Computational Neural Networks For Geophysical Data Processing

This book was primarily written for an audience that has heard about neural networks or has had some experience with the algorithms, but would like to gain a deeper understanding of the fundamental material. For those that already have a solid grasp of how to create a neural network application, this work can provide a wide range of examples of nuances in network design, data set design, testing strategy, and error analysis. Computational, rather than artificial, modifiers are used for neural networks in this book to make a distinction between networks that are implemented in hardware and those that are implemented in software. The term artificial neural network covers any implementation that is inorganic and is the most general term. Computational neural networks are only implemented in software but represent the vast majority of applications. While this book cannot provide a blue print for every conceivable geophysics application, it does outline a basic approach that has been used successfully.

This book provides a practical guide to applying soft-computing methods to interpret geophysical data. It discusses the design of neural networks with Matlab for geophysical data, as well as fuzzy logic and neuro-fuzzy concepts and their applications. In addition, it describes genetic algorithms for the automatic and/or intelligent processing and interpretation of geophysical data.

The past fifteen years has witnessed an explosive growth in the fundamental research and applications of artificial neural networks (ANNs) and fuzzy logic (FL). The main impetus behind this growth has been the ability of such methods to offer solutions not amenable to conventional techniques, particularly in application domains involving pattern recognition, prediction and control. Although the origins of ANNs and FL may be traced back to the 1940s and 1960s, respectively, the most rapid progress has only been achieved in the last fifteen years. This has been due to significant theoretical advances in our understanding of ANNs and FL, complemented by major technological developments in high-speed computing. In geophysics, ANNs and FL have enjoyed significant success and are now employed routinely in the following areas (amongst others): 1. Exploration Seismology. (a) Seismic data processing (trace editing; first break picking; deconvolution and multiple suppression; wavelet estimation; velocity analysis; noise identification/reduction; statics analysis; dataset matching/prediction, attenuation), (b) AVO analysis, (c) Chimneys, (d) Compression I dimensionality reduction, (e) Shear-wave analysis, (f) Interpretation (event tracking; lithology prediction and well-log analysis; prospect appraisal; hydrocarbon prediction; inversion; reservoir characterisation; quality assessment; tomography). 2. Earthquake Seismology and Subterranean Nuclear Explosions. 3. Mineral Exploration. 4. Electromagnetic I Potential Field Exploration. (a) Electromagnetic methods, (b) Potential field methods, (c) Ground penetrating radar, (d) Remote sensing, (e) inversion.

Advances in Geophysics, Volume 61 - Machine Learning and Artificial Intelligence in Geosciences, the latest release in this highly-respected publication in the field of geophysics, contains new chapters on a variety of topics, including a historical review on the development of machine learning, machine learning to investigate fault rupture on various scales, a review on machine learning techniques to describe fractured media, signal augmentation to improve the generalization of deep neural networks, deep generator priors for Bayesian seismic inversion, as well as a review on homogenization for seismology, and more. Provides high-level reviews of the latest innovations in geophysics Written by recognized experts in the field Presents an essential publication for researchers in all fields of geophysics

In the middle of the 20th century, Genrich Altshuller, a Russian engineer, analysed hundreds of thousands of patents and scientific publications. From this analysis, he developed TRIZ (G. Altshuller, "40 Principles: TRIZ Keys to Technical Innovation. TRIZ Tools," Volume 1, First Edition, Technical Innovation Center, Inc. , Worcester, MA, January 1998; Y. Salamatov, "TRIZ: The Right Solution at the Right Time. A Guide to Innovative Problem Solving. " Insytec B. V. , 1999), the theory of inventive problem solving, together with a series of practical tools for helping engineers solving technical problems. Among these tools and theories, the substance-field theory gives a structured way of representing problems, the patterns of evolution show the lifecycle of technical systems, the contradiction matrix tells you how to resolve technical contradictions, using the forty principles that describe common ways of improving technical systems. For example, if you want to increase the strength of a device, without adding too much extra weight to it, the contradiction matrix tells you that you can use "Principle 1: Segmentation," or "Principle 8: Counterweight," or "Principle 15: Dynamicity," or "Principle 40: Composite Materials. " I really like two particular ones: "Principle 1: Segmentation," and Principle 15: Dynamicity. " "Segmentation" shows how systems evolve from an initial monolithic form into a set of independent parts, then eventually increasing the number of parts until each part becomes small enough that it cannot be identified anymore.

Applying machine learning to the interpretation of seismic data Seismic data gathered on the surface can be used to generate numerous seismic attributes that enable better understanding of subsurface geological structures and stratigraphic features. With an ever-increasing volume of seismic data available, machine learning augments faster data processing and interpretation of complex subsurface geology. Meta-Attributes and Artificial Networking: A New Tool for Seismic Interpretation explores how artificial neural networks can be used for the automatic interpretation of 2D and 3D seismic data. Volume highlights include: Historic evolution of seismic attributes Overview of meta-attributes and how to design them Workflows for the computation of meta-attributes from seismic data Case studies demonstrating the application of meta-attributes Sets of exercises with solutions provided Sample data sets available for hands-on exercises The American Geophysical Union promotes discovery in Earth and space science for the benefit of humanity. Its publications disseminate scientific knowledge and provide resources for researchers, students, and professionals.

Computational Geo-Electromagnetics: Methods, Models, and Forecasts, Volume Five in the Computational Geophysics series, is devoted to techniques for building of geoelectrical models from electromagnetic data, featuring Bayesian statistical analysis and neural network algorithms. These models are applied to studying the geoelectrical structure of famous volcanoes (i.e., Vesuvio, Kilauea, Elbrus, Komagatake, Hengill) and geothermal zones (i.e., Travale, Italy; Soultz-sous-Forets, Elsace). Methodological recommendations are given on electromagnetic sounding of faults as well as geothermal and hydrocarbon reservoirs. Techniques for forecasting of petrophysical properties from the electrical resistivity as proxy parameter are also considered. Computational Geo-Electromagnetics: Methods, Models, and Forecasts offers techniques and algorithms for building geoelectrical models under conditions of rare or irregularly distributed EM data and/or lack of prior geological and geophysical information. This volume also includes methodological guidelines on interpretation of electromagnetic sounding data depending on goals of the study. Finally, it details computational algorithms for using electrical resistivity for properties beyond boreholes. Provides algorithms for inversion of incomplete, rare or irregularly distributed EM data Features methodological issues of building geoelectrical models Offers techniques for retrieving petrophysical properties from EM sounding data and well logs