Velocity-model building is a key step in hydrocarbon exploration. The main product of velocity-model an initial model the subsurface that subsequently used seismic imaging and interpretation workflows. Reflection or refraction tomography full-waveform inversion (FWI) are most commonly techniques building. On one hand, time-consuming activity relies on successive updates highly human-curated analysis gathers. other FWI very computationally demanding with no guarantees global convergence. We...

Learning to predict multi-label outputs is challenging, but in many problems there a natural metric on the that can be used improve predictions. In this paper we develop loss function for learning, based Wasserstein distance. The distance provides notion of dissimilarity probability measures. Although optimizing with respect exact costly, recent work has described regularized approximation efficiently computed. We describe an efficient learning algorithm regularization, as well novel...

For hydrocarbon exploration, large volumes of data are acquired and used in physical modeling-based workflows to identify geologic features interest such as fault networks, salt bodies, or, general, elements petroleum systems. The adjoint modeling step, which transforms the into model space, subsequent interpretation can be very expensive, both terms computing resources domain-expert time. We propose implement a unique approach that bypasses these demanding steps, directly assisting...

Seismic inversion is a fundamental tool in geophysical analysis, providing window into Earth. In particular, it enables the reconstruction of large-scale subsurface Earth models for hydrocarbon exploration, mining, earthquake shallow hazard assessment, and other tasks.

Summary Initial stages of velocity model building (VMB) start off from smooth models that capture geological assumptions the subsurface region under analysis. Acceptable result successive iterations interpretation and seismic data processing. The interpreters ensure additions/ corrections made by processing are compliant with geophysical knowledge. Seismic adds to structural elements, faults one most relevant those events since they can signal reservoir boundaries or hydrocarbon traps....

Exploration seismic data are heavily manipulated before human interpreters able to extract meaningful information regarding subsurface structures. This manipulation adds modeling and biases is limited by methodological shortcomings. Alternatively, using directly becoming possible thanks deep learning (DL) techniques. A DL-based workflow introduced that uses analog velocity models realistic raw waveforms as input produces output. When insufficient used for training, DL algorithms tend overfit...

Oil and gas companies trust Reverse Time Migration (RTM), the most advanced seismic imaging technique, with crucial decisions on drilling investments. The economic value of oil reserves that require RTM to be localized is in order 10^{13} dollars. But requires vast computational power, which somewhat hindered its practical success. Although, accelerator-based architectures deliver enormous little attention has been devoted assess implementations effort. aim this paper identify major...

Depth imaging projects dedicated to hydrocarbon exploration or field development rely heavily on velocity model building. When salt bodies are present, their accurate delineation is crucial ensure the quality of seismic images, especially for sub-salt targets. We investigate a supervised deep learning (DL) approach which predicts geometry by using and electromagnetic data simultaneously. Different network architectures were designed incorporate these distinct types tested assess best...

Migration techniques are an integral part of seismic imaging workflows. Least-squares reverse time migration (LSRTM) overcomes some the shortcomings conventional algorithms by compensating for illumination and removing sampling artifacts to increase spatial resolution. However, computational cost associated with iterative LSRTM is high convergence can be slow in complex media. We implement prestack a deep-learning framework adopt strategies from data science domain accelerate convergence....

We introduce two algorithms that invert simulated gravity data to 3D subsurface rock/flow properties. The first algorithm is a data-driven, deep learning-based approach, and the second also data-driven but considers temporal evolution of surface events. target application these proposed prediction CO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> plumes as complementary tool for monitoring sequestration deployments. Each outperforms...

Reverse-Time Migration (RTM) is a state-of-the-art technique in seismic acoustic imaging, because of the quality and integrity images it provides. Oil gas companies trust RTM with crucial decisions on multi-million-dollar drilling investments. But requires vastly more computational power than its predecessor techniques, this has somewhat hindered practical success. On other hand, despite multi-core architectures promise to deliver unprecedented power, little attention been devoted mapping...

Summary We explore the feasibility of a deep learning approach for tomography by comparing it with current velocity prediction techniques used in industry. This is accomplished through quantitative and qualitative comparisons models predicted Machine Learning (ML) system those two variations full-waveform inversion (FWI). Additionally, we compare computational aspects approaches. The results show that ML-based reconstructed are competitive to FWI-produced terms selected metrics, widely less...

