive signals, virtual shot gather.</div></div>","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"245 ","pages":"Article 106076"},"PeriodicalIF":2.1,"publicationDate":"2025-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840195","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Effect of data filtering on source mechanisms inverted from surface microseismic monitoring array","authors":"Toufik Chtouki ,&nbsp;Matej Petružálek ,&nbsp;Frantisek Staněk ,&nbsp;Leo Eisner ,&nbsp;Zuzana Jechumtálová ,&nbsp;Naveed Iqbal ,&nbsp;Umair bin Waheed","doi":"10.1016/j.jappgeo.2025.106070","DOIUrl":"10.1016/j.jappgeo.2025.106070","url":null,"abstract":"<div><div>The source mechanisms of induced microseismic events help understanding underground operations and mitigating hazards associated with induced seismicity. However, the uncertainty in the inverted source mechanisms is not well understood. In this study, we examine the impact of digital filters applied to dense surface monitoring data on the inverted source mechanisms derived from P-wave amplitudes. Ten filters, designed and used to increase signal to noise ratio, were tested. Filtering strongly affects both the shear and non-shear components of the full moment tensor. The differences in shear component orientation can exceed 20° in Kagan angle for some filters, despite the excellent coverage provided by the monitoring network. By constraining the inversion to pure shear mechanisms, the orientation was more stable. The smallest errors were observed with bandpass, interferometry, wavelet (with a well-chosen wavelet), and Wiener filters. On the other hand, the SVD and AGC filters resulted in largest changes in source mechanisms. Our results show that data filtering can lead to significant errors in the source mechanisms, which could potentially be misinterpreted if used to infer stress or other reservoir parameters.</div></div>","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"245 ","pages":"Article 106070"},"PeriodicalIF":2.1,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884367","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Rock physics modeling and seismic responses of deep carbonate rocks in Tahe oilfield","authors":"Chenglong Wu ,&nbsp;Hemin Yuan ,&nbsp;Xin Zhang","doi":"10.1016/j.jappgeo.2025.106072","DOIUrl":"10.1016/j.jappgeo.2025.106072","url":null,"abstract":"<div><div>Controlled by multi-stage tectonic-karst processes and high-temperature high-pressure (HTHP) environment, the deep carbonate reservoir in Tahe oilfield has complex seismic responses. To characterize these intricate seismic responses, we integrate a comprehensive rock physics modeling workflow with seismic forward modeling to bridge micro-scale elastic properties and macro-scale seismic signatures. We constructed a dual-porosity rock model by combining the differential effective medium (DEM) model and Gassmann equation. Critically, we incorporated temperature-pressure-salinity corrections for fluid properties and modeled the effects of pressure and temperature on the rock frame. The model was then used to generate reservoir parameters for seismic forward modeling. The modeling results demonstrated that the Gassmann equation outperformed the DEM model, and P-wave velocity prediction was improved by adding HTHP corrections and salinity. The seismic forward modeling results revealed that porosity and pore structure are the dominant controls on seismic features, with fluid type being minor unless gas is present. This study quantitatively characterized the seismic rock physics properties of the deep carbonates in the Tahe Oilfield, providing a robust method for accurately predicting velocities and seismic responses of carbonates in similar geological settings.</div></div>","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"245 ","pages":"Article 106072"},"PeriodicalIF":2.1,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840307","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Failure laws of the backfill borehole surrounding rock under three-dimensional non-hydrostatic stress field: Insight into the first principal deviatoric stress and strain energy density","authors":"Chunkang Liu ,&nbsp;Hongjiang Wang ,&nbsp;Shuangcheng Du ,&nbsp;Yafei Zhai","doi":"10.1016/j.jappgeo.2025.106071","DOIUrl":"10.1016/j.jappgeo.2025.106071","url":null,"abstract":"&lt;div&gt;&lt;div&gt;Effectively analyzing the failure risk in backfill borehole (BFB) is important in exploiting metal and coal resources in complex deep formations. The distribution laws of the first principal deviatoric stress (FPDS) field and strain energy density (SED) in BFB surrounding rock are analyzed theoretically. The analysis is conducted under a three-dimensional (3D) non-hydrostatic stress field to explore the failure laws of the surrounding rock. The correctness of theoretical results is verified by numerical simulation of plastic zone and engineering case. Results show that the FPDS field in the BFB surrounding rock varies obviously in different dominant stress fields. The maximum of FPDS occurs within the range of 55° to 65° under &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;H&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;&gt;&lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;V&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;&gt;&lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;h&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;, and the field displays an ‘8’-shaped distribution, indicating a potential ‘8’-shaped failure zone. When &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;V&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;&gt;&lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;H&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;&gt;&lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;h&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt; or &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;H&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;&gt;&lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;h&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;&gt;&lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;V&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;, the