Pub Date : 2025-02-01DOI: 10.1016/j.cageo.2024.105815
Parth Hasabnis , Enhedelihai Alex Nilot , Yunyue Elita Li
In this paper, Urban Seismic Event Detection (USED), a deep learning-based technique, is introduced to extract information about urban seismic events. As large labelled datasets for this research are not publicly available, a scheme is presented to synthesize training data by using a small batch of manually labelled field data. Unlabelled field data can also be leveraged while training using semi-supervised learning, and a mean-teacher approach is discussed. The trained models are tested using synthetic and real data. It is successfully demonstrated that deep learning models can identify urban seismic events when trained solely on synthetic data. The insights and shortcomings of this approach are also discussed while providing potential directions for future research.
{"title":"Introducing USED: Urban Seismic Event Detection","authors":"Parth Hasabnis , Enhedelihai Alex Nilot , Yunyue Elita Li","doi":"10.1016/j.cageo.2024.105815","DOIUrl":"10.1016/j.cageo.2024.105815","url":null,"abstract":"<div><div>In this paper, Urban Seismic Event Detection (USED), a deep learning-based technique, is introduced to extract information about urban seismic events. As large labelled datasets for this research are not publicly available, a scheme is presented to synthesize training data by using a small batch of manually labelled field data. Unlabelled field data can also be leveraged while training using semi-supervised learning, and a mean-teacher approach is discussed. The trained models are tested using synthetic and real data. It is successfully demonstrated that deep learning models can identify urban seismic events when trained solely on synthetic data. The insights and shortcomings of this approach are also discussed while providing potential directions for future research.</div></div>","PeriodicalId":55221,"journal":{"name":"Computers & Geosciences","volume":"196 ","pages":"Article 105815"},"PeriodicalIF":4.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143093173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.cageo.2024.105840
Ce Qin , Wenliang Fu , Ning Zhao , Jun Zhou
The direct current (DC) resistivity method is an effective method for detecting subsurface structures with conductive differences. The three-dimensional forward modeling of DC resistivity is essential for data processing and interpretation. In the forward modeling process, solving the large-scale linear system is the most time-consuming step. Traditional algebraic multigrid (AMG) method has been successfully applied to solve the linear system. However, the performance of AMG will deteriorate when dealing with high-order discretization and highly stretched grids. Additionally, in the context of adaptive mesh refinement, the hanging nodes will further complicate the solving process. To address these challenges, we propose a novel geometric multigrid (GMG) method combined with local smoothing for solving the linear system in three-dimensional DC resistivity forward modeling. In this work, we employ high-order basis functions to discretize the problem. To further enhance the accuracy of the numerical solution, the mesh is adaptively refined based on the goal-oriented posterior error estimator. We utilize a V-cycle geometric multigrid on locally refined grids and the hanging node issue is effectively addressed through local smoothing. We also employ the mesh partitioning technique to parallel the solution process. The efficiency, robustness, and parallel performance of our algorithm are verified through various numerical examples.
