Show newer

Physics-Driven Self-Supervised Deep Learning for Free-Surface Multiple Elimination arxiv.org/abs/2502.05189

Physics-Driven Self-Supervised Deep Learning for Free-Surface Multiple Elimination

In recent years, deep learning (DL) has emerged as a promising alternative approach for various seismic processing tasks, including primary estimation (or multiple elimination), a crucial step for accurate subsurface imaging. In geophysics, DL methods are commonly based on supervised learning from large amounts of high-quality labelled data. Instead of relying on traditional supervised learning, in the context of free-surface multiple elimination, we propose a method in which the DL model learns to effectively parameterize the free-surface multiple-free wavefield from the full wavefield by incorporating the underlying physics into the loss computation. This, in turn, yields high-quality estimates without ever being shown any ground truth data. Currently, the network reparameterization is performed independently for each dataset. We demonstrate its effectiveness through tests on both synthetic and field data. We employ industry-standard Surface-Related Multiple Elimination (SRME) using, respectively, global least-squares adaptive subtraction and local least-squares adaptive subtraction as benchmarks. The comparison shows that the proposed method outperforms the benchmarks in estimation accuracy, achieving the most complete primary estimation and the least multiple energy leakage, but at the cost of a higher computational burden.

arXiv.org

Physics-Trained Neural Network as Inverse Problem Solver for Potential Fields: An Example of Downward Continuation between Arbitrary Surfaces arxiv.org/abs/2502.05190

Physics-Trained Neural Network as Inverse Problem Solver for Potential Fields: An Example of Downward Continuation between Arbitrary Surfaces

Downward continuation is a critical task in potential field processing, including gravity and magnetic fields, which aims to transfer data from one observation surface to another that is closer to the source of the field. Its effectiveness directly impacts the success of detecting and highlighting subsurface anomalous sources. We treat downward continuation as an inverse problem that relies on solving a forward problem defined by the formula for upward continuation, and we propose a new physics-trained deep neural network (DNN)-based solution for this task. We hard-code the upward continuation process into the DNN's learning framework, where the DNN itself learns to act as the inverse problem solver and can perform downward continuation without ever being shown any ground truth data. We test the proposed method on both synthetic magnetic data and real-world magnetic data from West Antarctica. The preliminary results demonstrate its effectiveness through comparison with selected benchmarks, opening future avenues for the combined use of DNNs and established geophysical theories to address broader potential field inverse problems, such as density and geometry modelling.

arXiv.org

Descriptor: Five years of meteorological surface data at Oak Ridge Reserve in Tennessee arxiv.org/abs/2502.05191

Descriptor: Five years of meteorological surface data at Oak Ridge Reserve in Tennessee

Access to continuous, quality assessed meteorological data is critical for understanding the climatology and atmospheric dynamics of a region. Research facilities like Oak Ridge National Laboratory (ORNL) rely on such data to assess site-specific climatology, model potential emissions, establish safety baselines, and prepare for emergency scenarios. To meet these needs, on-site towers at ORNL collect meteorological data at 15-minute and hourly intervals. However, data measurements from meteorological towers are affected by sensor sensitivity, degradation, lightning strikes, power fluctuations, glitching, and sensor failures, all of which can affect data quality. To address these challenges, we conducted a comprehensive quality assessment and processing of five years of meteorological data collected from ORNL at 15-minute intervals, including measurements of temperature, pressure, humidity, wind, and solar radiation. The time series of each variable was pre-processed and gap-filled using established meteorological data collection and cleaning techniques, i.e., the time series were subjected to structural standardization, data integrity testing, automated and manual outlier detection, and gap-filling. The data product and highly generalizable processing workflow developed in Python Jupyter notebooks are publicly accessible online. As a key contribution of this study, the evaluated 5-year data will be used to train atmospheric dispersion models that simulate dispersion dynamics across the complex ridge-and-valley topography of the Oak Ridge Reservation in East Tennessee.

arXiv.org

Implementation of Machine Learning Algorithms for Seismic Events Classification arxiv.org/abs/2502.05197