maximum of FPDS appears in the direction parallel to &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;H&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt; or &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;h&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;, and the field presents either an ‘8’-shaped or elliptical distribution. Moreover, the distribution characteristics of SED partially reveal the failure pattern of the BFB, and the theoretical SED analysis results show good agreement with the numerical simulation of the BFB plastic zone, thereby confirming the theoretical accuracy. The SED parallel to the direction of &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;h&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt; and &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;H&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt; differs significantly in both magnitude and its variation with changes in &lt;em&gt;λ&lt;/em&gt;. Especially, under the conditions of equal &lt;em&gt;p&lt;/em&gt; and unequal &lt;em&gt;q&lt;/em&gt;, the SED decreases with &lt;em&gt;λ&lt;/em&gt; within the range of 0° to 15° in the &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;H&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt; dominant stress field with &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;V&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;&gt;&lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mi&gt;h&lt;/mi&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;, while which increases with &lt;em&gt;λ&lt;/em&gt; under other conditions. A case study of BFB in a copper mine mutually verifies the consistency the FPDS field and SED distribution of the surrounding rock. Furthermore, the minimum mud-wall pressure is calculated using a Lode angle function considering the third invariant of the deviatoric stress, bringing a 24.206 % reduction of the growth rate. By extending classical mechanics with 3D stress and energy considerations, the r","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"245 ","pages":"Article 106071"},"PeriodicalIF":2.1,"publicationDate":"2025-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840193","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A stochastic deconvolution via trans-dimensional Markov-chain Monte Carlo","authors":"Seunghoon Han ,&nbsp;Seokjoon Moon ,&nbsp;Hyunggu Jun ,&nbsp;Youngseo Kim ,&nbsp;Yi Shen ,&nbsp;Yongchae Cho","doi":"10.1016/j.jappgeo.2025.106063","DOIUrl":"10.1016/j.jappgeo.2025.106063","url":null,"abstract":"<div><div>Deconvolution is a crucial data processing step for enhancing the resolution of seismic exploration data, thereby enabling subsurface structures to be accurately interpreted. However, traditional deconvolution methods using an inverse filter of source wavelets provide unique results that do not account for the natural attenuation of wavelets with depth, leading to inherent accuracy limitations and difficulty in evaluating the uncertainty. This paper introduces a novel deconvolution method called Stochastic-Decon, which processes data through stochastic inversion rather than the traditional approach of applying an inverse filter. The method estimates the positions of stratigraphic boundaries from the posterior distribution of interface boundaries obtained through inversion. And it calculates the reflection coefficients from the posterior distribution of the impedance model. To evaluate the proposed stochastic deconvolution algorithm, we created a 1D model and verified the algorithm through application to a synthetic example. The algorithm was subsequently applied to 3D data from the Norne field to assess its applicability to real data. The results with spectral analysis and well-log data demonstrated that the proposed algorithm distinctly delineates stratigraphic boundaries, enhancing data resolution and suppressing source wavelets. These findings are expected to help identify stratigraphic boundaries and physical properties contrasts in future seismic exploration results. This paper also presents discussions and studies on the parameter settings necessary for detecting interlayer boundaries.</div></div>","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"245 ","pages":"Article 106063"},"PeriodicalIF":2.1,"publicationDate":"2025-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790736","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine Learning aids seismic inversion in reservoir characterization: A case study","authors":"Mohammed Farfour","doi":"10.1016/j.jappgeo.2025.106065","DOIUrl":"10.1016/j.jappgeo.2025.106065","url":null,"abstract":"<div><div>Seismic inversion plays a pivotal role in reservoir characterization, enabling interpreters to transform seismic data into physical, elastic, and petrophysical properties directly related to reservoir lithology and fluid content. From seismic inversion products (e.g., P-wave and S-wave impedances and density), a wide range of reservoir attributes can be derived. These include Vp/Vs ratios, Poisson's ratio, bulk modulus, porosity, water saturation, effective stress, and pore pressure, among others. Successful seismic inversion relies on high-quality seismic data and a sufficient number of wells with the necessary logging data. However, interpreters often face challenges due to the lack of critical well logs, such as P-wave and S-wave velocity logs. To address this, several approaches, including machine learning, have been developed. In this study, seismic data and well logs from offshore Australia were prepared for seismic inversion to extract various attributes related to reservoir lithology and fluid content. Three reservoirs were identified in the study area using petrophysical logs such as gamma ray, neutron, density porosity, and resistivity. However, P-wave and S-wave logs were available for only two of the reservoirs. To overcome this limitation, machine learning—specifically an artificial neural network (ANN)—was utilized to predict the missing logs for the third reservoir. All available logs were used for training and testing the ANN. The trained ANN model was subsequently validated on wells excluded from the training process and demonstrated high accuracy in predicting the P-wave and S-wave logs. Following this validation, the ANN was applied to generate the missing logs for the target reservoir. Using the complete set of logs, a new seismic inversion was conducted to produce P-wave and S-wave impedance volumes. These impedance volumes were further used to derive additional elastic properties and facilitate comprehensive geophysical reservoir characterization.