{"title":"3D adaptive finite-element forward modeling for direct current resistivity method using geometric multigrid solver","authors":"Ce Qin , Wenliang Fu , Ning Zhao , Jun Zhou","doi":"10.1016/j.cageo.2024.105840","DOIUrl":"10.1016/j.cageo.2024.105840","url":null,"abstract":"<div><div>The direct current (DC) resistivity method is an effective method for detecting subsurface structures with conductive differences. The three-dimensional forward modeling of DC resistivity is essential for data processing and interpretation. In the forward modeling process, solving the large-scale linear system is the most time-consuming step. Traditional algebraic multigrid (AMG) method has been successfully applied to solve the linear system. However, the performance of AMG will deteriorate when dealing with high-order discretization and highly stretched grids. Additionally, in the context of adaptive mesh refinement, the hanging nodes will further complicate the solving process. To address these challenges, we propose a novel geometric multigrid (GMG) method combined with local smoothing for solving the linear system in three-dimensional DC resistivity forward modeling. In this work, we employ high-order basis functions to discretize the problem. To further enhance the accuracy of the numerical solution, the mesh is adaptively refined based on the goal-oriented posterior error estimator. We utilize a V-cycle geometric multigrid on locally refined grids and the hanging node issue is effectively addressed through local smoothing. We also employ the mesh partitioning technique to parallel the solution process. The efficiency, robustness, and parallel performance of our algorithm are verified through various numerical examples.</div></div>","PeriodicalId":55221,"journal":{"name":"Computers & Geosciences","volume":"196 ","pages":"Article 105840"},"PeriodicalIF":4.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143093408","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.cageo.2024.105838
L. Castrogiovanni, P. Sternai, N. Piana Agostinetti, C. Pasquero
Future goals and strategies for mitigating ongoing climate changes rely on the understanding of the global carbon cycle and its connections to climate. Evidence from ice cores regarding past atmospheric CO2 and temperature changes through glacial-interglacial oscillations provide crucial insight into the natural variability of carbon cycling. However, poor constraints on atmospheric CO2 input and output fluxes limit our quantitative understanding of late Pleistocene carbon cycling and climate changes. In this study, we describe an inversion method based on a reversible-jump Markov chain Monte Carlo (rj-McMC) algorithm and a general formulation of the geological carbon cycle to estimate paleo-fluxes of CO2. We present results from two synthetic tests and a real case study based on data from the ice core of Dome Fuji. Results from synthetic tests demonstrate the capability of the algorithm to retrieve reliable estimates of atmospheric CO2 input and output fluxes inverting the time derivative of the atmospheric CO2 record and using its temperature time series as a further constraint. Results from the Dome Fuji case study underscore systematic pulses of input CO2 fluxes into the atmosphere during deglaciations predating peaks of T and output CO2 fluxes by ∼2.5 kyrs. The retrieved surface source and sink CO2 fluxes as well as future applications of the algorithm presented here will provide new insights to assess past climate driving mechanisms and inform projections of future climatic trajectories.
{"title":"A reversible-jump Markov chain Monte Carlo algorithm to estimate paleo surface CO2 fluxes linking temperature to atmospheric CO2 concentration time series","authors":"L. Castrogiovanni, P. Sternai, N. Piana Agostinetti, C. Pasquero","doi":"10.1016/j.cageo.2024.105838","DOIUrl":"10.1016/j.cageo.2024.105838","url":null,"abstract":"<div><div>Future goals and strategies for mitigating ongoing climate changes rely on the understanding of the global carbon cycle and its connections to climate. Evidence from ice cores regarding past atmospheric CO<sub>2</sub> and temperature changes through glacial-interglacial oscillations provide crucial insight into the natural variability of carbon cycling. However, poor constraints on atmospheric CO<sub>2</sub> input and output fluxes limit our quantitative understanding of late Pleistocene carbon cycling and climate changes. In this study, we describe an inversion method based on a reversible-jump Markov chain Monte Carlo (rj-McMC) algorithm and a general formulation of the geological carbon cycle to estimate paleo-fluxes of CO<sub>2</sub>. We present results from two synthetic tests and a real case study based on data from the ice core of Dome Fuji. Results from synthetic tests demonstrate the capability of the algorithm to retrieve reliable estimates of atmospheric CO<sub>2</sub> input and output fluxes inverting the time derivative of the atmospheric CO<sub>2</sub> record and using its temperature time series as a further constraint. Results from the Dome Fuji case study underscore systematic pulses of input CO<sub>2</sub> fluxes into the atmosphere during deglaciations predating peaks of T and output CO<sub>2</sub> fluxes by ∼2.5 kyrs. The retrieved surface source and sink CO<sub>2</sub> fluxes as well as future applications of the algorithm presented here will provide new insights to assess past climate driving mechanisms and inform projections of future climatic trajectories.</div></div>","PeriodicalId":55221,"journal":{"name":"Computers & Geosciences","volume":"196 ","pages":"Article 105838"},"PeriodicalIF":4.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143093412","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.cageo.2024.105830
Zujian Yang , Xiao Tian , Xiangteng Wang , Yue Wang , Xiong Zhang
The discrimination of natural and unnatural seismic events is an important part of earthquake monitoring and early warning. Deep learning algorithms, with their powerful feature extraction and classification capabilities, are extensively applied in seismic event identification. In this study, we utilized the DiTing 2.0 dataset to develop binary-class networks for distinguishing low-magnitude earthquakes from explosions, as well as three-class networks for identifying low-magnitude earthquakes, explosions, and collapses. The accuracies achieved for discriminating earthquakes from explosions using waveform and spectrogram datasets are 94% and 87%, respectively. The accuracies for discriminating earthquakes, explosions, and collapses using waveform and spectrogram datasets are 85% and 83%, respectively. We then apply the trained three-class model to discriminate explosions and collapses in four different regions in China. The prediction results indicate that the trained model can accurately identify event types and exhibits a good performance in low-magnitude seismic event ( <5) discrimination, demonstrating the effectiveness and generality of the models developed in this study.