Implementation of Machine Learning Algorithms for Seismic Events Classification

The classification of seismic events has been crucial for monitoring underground nuclear explosions and unnatural seismic events as well as natural earthquakes. This research is an attempt to apply different machine learning (ML) algorithms to classify various types of seismic events into chemical explosions, collapses, nuclear explosions, damaging earthquakes, felt earthquakes, generic earthquakes and generic explosions for a dataset obtained from IRIS-DMC. One major objective of this research has been to identify some of the best ML algorithms for such seismic events classification. The ML algorithms we are implementing in this study include logistic regression, support vector machine (SVM), Naïve Bayes, random forest, K-nearest neighbors (KNN), decision trees, and linear discriminant analysis. Our implementation of the above ML classifier algorithms required to prepare and preprocess the dataset we obtained so that it will be fit for the ML training and testing applications we sought. After the implementation of the ML algorithms, we were able to classify the seismic event types into seven classes in the dataset, and a comparison of each classifier is made to identify the best algorithm for the seismic data classification. Finally, we made predictions of the different event types using the different classifier algorithms, and evaluated each of the various classifier algorithms for seismic prediction using different evaluation metrics. These evaluation metrics helped us to measure the performance of each algorithm. After implementing the seven ML algorithms and a comparison among those various ML algorithms, it has been demonstrated that the best accuracy among these classifiers happened for the Random Forest (RF) algorithm, with an accuracy of 93.5%.

arXiv.org

A finite element-based machine learning model for hydro-mechanical analysis of swelling behavior in clay-sulfate rocks arxiv.org/abs/2502.05198

A finite element-based machine learning model for hydro-mechanical analysis of swelling behavior in clay-sulfate rocks

The hydro-mechanical behavior of clay-sulfate rocks, especially their swelling properties, poses significant challenges in geotechnical engineering. This study presents a hybrid constrained machine learning (ML) model developed using the categorical boosting algorithm (CatBoost) tuned with a Bayesian optimization algorithm to predict and analyze the swelling behavior of these complex geological materials. Initially, a coupled hydro-mechanical model based on the Richards' equation coupled to a deformation process with linear kinematics implemented within the finite element framework OpenGeoSys was used to simulate the observed ground heave in Staufen, Germany, caused by water inflow into the clay-sulfate bearing Triassic Grabfeld Formation. A systematic parametric analysis using Gaussian distributions of key parameters, including Young's modulus, Poisson's ratio, maximum swelling pressure, permeability, and air entry pressure, was performed to construct a synthetic database. The ML model takes time, spatial coordinates, and these parameter values as inputs, while water saturation, porosity, and vertical displacement are outputs. In addition, penalty terms were incorporated into the CatBoost objective function to enforce physically meaningful predictions. Results show that the hybrid approach effectively captures the nonlinear and dynamic interactions that govern hydro-mechanical processes. The study demonstrates the ability of the model to predict the swelling behavior of clay-sulfate rocks, providing a robust tool for risk assessment and management in affected regions. The results highlight the potential of ML-driven models to address complex geotechnical challenges.

arXiv.org

Progress on surface curvature analysis for describing atomization using 2P-LIF images arxiv.org/abs/2502.05207

Progress on surface curvature analysis for describing atomization using 2P-LIF images

To describe atomization completely it is necessary to track the liquid-gas interface morphology at any stage of the atomization process. Typically, instability analysis focuses on generic and simplified morphology: cylindrical jet, liquid sheet, ligament, and droplet to determine their stability and subsequent instability. On the other side sprays composed of spherical droplets are analyzed through their diameter distribution. However, between these situations the liquidgas interface experiences complex morphology that is more and more accessible through numerical simulation and advanced experimental imagery. To take advantage of this new information and to describe synthetically such data new analyses have been proposed. Here, we aim to analyze complex interface morphology with the surface curvature distribution (SCD) [1] but other possibilities exist [2]. The SCD allows us to describe continuously the destabilization of the initial liquid structure, through complex interfaces such as ligaments, blobs, and liquid sheets until the apparition of the first spherical structures which ultimately become droplets. Beyond the description of the interface, it has been possible to show that a careful analysis of the liquid-gas surface through the SCD allows for determining at the early stage of the atomization process the final characteristics of the spray, even its diameter distribution [3]. In the present work, we are using experimental measurements to assess the characteristics of the spray. With these data, it is possible to observe and describe the atomization process at all stages of the atomization using curvature analysis and image processing techniques.

