Wang Mingwei, Li Yong, Liu Yingtian, Peng Junheng, Li Huating
When dealing with seismic data, diffusion models often face challenges in adequately capturing local features and expressing spatial relationships. This limitation makes it difficult for diffusion models to remove noise from complex structures effectively. To tackle this issue, we propose a novel convolutional attention mechanism Multi-head Self-attention mechanism based on Deformable convolution (DCMSA) achieving efficient fusion of diffusion models with convolutional attention. The implementation of DCMSA is as follows: First, we integrate DCMSA into the UNet architecture to enhance the network's capability in recognizing and processing complex seismic data. Next, the diffusion model utilizes the UNet enhanced with DCMSA to process noisy data. The results indicate that this method addresses the shortcomings of diffusion models in capturing local features and expressing spatial relationships effectively, proving superior to traditional diffusion models and standard neural networks in noise suppression and preserving meaningful seismic data information.
{"title":"DCMSA: Multi-Head Self-Attention Mechanism Based on Deformable Convolution For Seismic Data Denoising","authors":"Wang Mingwei, Li Yong, Liu Yingtian, Peng Junheng, Li Huating","doi":"arxiv-2408.06963","DOIUrl":"https://doi.org/arxiv-2408.06963","url":null,"abstract":"When dealing with seismic data, diffusion models often face challenges in\u0000adequately capturing local features and expressing spatial relationships. This\u0000limitation makes it difficult for diffusion models to remove noise from complex\u0000structures effectively. To tackle this issue, we propose a novel convolutional\u0000attention mechanism Multi-head Self-attention mechanism based on Deformable\u0000convolution (DCMSA) achieving efficient fusion of diffusion models with\u0000convolutional attention. The implementation of DCMSA is as follows: First, we\u0000integrate DCMSA into the UNet architecture to enhance the network's capability\u0000in recognizing and processing complex seismic data. Next, the diffusion model\u0000utilizes the UNet enhanced with DCMSA to process noisy data. The results\u0000indicate that this method addresses the shortcomings of diffusion models in\u0000capturing local features and expressing spatial relationships effectively,\u0000proving superior to traditional diffusion models and standard neural networks\u0000in noise suppression and preserving meaningful seismic data information.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"15 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142211795","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yingtian Liu, Yong Li, Junheng Peng, Huating Li, Mingwei Wang
Thin layers and reservoirs may be concealed in areas of low seismic reflection amplitude, making them difficult to recognize. Deep learning (DL) techniques provide new opportunities for accurate impedance prediction by establishing a nonlinear mapping between seismic data and impedance. However, existing methods primarily use time domain seismic data, which limits the capture of frequency bands, thus leading to insufficient resolution of the inversion results. To address these problems, we introduce a new time-frequency-phase (TFP) mixed-domain closed-loop seismic inversion network (TFP-CSIN) to improve the identification of thin layers and reservoirs. First, the inversion network and closed-loop network are constructed by using bidirectional gated recurrent units (Bi-GRU) and convolutional neural network (CNN) architectures, enabling bidirectional mapping between seismic data and impedance data. Next, to comprehensive learning across the entire frequency spectrum, the Fourier transform is used to capture frequency information and establish frequency domain constraints. At the same time, the phase domain constraint is introduced through Hilbert transformation, which improves the method's ability to recognize the weak reflection region features. Both experiments on the synthetic data show that TFP-CSIN outperforms the traditional supervised learning method and time domain semi-supervised learning methods in seismic inversion. The field data further verify that the proposed method improves the identification ability of weak reflection areas and thin layers.
