SEISMIC FIRST ARRIVAL TRAVELTIME INVERSION HARNESSING PHYSICS INFORMED NEURAL NETWORKS

Download

MSc_thesis.pdf

Date

2022-2-11

Author

Yıldırım, İsa Eren

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

384
views

388
downloads

In Seismic prospecting, huge amounts of data are collected and processed to infer the structural and lithological composition of the subsurface. The key step in this procedure is velocity model building. First arrival traveltime inversion is one of the velocity model building tools commonly used for predicting near-surface velocity structures in seismic exploration. Conventionally, the inversion is carried out using ray-based methods or gradient-based algorithms. Though the gradient-based algorithms find the gradient that is needed to update the model parameters without requiring ray tracing, it can be computationally demanding. On the other hand, despite its robustness and efficiency ray-based methods suffer from complex regions as the ray theory relies on the high-frequency approximation. Instead of using these approaches for a traveltime inversion problem, we propose a machine learning based approach, specifically harnessing the physics informed neural networks exploiting the mathematical model represented by the eikonal equation to estimate the near-surface subsurface velocities. Through synthetic tests and the application of real data, we show the reliability of the physics informed machine learning based traveltime inversion which can be a potential alternative tool to the traditional tomography frameworks.

Subject Keywords

inverse problems, machine learning, physics informed neural networks

URI

https://hdl.handle.net/11511/96711

Collections

Graduate School of Applied Mathematics, Thesis

Suggestions

OpenMETU
Core

Flow dynamics at the origin of thin clayey sand lacustrine turbidites: Examples from Lake Hazar, Turkey Hage, Sophie; Hubert-Ferrari, Aurelia; Lamaır, Laura; Avşar, Ulaş; El Ouahabı, Meriam; Van Daele, Maarten; Boulvaın, Frederic; Bahrı, Mohamed Ali; Seret, Alain; Plenevaux, Alain (2017-12-01) Turbidity currents and their deposits can be investigated using several methods, i.e. direct monitoring, physical and numerical modelling, sediment cores and outcrops. The present study focused on thin clayey sand turbidites found in Lake Hazar (Turkey) occurring in eleven clusters of closely spaced thin beds. Depositional processes and sources for three of those eleven clusters are studied at three coring sites. Bathymetrical data and seismic reflection profiles are used to understand the specific geomorph...
Hourly-scale coastal sea level modeling in a changing climate using long short-term memory neural network Ishida, Kei; Tsujimoto, Gozo; Ercan, Ali; Tu, Tongbi; Kiyama, Masato; Amagasaki, Motoki (2020-06-01) In this study, a coastal sea level estimation model was developed at an hourly temporal scale using the long short-termmemory (LSTM) network, which is a type of recurrent neural networks. The model incorporates the effects of various phenomena on the coastal sea level such as the gravitational attractions of the sun and the moon, seasonality, storm surges, and changing climate. The relative positions of the moon and the sun from the target location at each hour were utilized to reflect the gravitational att...
Singularly perturbed diffusion-advection-reaction processes on extremely large three-dimensional curvilinear networks with a periodic microstructure -- efficient solution strategies based on homogenization theory Kropat, Erik; Meyer-Nieberg, Silja; Weber, Gerhard-Wilhelm (American Institute of Mathematical Sciences (AIMS), 2016-8) Boundary value problems on large periodic networks arise in many applications such as soil mechanics in geophysics or the analysis of photonic crystals in nanotechnology. As a model example, singularly perturbed elliptic differential equations of second order are addressed. Typically, the length of periodicity is very small compared to the size of the covered region. The overall complexity of the networks raises serious problems on the computational side. The high density of the graph, the huge number of ed...
Thermally stimulated currents in n-InS single crystals Hasanlı, Nızamı; Yuksek, NS (Elsevier BV, 2003-03-24) Thermally stimulated current measurements are carried out on as-grown n-InS single crystals in the temperature range of 10-125 K. Experimental evidence is found for four trapping centers present in InS. They are located at 20, 35, 60 and 130 meV. The trap parameters have been determined by various methods of analysis, and they agree well with each other.
Coupled physical and biochemical data driven simulations of Black Sea in spring-summer: real-time forecast and data assimilation Besiktepe, ST (2002-12-06) Data driven simulations in the Black Sea based upon observations during May-June 2001 in the SW part of the basin and coupled 3D physical and biochemical models have been carried out. The model was initialised with the data obtained during 22-28 May, 2001 and ran until 15 June, 2001. The data obtained in the second leg during 12-18 June, 2001 was assimilated into the model. At the time of the assimilation, the model forecast and the data were also compared. Quantitative and qualitative comparisons of the co...

Citation Formats

İ. E. Yıldırım, “SEISMIC FIRST ARRIVAL TRAVELTIME INVERSION HARNESSING PHYSICS INFORMED NEURAL NETWORKS,” M.S. - Master of Science, Middle East Technical University, 2022.