{"title":"Discrimination of earthquakes, explosions, and collapses based on the deep learning: Applications to DiTing 2.0 dataset","authors":"Zujian Yang , Xiao Tian , Xiangteng Wang , Yue Wang , Xiong Zhang","doi":"10.1016/j.cageo.2024.105830","DOIUrl":"10.1016/j.cageo.2024.105830","url":null,"abstract":"<div><div>The discrimination of natural and unnatural seismic events is an important part of earthquake monitoring and early warning. Deep learning algorithms, with their powerful feature extraction and classification capabilities, are extensively applied in seismic event identification. In this study, we utilized the DiTing 2.0 dataset to develop binary-class networks for distinguishing low-magnitude earthquakes from explosions, as well as three-class networks for identifying low-magnitude earthquakes, explosions, and collapses. The accuracies achieved for discriminating earthquakes from explosions using waveform and spectrogram datasets are 94% and 87%, respectively. The accuracies for discriminating earthquakes, explosions, and collapses using waveform and spectrogram datasets are 85% and 83%, respectively. We then apply the trained three-class model to discriminate explosions and collapses in four different regions in China. The prediction results indicate that the trained model can accurately identify event types and exhibits a good performance in low-magnitude seismic event (<span><math><mrow><msub><mi>M</mi><mi>L</mi></msub></mrow></math></span> <5) discrimination, demonstrating the effectiveness and generality of the models developed in this study.</div></div>","PeriodicalId":55221,"journal":{"name":"Computers & Geosciences","volume":"196 ","pages":"Article 105830"},"PeriodicalIF":4.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143093816","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.cageo.2025.105868
Huan Cao , Bin Xu , Congyu Wang , Jun Hu , Quanfeng Wang , Jun Feng
Seismic event detection is a basic and crucial task in seismic data processing. With the gradual increase in seismic observation data, how to detect seismic events from seismic records automatically and accurately has become an urgent problem. However, due to the complexity and variability of the seismic observation environment, acquired seismic records are always accompanied by various noises, compromising detection accuracy. Considering the different noises contained in seismic records acquired by different seismic sensors, herein, a deep residual shrinkage network (DRSN) was constructed to detect seismic events in low signal-to-noise ratio (SNR) seismic records. To test the performance of our model, two types of experiments were conducted. Results demonstrated that the DRSN uses a soft thresholding function to eliminate noise interference while retaining effective signal features; it also introduces an attention mechanism to enhance the focus on significant features and adaptively adjusts the denoising threshold. Consequently, the DRSN effectively eliminates the effect of different noises on seismic event recognition according to the characteristics of different signals, thereby resulting in good overall performance. In detecting the Stanford earthquake dataset and microseismic signals, the DRSN achieved accuracies of 99.08% and 95.43%, respectively, outperforming the short-term average over long-term average, convolutional neural network, earthquake transformer, and sequential attention network. The DRSN can be applied to the automatic and accurate detection of seismic events, especially under low SNR conditions, such as for microseismic signals. Moreover, the DRSN requires no manual setting of the optimal denoising threshold, making the model operable and universal.