{"title":"High-resolution closed-loop seismic inversion network in time-frequency phase mixed domain","authors":"Yingtian Liu, Yong Li, Junheng Peng, Huating Li, Mingwei Wang","doi":"arxiv-2408.04932","DOIUrl":"https://doi.org/arxiv-2408.04932","url":null,"abstract":"Thin layers and reservoirs may be concealed in areas of low seismic\u0000reflection amplitude, making them difficult to recognize. Deep learning (DL)\u0000techniques provide new opportunities for accurate impedance prediction by\u0000establishing a nonlinear mapping between seismic data and impedance. However,\u0000existing methods primarily use time domain seismic data, which limits the\u0000capture of frequency bands, thus leading to insufficient resolution of the\u0000inversion results. To address these problems, we introduce a new\u0000time-frequency-phase (TFP) mixed-domain closed-loop seismic inversion network\u0000(TFP-CSIN) to improve the identification of thin layers and reservoirs. First,\u0000the inversion network and closed-loop network are constructed by using\u0000bidirectional gated recurrent units (Bi-GRU) and convolutional neural network\u0000(CNN) architectures, enabling bidirectional mapping between seismic data and\u0000impedance data. Next, to comprehensive learning across the entire frequency\u0000spectrum, the Fourier transform is used to capture frequency information and\u0000establish frequency domain constraints. At the same time, the phase domain\u0000constraint is introduced through Hilbert transformation, which improves the\u0000method's ability to recognize the weak reflection region features. Both\u0000experiments on the synthetic data show that TFP-CSIN outperforms the\u0000traditional supervised learning method and time domain semi-supervised learning\u0000methods in seismic inversion. The field data further verify that the proposed\u0000method improves the identification ability of weak reflection areas and thin\u0000layers.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"4 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141949408","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Uri Malamud, Christoph M. Schafer, Irina Luciana San Sebastian, Maximilian Timpe, Karl Alexander Essink, Christopher Kreuzig, Gerwin Meier, Jürgen Blum, Hagai B. Perets, Christoph Burger
Crush curves are of fundamental importance to numerical modeling of small and porous astrophysical bodies. The empirical literature often measures them for silica grains, and different studies have used various methods, sizes, textures, and pressure conditions. Here we review past studies and supplement further experiments in order to develop a full and overarching understanding of the silica crush curve behavior. We suggest a new power-law function that can be used in impact simulations of analog materials similar to micro-granular silica. We perform a benchmarking study to compare this new crush curve to the parametric quadratic crush curve often used in other studies, based on the study case of the DART impact onto the asteroid Dimorphos. We find that the typical quadratic crush curve parameters do not closely follow the silica crushing experiments, and as a consequence they under (over) estimate compression close (far) from the impact site. The new crush curve presented here, applicable to pressures between a few hundred Pa and up to 1.1 GPa, might therefore be more precise. Additionally, it is not calibrated by case-specific parameters, and can be used universally for comet- or asteroid-like bodies, given an assumed composition similar to micro-granular silica.
{"title":"New versus past silica crush curve experiments: application to Dimorphos benchmarking impact simulations","authors":"Uri Malamud, Christoph M. Schafer, Irina Luciana San Sebastian, Maximilian Timpe, Karl Alexander Essink, Christopher Kreuzig, Gerwin Meier, Jürgen Blum, Hagai B. Perets, Christoph Burger","doi":"arxiv-2408.04014","DOIUrl":"https://doi.org/arxiv-2408.04014","url":null,"abstract":"Crush curves are of fundamental importance to numerical modeling of small and\u0000porous astrophysical bodies. The empirical literature often measures them for\u0000silica grains, and different studies have used various methods, sizes,\u0000textures, and pressure conditions. Here we review past studies and supplement\u0000further experiments in order to develop a full and overarching understanding of\u0000the silica crush curve behavior. We suggest a new power-law function that can\u0000be used in impact simulations of analog materials similar to micro-granular\u0000silica. We perform a