arXiv.org

Construction of a Small-Scale Vacuum Generation System and Using It as an Educational Device to Demonstrate Features of the Vacuum arxiv.org/abs/2502.04340

Derivation and thermodynamiclly consistent coupling of a Debye-H\"uckel type energy to a steric electrolyte model and application to the apparent molar volume and phase boundaries arxiv.org/abs/2502.04344

Derivation and thermodynamiclly consistent coupling of a Debye-Hückel type energy to a steric electrolyte model and application to the apparent molar volume and phase boundaries

We propose to determine the size parameters of a steric electrolyte model from the experimental data of apparent molar volume obtained from mass density measurements. Thereby, we avoid the difficulties associated with modeling the complex structure of the double layer. To this end, we derive a Debye-Hückel-like model of the electric ion-ion interaction for non-constant dielectric susceptibility, which does not depend on any kind of charging process due to its foundation in the general framework of non-equilibrium electro-thermodynamics. The derivation, however, leads to a novel thermodynamic consistency condition for the temperature dependence of the susceptibility. Due to its contributions to the total pressure, the consistent coupling of this new contribution to the free energy requires subtle modifications in the derivation of the simple mixing model for electrolytes. The coupled model is then applied to the apparent molar volume for various related monovalent salts over a wider range of salt concentrations and temperatures, and classical tests of the electrolyte theory at phase boundaries are investigated.

arXiv.org
The Numerics of VMEC++

VMEC++ is a Python-friendly, from-scratch reimplementation in C++ of the Variational Moments Equilibrium Code (VMEC), a fixed- and free-boundary ideal-MHD equilibrium solver for stellarators and tokamaks. The first VMEC implementation was written by Steven P. Hirshman and colleagues in the 1980s and 1990s and its latest Fortran incarnation (PARVMEC, https://github.com/ORNL-Fusion/PARVMEC) is widely used in stellarator optimization systems. Our work improves on previous implementations with regard to various critical aspects: special care has been put in providing an idiomatic Python experience, from installation to actual usage; VMEC++ has a zero-crash policy; it supports inputs in the classic INDATA format as well as friendlier JSON files. VMEC++ execution times are typically less than or equal to previous implementations, and time to convergence can be decreased dramatically by leveraging its hot-restart feature: by providing the output of a VMEC++ run as initial state for a subsequent one, VMEC++ is initialized using the previously converged equilibrium. This can dramatically decrease runtimes when running on many similar magnetic configurations as it typically happens in stellarator optimization pipelines. On the flip side, some features of the original Fortran VMEC implementation are not yet available in VMEC++, such as support for non-stellarator-symmetric configurations. This contribution presents the internal numerics of the open-source VMEC++ package publicly for the first time.

arXiv.org

Spatiotemporal dynamics of nanosecond pulsed discharge in the form of a fast ionization wave: self-consistent two-dimensional modeling and comparison with experiments under negative and positive polarity arxiv.org/abs/2502.04453

Spatiotemporal dynamics of nanosecond pulsed discharge in the form of a fast ionization wave: self-consistent two-dimensional modeling and comparison with experiments under negative and positive polarity

Nanosecond discharges are characterized by a shift in energy branching toward the excitation of electronic levels and dissociation, making them particularly attractive for plasma chemistry. Understanding the spatiotemporal structure of these discharges is especially important. This paper presents a detailed 2D-axisymmetric numerical analysis of a nanosecond discharge propagating in a long tube and in pure nitrogen. The modeling is conducted using a self-consistent plasma fluid solver under the local mean energy approximation (LMEA), including photoionization. The discharge develops at moderate pressures, 1 - 10 Torr, in the form of a fast ionization wave (FIW). Simulations are performed for both negative and positive polarities of the voltage pulse applied to the high-voltage electrode. The computational results are validated against available experimental data, including FIW velocity within the studied pressure range, electron density, longitudinal electric field, and the radial distribution of N2(C) emission on a nanosecond timescale.