{"title":"Automatic seismic event detection in low signal-to-noise ratio seismic signal based on a deep residual shrinkage network","authors":"Huan Cao , Bin Xu , Congyu Wang , Jun Hu , Quanfeng Wang , Jun Feng","doi":"10.1016/j.cageo.2025.105868","DOIUrl":"10.1016/j.cageo.2025.105868","url":null,"abstract":"<div><div>Seismic event detection is a basic and crucial task in seismic data processing. With the gradual increase in seismic observation data, how to detect seismic events from seismic records automatically and accurately has become an urgent problem. However, due to the complexity and variability of the seismic observation environment, acquired seismic records are always accompanied by various noises, compromising detection accuracy. Considering the different noises contained in seismic records acquired by different seismic sensors, herein, a deep residual shrinkage network (DRSN) was constructed to detect seismic events in low signal-to-noise ratio (SNR) seismic records. To test the performance of our model, two types of experiments were conducted. Results demonstrated that the DRSN uses a soft thresholding function to eliminate noise interference while retaining effective signal features; it also introduces an attention mechanism to enhance the focus on significant features and adaptively adjusts the denoising threshold. Consequently, the DRSN effectively eliminates the effect of different noises on seismic event recognition according to the characteristics of different signals, thereby resulting in good overall performance. In detecting the Stanford earthquake dataset and microseismic signals, the DRSN achieved accuracies of 99.08% and 95.43%, respectively, outperforming the short-term average over long-term average, convolutional neural network, earthquake transformer, and sequential attention network. The DRSN can be applied to the automatic and accurate detection of seismic events, especially under low SNR conditions, such as for microseismic signals. Moreover, the DRSN requires no manual setting of the optimal denoising threshold, making the model operable and universal.</div></div>","PeriodicalId":55221,"journal":{"name":"Computers & Geosciences","volume":"196 ","pages":"Article 105868"},"PeriodicalIF":4.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143093823","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.cageo.2024.105842
Fuxiang Liu , Shengqing Xiong , Hai Yang
Copper-nickel (Cu-Ni) deposits are short-supply strategic mineral resources. Ultrabasic rocks are the important ore-forming geological bodies of Cu-Ni deposits. The accurate identification of ultrabasic rocks is important for predicting Cu-Ni deposits. With the high-level prospecting, the outcropped ultrabasic rocks are mostly discovered, while the concealed ones are difficult to detect by geological mapping. Geophysical data contains deep information that is important in concealed ultrabasic rock prospecting. Based on the physical properties (high magnetic susceptibility and low radioactivity), we use large-range, high-precision airborne magnetic and radioactive data to predict undiscovered ultrabasic rocks. With the development of artificial intelligence technology, support vector machines (SVM) have the advantage of strong classification ability and are suitable for small samples. Because of the influence of several parameters, it is difficult to get a prediction result with high geological interpretability. Therefore, genetic algorithms (GA) with global optimization are used in this study to obtain better SVM parameters. The geological factors are first proposed to describe ultrabasic rock characteristics and added into the fitness functions of GA to improve accuracy and geological interpretability. The geological factors consist of shape (Fshape), area (Farea), and scatter (Fsactter). Consequently, support vector machines optimized by genetic algorithms (GASVM) are applied to predict ultrabasic rocks in the Qilian orogen. The airborne magnetic and radioactive data consider different geophysical properties and reveal geological characteristics of ultrabasic rocks from deep to shallow. The results are better than the prediction by a single dataset. In addition, we test three standardization methods, the rising ridge distribution standardization method could effectively and efficiently improve the prediction accuracy. The perdition accuracies are higher than 85% compared with the known ultrabasic rocks. The prediction ranges are concentrated in less than 10% of the whole area. Two ultrabasic rocks are first predicted in this study which is confirmed by field investigation. Meanwhile, concealed ultrabasic rocks are predicted in the southeast of the study area, which are not shown on the geological map. The results can provide general and reliable distributions of ultrabasic rock and rapidly delineate the favorite area for Cu-Ni deposit prospecting.