benchmarking study to compare this new crush curve to the\u0000parametric quadratic crush curve often used in other studies, based on the\u0000study case of the DART impact onto the asteroid Dimorphos. We find that the\u0000typical quadratic crush curve parameters do not closely follow the silica\u0000crushing experiments, and as a consequence they under (over) estimate\u0000compression close (far) from the impact site. The new crush curve presented\u0000here, applicable to pressures between a few hundred Pa and up to 1.1 GPa, might\u0000therefore be more precise. Additionally, it is not calibrated by case-specific\u0000parameters, and can be used universally for comet- or asteroid-like bodies,\u0000given an assumed composition similar to micro-granular silica.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"30 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141949410","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The total energy of a fireball is commonly obtained from optical measurements with an assumed value for luminous efficiency. Acoustic energy measurements offer an independent means of energy estimation. Here we combine optical and acoustic methods to validate the luminous efficiency model of Boroviv{c}ka et al. (2020). Our goal is to compare these models with acoustic measurements of meteoroid energy deposition. Employing theoretical blast scaling laws following the approach of McFadden et al. (2021), we determine explosive yields for both fireball fragmentation events and cylindrical shocks for four different bright fireballs. We model fireballs using the MetSim software (Vida et al., 2023) and find that the Boroviv{c}ka et al. (2020) model produces agreement better than a factor of two for our three chondritic fireball case studies. The major exception is an iron meteorite-producing fireball where the luminous efficiency is an order of magnitude higher than model predictions calibrated with stony fireballs. We suggest that large disparities between optical and acoustic energies could be a signature of iron fireballs and hence useful as a discriminant of that population.
{"title":"A Comparison of Fireball Luminous Efficiency Models using Acoustic Records","authors":"Luke McFadden, Peter Brown, Denis Vida","doi":"arxiv-2408.04078","DOIUrl":"https://doi.org/arxiv-2408.04078","url":null,"abstract":"The total energy of a fireball is commonly obtained from optical measurements\u0000with an assumed value for luminous efficiency. Acoustic energy measurements\u0000offer an independent means of energy estimation. Here we combine optical and\u0000acoustic methods to validate the luminous efficiency model of Boroviv{c}ka et\u0000al. (2020). Our goal is to compare these models with acoustic measurements of\u0000meteoroid energy deposition. Employing theoretical blast scaling laws following\u0000the approach of McFadden et al. (2021), we determine explosive yields for both\u0000fireball fragmentation events and cylindrical shocks for four different bright\u0000fireballs. We model fireballs using the MetSim software (Vida et al., 2023) and\u0000find that the Boroviv{c}ka et al. (2020) model produces agreement better than\u0000a factor of two for our three chondritic fireball case studies. The major\u0000exception is an iron meteorite-producing fireball where the luminous efficiency\u0000is an order of magnitude higher than model predictions calibrated with stony\u0000fireballs. We suggest that large disparities between optical and acoustic\u0000energies could be a signature of iron fireballs and hence useful as a\u0000discriminant of that population.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"28 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141949409","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In a previous paper, we had shown that because of varying angles of incidence there is a varying degree of convolution down a trace and across a gather, necessitating deconvolution operators varying with time and offset. This idea is examined further in $t$-$x$ as well as $tau$-$p$ domain. We suggest better ways to deconvolve data in $tau$-$p$ domain, taking into account varying degree of convolution in this domain. We derive formulae for periods of surface multiples in $tau$-$p$ domain, e.g., water column peg-legs and reverberations, which have a fixed period depending only on the value of $p$ -- and suggest a way to check/revise the picked velocity using the formulae, provided the multiples are well separated from the primary. Periodicity of two way surface multiples is also studied.