This paper introduces novel deep recurrent neural network architectures for Velocity Model Building (VMB), which is beyond what Araya-Polo et al 2018 pioneered with the Machine Learning-based seismic tomography built convolutional non-recurrent network. Our investigation includes utilization of basic (RNN) cells, as well Long Short Term Memory (LSTM) and Gated Recurrent Unit (GRU) cells. Performance evaluation reveals that salt bodies are consistently predicted more accurately by GRU...

Summary Current micro-CT image resolution is limited to ∼1-2 microns. A recent study has identified that at least 10 voxels are needed resolve pore throats, which limits the applicability of direct simulations using Digital Rock (DR) technology medium-to-coarse grained rocks (i.e., with permeability &gt; 100 mD). On other hand, 2D high-resolution colored images such as ones obtained from Scanning Electron Microscopy (SEM) deliver a much higher (∼0.5 microns). However, reliable and efficient...

Summary Faults are geological features that essential in considering the development of hydrocarbon reservoirs. Significant resources (appraisal costs and manpower) often deployed to locate them assess their connectivity more accurately so adequate investment decisions can be made. Early a project's life, data might sparse or time dedicated processing limited. An estimate impact faults derived by analogous fault patterns similar reservoir environments. A significant number discoveries...

Quantum and quantum-inspired machine learning has emerged as a promising challenging research field due to the increased popularity of quantum computing, especially with near-term devices. Theoretical contributions point toward generative modeling direction realize first examples real-world advantages from these technologies. A few empirical studies also demonstrate such potential, when considering models based on tensor networks. In this work, we apply tensor-network-based problem molecular...

This article introduces a benchmark application for seismic modeling using finite difference method, which is namedMiniMod, mini modeling. The purpose to provide suite that is, on one hand easy build and adapt the state of art in programming models changing high performance hardware landscape. On other hand, intention have proxy actual production geophysical exploration workloads Oil & Gas exploration, geosciences applications based wave equation. From top bottom, we describe design...

Inverting seismic data to build 3D geological structures is a challenging task due the overwhelming amount of acquired data, and very-high computational load iterative numerical solutions wave equation, as required by industry-standard tools such Full Waveform Inversion (FWI). For example, in an area with surface dimensions 4.5 km × km, hundreds shot-gather cubes are for model reconstruction, leading Terabytes recorded data. This paper presents deep learning solution reconstruction realistic...

Reverse Time Migration (RTM) has become the latest chapter in seismic imaging for geologically complex subsurface areas. In particular proven to be very useful subsaly oil plays of US Gulf Mexico. However, RTM cannot applied extensively due extreme computational demand. The recent availability multi‐core processors, homogeneous and heterogeneous, may provide required compute power. this paper, we benchmark an effective algorithm on several HPC platforms assess viability hardware.

The success of hydrocarbon exploration and production imaging projects relies heavily on the accuracy velocity model. standard methodology for model building is computationally expensive requires extensive human interpretation quality control. Several deep learning (DL) approaches have been proposed to reduce both computational labor costs, but they are all confronted with challenge limited available pertinent training data. In this paper, we propose a hybrid workflow tackle generalization...

The Fast Fourier Transform (FFT) is a widely used numerical algorithm. When N input data points lead to only k << non-zero coefficients in the transformed domain, algorithm clearly inefficient: FFT performs O(NlogN) operations on order calculate or large coefficients, and -- zero negligibly small ones. recently developed sparse (sFFT) provides solution this problem. As are those for FFT, sFFT algorithms complex still computationally challenging. computational difficulties mainly due memory...

We propose a 3D reverse‐time migration (RTM) using an hybrid Finite Difference (FD) pseudospectral algorithm to solve the two‐way acoustic equation. This mainly consists of forward‐backward 2D FFT in lateral dimensions (x‐y plane) and 1D FD depth dimension. allows us get high order accuracy simplifies computation cross derivatives. Therefore our RTM deal with case isotropic media, VTI media (Zhou et al., 2006b) TTI media. The lies on new anisotropic wave equations system (Lesage 2008), which...