{"title":"Prediction of ultrabasic rocks by support vector machine based on airborne magnetic and radioactivity data","authors":"Fuxiang Liu , Shengqing Xiong , Hai Yang","doi":"10.1016/j.cageo.2024.105842","DOIUrl":"10.1016/j.cageo.2024.105842","url":null,"abstract":"<div><div>Copper-nickel (Cu-Ni) deposits are short-supply strategic mineral resources. Ultrabasic rocks are the important ore-forming geological bodies of Cu-Ni deposits. The accurate identification of ultrabasic rocks is important for predicting Cu-Ni deposits. With the high-level prospecting, the outcropped ultrabasic rocks are mostly discovered, while the concealed ones are difficult to detect by geological mapping. Geophysical data contains deep information that is important in concealed ultrabasic rock prospecting. Based on the physical properties (high magnetic susceptibility and low radioactivity), we use large-range, high-precision airborne magnetic and radioactive data to predict undiscovered ultrabasic rocks. With the development of artificial intelligence technology, support vector machines (SVM) have the advantage of strong classification ability and are suitable for small samples. Because of the influence of several parameters, it is difficult to get a prediction result with high geological interpretability. Therefore, genetic algorithms (GA) with global optimization are used in this study to obtain better SVM parameters. The geological factors are first proposed to describe ultrabasic rock characteristics and added into the fitness functions of GA to improve accuracy and geological interpretability. The geological factors consist of shape (Fshape), area (Farea), and scatter (Fsactter). Consequently, support vector machines optimized by genetic algorithms (GASVM) are applied to predict ultrabasic rocks in the Qilian orogen. The airborne magnetic and radioactive data consider different geophysical properties and reveal geological characteristics of ultrabasic rocks from deep to shallow. The results are better than the prediction by a single dataset. In addition, we test three standardization methods, the rising ridge distribution standardization method could effectively and efficiently improve the prediction accuracy. The perdition accuracies are higher than 85% compared with the known ultrabasic rocks. The prediction ranges are concentrated in less than 10% of the whole area. Two ultrabasic rocks are first predicted in this study which is confirmed by field investigation. Meanwhile, concealed ultrabasic rocks are predicted in the southeast of the study area, which are not shown on the geological map. The results can provide general and reliable distributions of ultrabasic rock and rapidly delineate the favorite area for Cu-Ni deposit prospecting.</div></div>","PeriodicalId":55221,"journal":{"name":"Computers & Geosciences","volume":"196 ","pages":"Article 105842"},"PeriodicalIF":4.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143093104","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.cageo.2024.105820
Rishav Mallick , Brendan J. Meade
Fault slip during the earthquake cycle is often spatially heterogeneous and occurs on non-planar fault surfaces. In this study, we present an analytical method for calculating displacements and stresses resulting from spatially variable fault slip on faults with arbitrary geometry in a linear elastic medium. This method enforces that fault slip is spatially continuous and differentiable, in contrast to classical constant-slip Green’s function boundary element models which suffer from stress singularities at element boundaries. By eliminating these stress singularities, our approach improves mechanical interpretability and accuracy in strain energy calculations. We demonstrate the construction and application of continuous slip boundary element models in two dimensions for the Himalayan Range Front (HRF) faults in Nepal, and show that strain energy accumulates at Pa-m/m of convergence for the greater HRF region and grows quadratically with convergence amount. The strain energy released by the 2015 Gorkha earthquake was Pa-m, equivalent to the complete release of strain energy from only m of convergence as compared to nearly uniform 6–7 m of slip estimated geodetically. The discrepancy between coseismic slip and the equivalent convergence represents a roughly 30% increase in the total strain energy in the volume over the considered interval of the earthquake cycle.
{"title":"Smooth slip is all you need: A singularity-free boundary element method for fault slip problems","authors":"Rishav Mallick , Brendan J. Meade","doi":"10.1016/j.cageo.2024.105820","DOIUrl":"10.1016/j.cageo.2024.105820","url":null,"abstract":"<div><div>Fault slip during the earthquake cycle is often spatially heterogeneous and occurs on non-planar fault surfaces. In this study, we present an analytical method for calculating displacements and stresses resulting from spatially variable fault slip on faults with arbitrary geometry in a linear elastic medium. This method enforces that fault slip is spatially continuous and differentiable, in contrast to classical constant-slip Green’s function boundary element models which suffer from stress singularities at element boundaries. By eliminating these stress singularities, our approach improves mechanical interpretability and accuracy in strain energy calculations. We demonstrate the construction and application of continuous slip boundary element models in two dimensions for the Himalayan Range Front (HRF) faults in Nepal, and show that strain energy accumulates at <span><math><mrow><mn>5</mn><mo>.