{"title":"Certain aspects of prestack deconvolution","authors":"Jagmeet Singh","doi":"arxiv-2408.03089","DOIUrl":"https://doi.org/arxiv-2408.03089","url":null,"abstract":"In a previous paper, we had shown that because of varying angles of incidence\u0000there is a varying degree of convolution down a trace and across a gather,\u0000necessitating deconvolution operators varying with time and offset. This idea\u0000is examined further in $t$-$x$ as well as $tau$-$p$ domain. We suggest better\u0000ways to deconvolve data in $tau$-$p$ domain, taking into account varying\u0000degree of convolution in this domain. We derive formulae for periods of surface\u0000multiples in $tau$-$p$ domain, e.g., water column peg-legs and reverberations,\u0000which have a fixed period depending only on the value of $p$ -- and suggest a\u0000way to check/revise the picked velocity using the formulae, provided the\u0000multiples are well separated from the primary. Periodicity of two way surface\u0000multiples is also studied.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"48 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141949413","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Using nonlinear simulations in two settings, we demonstrate that QG$^mathrm{+1}$, a potential-vorticity (PV) based next-order-in-Rossby balanced model, captures several aspects of ocean submesoscale physics. In forced-dissipative 3D simulations under baroclinically unstable Eady-type background states, the statistical equilibrium turbulence exhibits long cyclonic tails and a plethora of rapidly-intensifying ageostrophic fronts. Despite that the model requires setting an explicit, small value for the fixed scaling Rossby number, the emergent flows are nevertheless characterized by $O(f)$ vorticity and convergence, as observed in upper-ocean submesoscale flows. Simulations of QG$^mathrm{+1}$ under the classic strain-induced frontogenesis set-up show realistic frontal asymmetry and a finite time blow-up, quantitatively comparable to simulations of the semigeostrophic equations. The inversions in the QG$^mathrm{+1}$ model are straightforward elliptic problems, allowing for the reconstruction of all flow fields from the PV and surface buoyancy, while avoiding the semigeostrophic coordinate transformation. Together, these results suggest QG$^mathrm{+1}$ as a useful tool for studying upper-ocean submesoscale dynamics.
{"title":"Next-order balanced model captures submesoscale physics and statistics","authors":"Ryan Shìjié Dù, K. Shafer Smith, Oliver Bühler","doi":"arxiv-2408.03422","DOIUrl":"https://doi.org/arxiv-2408.03422","url":null,"abstract":"Using nonlinear simulations in two settings, we demonstrate that\u0000QG$^mathrm{+1}$, a potential-vorticity (PV) based next-order-in-Rossby\u0000balanced model, captures several aspects of ocean submesoscale physics. In\u0000forced-dissipative 3D simulations under baroclinically unstable Eady-type\u0000background states, the statistical equilibrium turbulence exhibits long\u0000cyclonic tails and a plethora of rapidly-intensifying ageostrophic fronts.\u0000Despite that the model requires setting an explicit, small value for the fixed\u0000scaling Rossby number, the emergent flows are nevertheless characterized by\u0000$O(f)$ vorticity and convergence, as observed in upper-ocean submesoscale\u0000flows. Simulations of QG$^mathrm{+1}$ under the classic strain-induced\u0000frontogenesis set-up show realistic frontal asymmetry and a finite time\u0000blow-up, quantitatively comparable to simulations of the semigeostrophic\u0000equations. The inversions in the QG$^mathrm{+1}$ model are straightforward\u0000elliptic problems, allowing for the reconstruction of all flow fields from the\u0000PV and surface buoyancy, while avoiding the semigeostrophic coordinate\u0000transformation. Together, these results suggest QG$^mathrm{+1}$ as a useful\u0000tool for studying upper-ocean submesoscale dynamics.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141949411","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Seismic impedance inversion is a widely used technique for reservoir characterization. Accurate, high-resolution seismic impedance data form the foundation for subsequent reservoir interpretation. Deep learning methods have demonstrated significant potential in seismic impedance inversion. Traditional single semi-supervised networks, which directly input original seismic logging data, struggle to capture high-frequency weak signals. This limitation leads to low-resolution inversion results with inadequate accuracy and stability. Moreover, seismic wavelet uncertainty further constrains the application of these methods to real seismic data. To address these challenges, we propose ADDIN-I: an Attention-based Dual-branch Double-Inversion Network for Impedance prediction. ADDIN-I's dual-branch architecture overcomes the limitations of single-branch semi-supervised networks and improves the extraction of high-frequency weak signal features in sequence modeling. The network incorporates an attention mechanism to further enhance its feature extraction capabilities. To adapt the method for real seismic data applications, a deep learning forward operator is employed to fit the wavelet adaptively. ADDIN-I demonstrates excellent performance in both synthetic and real data applications.
