Abstract

Abstract An image processing technique called the cellular neural network (CNN) approach is used in this study to locate geological features giving rise to gravity anomalies such as faults or the boundary of two geologic zones. CNN is a stochastic image processing technique based on template optimization using the neighborhood relationships of cells. These cells can be characterized by a functional block diagram that is typical of neural network theory. The functionality of CNN is described in its entirety by a number of small matrices (A, B and I) called the cloning template. CNN can also be considered to be a nonlinear convolution of these matrices. This template describes the strength of the nearest neighbor interconnections in the network. The recurrent perceptron learning algorithm (RPLA) is used in optimization of cloning template. The CNN and standard Canny algorithms were first tested on two sets of synthetic gravity data with the aim of checking the reliability of the proposed approach. The CNN method was compared with classical derivative techniques by applying the cross-correlation method (CC) to the same anomaly map as this latter approach can detect some features that are difficult to identify on the Bouguer anomaly maps. This approach was then applied to the Bouguer anomaly map of Biga and its surrounding area, in Turkey. Structural features in the area between Bandirma, Biga, Yenice and Gonen in the southwest Marmara region are investigated by applying the CNN and CC to the Bouguer anomaly map. Faults identified by these algorithms are generally in accordance with previously mapped surface faults. These examples show that the geologic boundaries can be detected from Bouguer anomaly maps using the cloning template approach. A visual evaluation of the outputs of the CNN and CC approaches is carried out, and the results are compared with each other. This approach provides quantitative solutions based on just a few assumptions, which makes the method more powerful than the classical methods.

Highlights

  • One of the most common problems encountered in geophysical studies is how to determine the subsurface geological features, namely faults or the boundary of two geologic zones at various depths, as it difficult to establish the geological boundaries hidden beneath surface materials

  • cellular neural network (CNN) is a stochastic image processing technique based on template optimization using the neighborhood relationships of cells

  • The CNN method was compared with classical derivative techniques by applying the cross-correlation method (CC) to the same anomaly map as this latter approach can detect some features that are difficult to identify on the Bouguer anomaly maps

Read more

Summary

Introduction

One of the most common problems encountered in geophysical studies is how to determine the subsurface geological features, namely faults or the boundary of two geologic zones at various depths, as it difficult to establish the geological boundaries hidden beneath surface materials. Gravity is an effective geophysical method by which to identify these, and numerous techniques exist in the field of geophysics for analyzing gravity anomalies produced by faults or the boundary of two geologic zones. In this context, boundary analyses of gravity or magnetic anomalies are carried out using various combinations of directional derivative (gradient) techniques. Geologists and geophysicists are interested in linear anomalies in gravity and magnetic maps that may correspond to subsurface faults, a boundary of two geologic zones and other structural features. Several edge detection algorithms can be found in earlier publications, all of which are based on image gradi-

AYDOGAN
Geological Features of the Southern Marmara Region
Discussion and Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call