</mo><mn>6</mn><mo>×</mo><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mn>13</mn></mrow></msup></mrow></math></span> Pa-m/m of convergence for the greater HRF region and grows quadratically with convergence amount. The strain energy released by the 2015 <span><math><mrow><msub><mrow><mi>M</mi></mrow><mrow><mi>W</mi></mrow></msub><mo>=</mo><mn>7</mn><mo>.</mo><mn>8</mn></mrow></math></span> Gorkha earthquake was <span><math><mrow><mo>∼</mo><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mn>15</mn></mrow></msup></mrow></math></span> Pa-m, equivalent to the complete release of strain energy from only <span><math><mrow><mo>∼</mo><mn>4</mn></mrow></math></span> m of convergence as compared to nearly uniform 6–7 m of slip estimated geodetically. The discrepancy between coseismic slip and the equivalent convergence represents a roughly 30% increase in the total strain energy in the volume over the considered interval of the earthquake cycle.</div></div>","PeriodicalId":55221,"journal":{"name":"Computers & Geosciences","volume":"196 ","pages":"Article 105820"},"PeriodicalIF":4.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143093109","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.cageo.2024.105832
Hongzhu Cai , Siyuan He , Ziang He , Shuang Liu , Lichao Liu , Xiangyun Hu
Recovering basement relief from gravity data plays a crucial role in understanding regional tectonics and advancing resource exploration. Traditional inversion methods typically assume a known density contrast between sedimentary layers and basement rocks to simplify the inverse problem, despite the reality that this contrast varies significantly. To overcome this limitation, we propose a deep learning approach to estimate basement relief from gravity data without requiring a fixed density contrast. We develop two distinct model generation methods to prepare the dataset and validate our neural network through comprehensive synthetic studies. Utilizing a CNN-LSTM architecture, which performs robustly across all tests, we apply this method to both synthetic and field case studies. The results demonstrate that our approach accurately estimates basement relief under variable density contrasts. Furthermore, our testing framework identifies the most effective network architectures and model generation strategies for tackling complex, multi-source geophysical problems.
{"title":"Effective gravity inversion of basement relief with unfixed density contrast using deep learning","authors":"Hongzhu Cai , Siyuan He , Ziang He , Shuang Liu , Lichao Liu , Xiangyun Hu","doi":"10.1016/j.cageo.2024.105832","DOIUrl":"10.1016/j.cageo.2024.105832","url":null,"abstract":"<div><div>Recovering basement relief from gravity data plays a crucial role in understanding regional tectonics and advancing resource exploration. Traditional inversion methods typically assume a known density contrast between sedimentary layers and basement rocks to simplify the inverse problem, despite the reality that this contrast varies significantly. To overcome this limitation, we propose a deep learning approach to estimate basement relief from gravity data without requiring a fixed density contrast. We develop two distinct model generation methods to prepare the dataset and validate our neural network through comprehensive synthetic studies. Utilizing a CNN-LSTM architecture, which performs robustly across all tests, we apply this method to both synthetic and field case studies. The results demonstrate that our approach accurately estimates basement relief under variable density contrasts. Furthermore, our testing framework identifies the most effective network architectures and model generation strategies for tackling complex, multi-source geophysical problems.</div></div>","PeriodicalId":55221,"journal":{"name":"Computers & Geosciences","volume":"196 ","pages":"Article 105832"},"PeriodicalIF":4.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143093179","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.cageo.2024.105831
Qiang Liu , Changli Yao , Guangjing Xu , Yao Luo , Xianzhe Yin
Normalized source strength (NSS) is often applied to interpret magnetic anomalies due to its low sensitivity to magnetization direction. However, when calculating NSS, it is usually necessary to calculate the magnetic potential based on the direction of the geomagnetic field. Similar to the reduction-to-the-pole method routinely computed in the wavenumber domain, NSS is unstable at low latitudes. Therefore, we proposed a new method called the low-latitude normalized source strength (LLNSS); with this approach, the NSS is calculated using the magnetic anomaly instead of the magnetic potential. This approach expands the range of application of the NSS method. The proposed method does not depend on the direction of the geomagnetic field, making it suitable for processing and interpreting magnetic data in the presence of strong residual magnetization, particularly in low-latitude areas. This method was tested on both synthetic and field datasets. Comparative model test results showed that our algorithm had better calculation stability, lower magnetization direction sensitivity, and stronger field-source positioning ability. Real data processing results further validated the effectiveness and practicality of our method.