{"title":"Acoustic Impedance Prediction Using an Attention-Based Dual-Branch Double-Inversion Network","authors":"Wen Feng, Yong Li, Yingtian Liu, Huating Li","doi":"arxiv-2408.02524","DOIUrl":"https://doi.org/arxiv-2408.02524","url":null,"abstract":"Seismic impedance inversion is a widely used technique for reservoir\u0000characterization. Accurate, high-resolution seismic impedance data form the\u0000foundation for subsequent reservoir interpretation. Deep learning methods have\u0000demonstrated significant potential in seismic impedance inversion. Traditional\u0000single semi-supervised networks, which directly input original seismic logging\u0000data, struggle to capture high-frequency weak signals. This limitation leads to\u0000low-resolution inversion results with inadequate accuracy and stability.\u0000Moreover, seismic wavelet uncertainty further constrains the application of\u0000these methods to real seismic data. To address these challenges, we propose\u0000ADDIN-I: an Attention-based Dual-branch Double-Inversion Network for Impedance\u0000prediction. ADDIN-I's dual-branch architecture overcomes the limitations of\u0000single-branch semi-supervised networks and improves the extraction of\u0000high-frequency weak signal features in sequence modeling. The network\u0000incorporates an attention mechanism to further enhance its feature extraction\u0000capabilities. To adapt the method for real seismic data applications, a deep\u0000learning forward operator is employed to fit the wavelet adaptively. ADDIN-I\u0000demonstrates excellent performance in both synthetic and real data\u0000applications.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"7 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141949414","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Noise suppression in seismic data processing is a crucial research focus for enhancing subsequent imaging and reservoir prediction. Deep learning has shown promise in computer vision and holds significant potential for seismic data processing. However, supervised learning, which relies on clean labels to train network prediction models, faces challenges due to the unavailability of clean labels for seismic exploration data. In contrast, self-supervised learning substitutes traditional supervised learning with surrogate tasks by different auxiliary means, exploiting internal input data information. Inspired by Self2Self with Dropout, this paper presents a self-supervised learning-based noise suppression method called Self-Supervised Deep Convolutional Networks (SSDCN), specifically designed for Common Reflection Point (CRP) gathers. We utilize pairs of Bernoulli-sampled instances of the input noisy image as surrogate tasks to leverage its inherent structure. Furthermore, SSDCN incorporates geological knowledge through the normal moveout correction technique, which capitalizes on the approximately horizontal behavior and strong self-similarity observed in useful signal events within CRP gathers. By exploiting the discrepancy in self-similarity between the useful signals and noise in CRP gathers, SSDCN effectively extracts self-similarity features during training iterations, prioritizing the extraction of useful signals to achieve noise suppression. Experimental results on synthetic and actual CRP gathers demonstrate that SSDCN achieves high-fidelity noise suppression.