arXiv.org

Contrasting the relative performance of RF photonic transversal signal processors based on microcombs using discrete components versus integrated devices arxiv.org/abs/2502.01641

Censor-Aware Semi-Supervised Lung Cancer Survival Time Prediction Using Clinical and Radiomics Feature arxiv.org/abs/2502.01661

Censor-Aware Semi-Supervised Lung Cancer Survival Time Prediction Using Clinical and Radiomics Feature

Purpose: Lung cancer poses a significant global health challenge, necessitating improved prognostic methods for personalized treatment. This study introduces a censor-aware semi-supervised learning (SSL) that integrates clinical and imaging data, addressing biases in traditional models handling censored data. Methods: We analyzed clinical, PET, and CT data from 199 lung cancer patients from TCIA and BC Cancer Agency, focusing on overall survival (OS) time as the primary outcome. Handcrafted (HRF) and Deep Radiomics features (DRF) were extracted after preprocessing using ViSERA software and were combined with clinical features (CF). Feature dimensions were optimized using Principal Component Analysis (PCA), followed by the application of supervised learning (SL) and SSL. SSL incorporated pseudo-labeling of censored data to improve performance. Seven regressors and three hazard ratio survival analysis (HRSA) algorithms were optimized using five-fold cross-validation, grid search and external test bootstrapping. Results: For PET HRFs, SSL reduced the mean absolute error (MAE) by 26.54%, achieving 1.55 years with PCA+decision tree regression, compared to SL's 2.11 years with PCA+KNNR (p<0.05). Combining HRFs (CT_HRF) and DRFs from CT images using SSL+PCA+KNNR achieved an MAE of 2.08.45 years, outperforming SL's 2.26 years by 7.96% (p<0.05). In HRSA, CT_HRF applied to PCA+Component Wise Gradient Boosting Survival Analysis achieved an external c-index of 0.65, effectively differentiating high- and low-risk groups. Conclusion: This study demonstrated that the SSL strategy significantly outperformed SL across PET, CT, and CF. Thereby, censor-aware SSL applied to HRFs from PET images significantly improved survival prediction performance by 26.54% compared to the SL approach.

arXiv.org

Energetically consistent localised APE budgets for local and regional studies of stratified flow energetics arxiv.org/abs/2502.01686

Energetically consistent localised APE budgets for local and regional studies of stratified flow energetics

Because it allows a rigorous separation between reversible and irreversible processes, the concept of available potential energy (APE) has become central to the study of turbulent stratified fluids. In ocean modelling, it is fundamental to the parameterisation of meso-scale ocean eddies and of the turbulent mixing of heat and salt. However, how to apply APE theory consistently to local or regional subdomains has been a longstanding source of confusion due to the globally defined Lorenz reference state entering the definition of APE and of buoyancy forces being generally thought to be meaningless in those cases. In practice, this is often remedied by introducing heuristic `localised' forms of APE density depending uniquely on region-specific reference states, possibly diverging significantly from the global Lorenz reference state. In this paper, we argue that this practice is problematic because it cannot consistently describes the turbulent APE cascades associated with the inter-scale energy transfers between the APE produced at large scales -- which depends on the global Lorenz reference state -- and the APE of smaller scales. To avoid this difficulty, we argue that localised forms of APE density should be defined as the eddy APE component of an exact mean/eddy decomposition of the APE density. The eddy APE density thus defined exhibits a much weaker dependency on the global Lorenz reference state than the mean APE, in agreement with physical intuition, but with a different structure than that of existing heuristic localised APE forms. The results are important, because they establish a rigorous physical basis for linking parameterised energy transfers to observable viscous and diffusive dissipation rates, which is a pivotal goal of energetically consistent ocean models.

arXiv.org
Show older
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.