{"title":"Estimation method for field source location in the presence of strong remanent magnetization in low-latitude regions","authors":"Qiang Liu , Changli Yao , Guangjing Xu , Yao Luo , Xianzhe Yin","doi":"10.1016/j.cageo.2024.105831","DOIUrl":"10.1016/j.cageo.2024.105831","url":null,"abstract":"<div><div>Normalized source strength (NSS) is often applied to interpret magnetic anomalies due to its low sensitivity to magnetization direction. However, when calculating NSS, it is usually necessary to calculate the magnetic potential based on the direction of the geomagnetic field. Similar to the reduction-to-the-pole method routinely computed in the wavenumber domain, NSS is unstable at low latitudes. Therefore, we proposed a new method called the low-latitude normalized source strength (LLNSS); with this approach, the NSS is calculated using the magnetic anomaly instead of the magnetic potential. This approach expands the range of application of the NSS method. The proposed method does not depend on the direction of the geomagnetic field, making it suitable for processing and interpreting magnetic data in the presence of strong residual magnetization, particularly in low-latitude areas. This method was tested on both synthetic and field datasets. Comparative model test results showed that our algorithm had better calculation stability, lower magnetization direction sensitivity, and stronger field-source positioning ability. Real data processing results further validated the effectiveness and practicality of our method.</div></div>","PeriodicalId":55221,"journal":{"name":"Computers & Geosciences","volume":"196 ","pages":"Article 105831"},"PeriodicalIF":4.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143093820","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.cageo.2025.105851
Dan Song , Yu Wang , Wenhui Li , Wen Liu , Zhiqiang Wei , An-An Liu
Accurate precipitation nowcasting holds great significance for daily life. In recent years, deep learning networks have demonstrated excellent performance in the field of precipitation nowcasting. However, they did not fully harness important prior information such as the experience acquired from pre-trained model and the effects caused by terrain. In this paper, we propose a prior information assisted multi-scale network for precipitation nowcasting. Firstly, we employ a cross-attention mechanism to model the correlation between terrain elevation and radar echoes, enhancing the feature representation of the input. Subsequently, we introduce a teacher–student network, leveraging the pre-trained model’s capability in modeling echo movement as prior information to assist in the prediction. Finally, a multi-scale UNet network is proposed to cross-fuse large-scale and small-scale features so that the predicted images retain global information and more local details. We conduct precipitation nowcasting tests using real radar echo datasets within the 0–2 h range. Compared with the second best results(i.e., REMNet (Jing et al., 2022) for Probability of Detection (POD) and RainNet (Ayzel et al., 2020) for Critical Success Index (CSI)), our method improves the POD and CSI by 15.4% and 27.7%, respectively, demonstrating the superiority of our method.
{"title":"Prior information assisted multi-scale network for precipitation nowcasting","authors":"Dan Song , Yu Wang , Wenhui Li , Wen Liu , Zhiqiang Wei , An-An Liu","doi":"10.1016/j.cageo.2025.105851","DOIUrl":"10.1016/j.cageo.2025.105851","url":null,"abstract":"<div><div>Accurate precipitation nowcasting holds great significance for daily life. In recent years, deep learning networks have demonstrated excellent performance in the field of precipitation nowcasting. However, they did not fully harness important prior information such as the experience acquired from pre-trained model and the effects caused by terrain. In this paper, we propose a prior information assisted multi-scale network for precipitation nowcasting. Firstly, we employ a cross-attention mechanism to model the correlation between terrain elevation and radar echoes, enhancing the feature representation of the input. Subsequently, we introduce a teacher–student network, leveraging the pre-trained model’s capability in modeling echo movement as prior information to assist in the prediction. Finally, a multi-scale UNet network is proposed to cross-fuse large-scale and small-scale features so that the predicted images retain global information and more local details. We conduct precipitation nowcasting tests using real radar echo datasets within the 0–2 h range. Compared with the second best results(i.e., REMNet (Jing et al., 2022) for Probability of Detection (POD) and RainNet (Ayzel et al., 2020) for Critical Success Index (CSI)), our method improves the POD and CSI by 15.4% and 27.7%, respectively, demonstrating the superiority of our method.</div></div>","PeriodicalId":55221,"journal":{"name":"Computers & Geosciences","volume":"196 ","pages":"Article 105851"},"PeriodicalIF":4.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143093824","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号