地震数据处理中的噪声抑制是促进后续成像和储层预测的关键研究重点。深度学习在计算机视觉领域大有可为，在地震数据处理方面也具有巨大潜力。然而，依赖于干净标签来训练网络预测模型的监督学习面临着挑战，因为地震勘探数据无法获得干净标签。相比之下，自监督学习（self-supervised learning）通过不同的辅助手段，利用内部输入数据信息，以代用任务取代传统的监督学习。受 "自我对自我"（Self2Self with Dropout）的启发，本文提出了一种基于自我监督学习的噪声抑制方法，称为 "自我监督深度卷积网络"（Self-Supervised Deep Convolutional Networks，SSDCN），专门用于共反射点（CRP）采集。我们利用输入噪声图像的一对伯努利采样实例替代任务，以充分利用其固有结构。此外，SSDCN 还通过正常偏移校正技术纳入了地质知识，该技术利用了在 CRP 采集中有用信号事件中观察到的近似水平行为和较强的自相似性。通过利用 CRP 采集中有用信号与噪声之间的自相似性差异，SSDCN 在训练迭代过程中有效地提取了自相似性特征，优先提取有用信号以实现噪声抑制。在合成和实际 CRP 收集上的实验结果表明，SSDCN 实现了高保真噪声抑制。
{"title":"Noise Suppression for CRP Gathers Based on Self2Self with Dropout","authors":"Fei Li, Zhenbin Xia, Dawei Liu, Xiaokai Wang, Wenchao Chen, Juan Chen, Leiming Xu","doi":"arxiv-2408.02187","DOIUrl":"https://doi.org/arxiv-2408.02187","url":null,"abstract":"Noise suppression in seismic data processing is a crucial research focus for\u0000enhancing subsequent imaging and reservoir prediction. Deep learning has shown\u0000promise in computer vision and holds significant potential for seismic data\u0000processing. However, supervised learning, which relies on clean labels to train\u0000network prediction models, faces challenges due to the unavailability of clean\u0000labels for seismic exploration data. In contrast, self-supervised learning\u0000substitutes traditional supervised learning with surrogate tasks by different\u0000auxiliary means, exploiting internal input data information. Inspired by\u0000Self2Self with Dropout, this paper presents a self-supervised learning-based\u0000noise suppression method called Self-Supervised Deep Convolutional Networks\u0000(SSDCN), specifically designed for Common Reflection Point (CRP) gathers. We\u0000utilize pairs of Bernoulli-sampled instances of the input noisy image as\u0000surrogate tasks to leverage its inherent structure. Furthermore, SSDCN\u0000incorporates geological knowledge through the normal moveout correction\u0000technique, which capitalizes on the approximately horizontal behavior and\u0000strong self-similarity observed in useful signal events within CRP gathers. By\u0000exploiting the discrepancy in self-similarity between the useful signals and\u0000noise in CRP gathers, SSDCN effectively extracts self-similarity features\u0000during training iterations, prioritizing the extraction of useful signals to\u0000achieve noise suppression. Experimental results on synthetic and actual CRP\u0000gathers demonstrate that SSDCN achieves high-fidelity noise suppression.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"7 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141949417","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Traditional approaches to mathematically describe spontaneous imbibition are usually based on either macro-scale models, such as Richards equation, or simplified pore-scale models, such as the bundle of capillary tubes (BCTM) or pore-network modeling (PNM). It is well known that such models cannot provide full microscopic details of the multiphase flow processes and that many pore-scale mechanisms still lack proper mathematical descriptions. To improve the predictive capabilities of traditional models, a fundamental understanding of pore-scale dynamics is needed. The focus of this paper is obtaining detailed insight and consistent explanation of particular processes during capillary-controlled water imbibition into dry porous media. We use two-dimensional model geometries and perform fully dynamic volume-of-fluid based direct numerical simulations of air-water multiphase flow at the pore-scale, to study processes that generally are not considered in traditional models. More specifically, we investigate differences between converging and diverging geometries, dynamic pressure and meniscus reconfiguration during pore-filling events, and the influence of inertia and pore size on imbibition dynamics and the occurrence of capillary barriers. Furthermore, we perform a detailed comparison between non-interacting and interacting BCTM and study the impact of the narrow contractions on imbibition dynamics and the trapping of the non-wetting phase. Obtained knowledge can be used to improve predictive models, which are broadly relevant considering the importance of spontaneous imbibition in many different natural and industrial processes.