</div></div>","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"245 ","pages":"Article 106065"},"PeriodicalIF":2.1,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790735","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Simultaneously seismic dip estimation and random noise attenuation via dip-informed Radon dictionary","authors":"Kuijie Cai,&nbsp;Xing Zhang,&nbsp;Yuqing Wang,&nbsp;Shujuan Wang","doi":"10.1016/j.jappgeo.2025.106059","DOIUrl":"10.1016/j.jappgeo.2025.106059","url":null,"abstract":"<div><div>Seismic dip is a key attribute for subsurface interpretation and subsequent processing such as noise attenuation and seismic interpolation. However, existing estimation methods are often vulnerable to strong random noise and intersecting seismic events. Strong random noise can severely degrade data quality and complicate the interpretation of subsurface structures. We propose a Dip-Informed Radon Dictionary (DIRD) method that simultaneously estimates seismic dip and suppresses random noise within a unified iterative framework. The core idea is the joint estimation of dip values and the denoised seismic signals by linking them through a dip-informed Radon dictionary. Noise is attenuated via a sparse reconstruction algorithm based on the dip-informed Radon dictionary, while more accurate dip estimates are obtained during the iterative suppression process. The algorithm alternately refines the dip parameters and denoised signals via sparse optimization, which significantly improves its robustness to heavy random noise. Furthermore, the DIRD framework decomposes seismic patches into multiple dip components, providing a more accurate estimation for intersecting events within the same spatial region. Experiments on synthetic data with an input SNR of −3.99 dB show that the DIRD method achieves an output SNR of 3.27 dB, outperforming FXDECON(2.25 dB) and APF(2.50 dB). The DIRD method also demonstrates superior accuracy and robustness of dip estimation compared to nonlinear PWD, structure tensor, and Radon methods in terms of mean squared error, mean absolute error, and standard deviation.</div></div>","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"245 ","pages":"Article 106059"},"PeriodicalIF":2.1,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790734","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Detection of mining-induced microseismicity through a deep convolutional neural network","authors":"Sepideh Vafaei Shoushtari,&nbsp;Bernard Giroux,&nbsp;Erwan Gloaguen,&nbsp;Maher Nasr","doi":"10.1016/j.jappgeo.2025.106069","DOIUrl":"10.1016/j.jappgeo.2025.106069","url":null,"abstract":"<div><div>The underground extraction of mineral resources is often closely linked to induced microseismic events. The use of a seismic network to continuously monitor mining-induced seismicity to reduce risks and improve operational safety is common. For this monitoring to be effective, a comprehensive catalog of microseismic events, containing low-to high-magnitude events, is essential to evaluate the response of the rock mass to mining activities. However, detecting low-magnitude events based on manual picking or automated conventional approaches has been challenging in mining environments owing to the inherent noise level. Recent advancements in deep learning and data-driven methods, particularly Convolutional Neural Networks (CNNs) trained on extensive seismic datasets, have shown improved capabilities in automated event detection and arrival phase picking on seismic data recorded by regional seismic networks. In this study, we assessed the performance of PhaseNet, a deep learning arrival-time picking method, in detecting the P- and S-wave arrivals of mining-induced microseismic events at different noise levels. As access to high-quality, labeled microseismic datasets for such mining applications is rare, a realistic three-component synthetic dataset was generated using full-waveform modeling. This simulation accounted for the geological conditions and network geometry specific to a mine in Ontario, Canada. The mine, which integrates copper and nickel operations, experiences considerable mining-induced earthquakes annually, posing risks to miners and infrastructure. The simulation includes a variety of source mechanisms with different magnitudes and offers more than 270,000 labeled seismograms. The results from the PhaseNet-trained model, which utilized the simulated dataset, demonstrated its effectiveness in managing noisy waveforms. This capability allows the detection of low-magnitude events within the mine environment, which may be overlooked by traditional methods. Furthermore, the model shows high accuracy in picking both the P- and S-wave arrival times, achieving precision rates exceeding 0.9. Tests on real data were performed in three different scenarios. The first scenario involves training the model exclusively using real data. The second scenario combines synthetic and real data to retrain the model previously trained with synthetic data only. Finally, the third scenario focuses on retraining the pre-trained model using only synthetic data. All these trained models were used to evaluate the performance on the real test dataset. The results indicate that the model retrained with synthetic and real seismograms yielded the best arrival time predictions for the mine dataset.</div></div>","PeriodicalId":54882,"journal":{"name":"Journal of Applied Geophysics","volume":"245 ","pages":"Article 106069"},"PeriodicalIF":2.1,"publicationDate":"2025-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790708","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}