{"title":"Towards improved understanding of spontaneous imbibition into dry porous media using pore-scale direct numerical simulations","authors":"Luka Malenica, Zhidong Zhang, Ueli Angst","doi":"arxiv-2408.02831","DOIUrl":"https://doi.org/arxiv-2408.02831","url":null,"abstract":"Traditional approaches to mathematically describe spontaneous imbibition are\u0000usually based on either macro-scale models, such as Richards equation, or\u0000simplified pore-scale models, such as the bundle of capillary tubes (BCTM) or\u0000pore-network modeling (PNM). It is well known that such models cannot provide\u0000full microscopic details of the multiphase flow processes and that many\u0000pore-scale mechanisms still lack proper mathematical descriptions. To improve\u0000the predictive capabilities of traditional models, a fundamental understanding\u0000of pore-scale dynamics is needed. The focus of this paper is obtaining detailed\u0000insight and consistent explanation of particular processes during\u0000capillary-controlled water imbibition into dry porous media. We use\u0000two-dimensional model geometries and perform fully dynamic volume-of-fluid\u0000based direct numerical simulations of air-water multiphase flow at the\u0000pore-scale, to study processes that generally are not considered in traditional\u0000models. More specifically, we investigate differences between converging and\u0000diverging geometries, dynamic pressure and meniscus reconfiguration during\u0000pore-filling events, and the influence of inertia and pore size on imbibition\u0000dynamics and the occurrence of capillary barriers. Furthermore, we perform a\u0000detailed comparison between non-interacting and interacting BCTM and study the\u0000impact of the narrow contractions on imbibition dynamics and the trapping of\u0000the non-wetting phase. Obtained knowledge can be used to improve predictive\u0000models, which are broadly relevant considering the importance of spontaneous\u0000imbibition in many different natural and industrial processes.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"11 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141949318","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Joint inversion of geophysical datasets is instrumental in subsurface characterization and has garnered significant popularity, leveraging information from multiple geophysical methods. In this study, we implemented the joint inversion of DC resistivity with MT data using the Multi-Objective Grey Wolf Optimization (MOGWO) algorithm. As an extension of the widely-used Grey Wolf Optimization algorithm, MOGWO offers a suite of pareto optimal non-dominated solutions, eliminating the need for weighting parameters in the objective functions. This set of non-dominated predictions also facilitates the understanding of uncertainty in the predicted model parameters. Through a field case study in the region around Broken Hill in South Central Australia, the paper showcases MOGWO's capabilities in joint inversion, providing confident estimates of the model parameters (resistivity profiles), as indicated by a narrow spread in the suite of solutions. The obtained results are comparable to well established methodologies and highlight the efficacy of MOGWO as a reliable tool in geophysical exploration.
{"title":"Joint Inversion of DC Resistivity and MT Data using Multi-Objective Grey Wolf Optimization","authors":"Rohan Sharma, Divakar Vashisth, Kuldeep Sarkar, Upendra Kumar Singh","doi":"arxiv-2408.02414","DOIUrl":"https://doi.org/arxiv-2408.02414","url":null,"abstract":"Joint inversion of geophysical datasets is instrumental in subsurface\u0000characterization and has garnered significant popularity, leveraging\u0000information from multiple geophysical methods. In this study, we implemented\u0000the joint inversion of DC resistivity with MT data using the Multi-Objective\u0000Grey Wolf Optimization (MOGWO) algorithm. As an extension of the widely-used\u0000Grey Wolf Optimization algorithm, MOGWO offers a suite of pareto optimal\u0000non-dominated solutions, eliminating the need for weighting parameters in the\u0000objective functions. This set of non-dominated predictions also facilitates the\u0000understanding of uncertainty in the predicted model parameters. Through a field\u0000case study in the region around Broken Hill in South Central Australia, the\u0000paper showcases MOGWO's capabilities in joint inversion, providing confident\u0000estimates of the model parameters (resistivity profiles), as indicated by a\u0000narrow spread in the suite of solutions. The obtained results are comparable to\u0000well established methodologies and highlight the efficacy of MOGWO as a\u0000reliable tool in geophysical exploration.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"88 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141949416","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：