TERRAPUB Earth, Planets and Space
Back

Earth Planets Space, Vol. 56 (No. 2), pp. 217-227, 2004

Genetic Algorithm inversion of geomagnetic vector data using a 2.5-dimensional magnetic structure model

Michiko Yamamoto1, and Nobukazu Seama2

1GEOMAR Forschungszentrum fur marine Geowissenschaften, Wischhofstrasse 1-3, D-24148 Kiel, Germany
2Research Center for Inland Seas, Kobe University, 1-1 Rokkodai, Nada, Kobe 657-8501, Japan

(Received June 25, 2003; Revised December 18, 2003; Accepted January 23, 2004)

Abstract: We propose a new inversion method for vector magnetic field data, which uses the Genetic Algorithm in a space domain calculation to determine the best-fitting 2.5-dimensional (2.5-D) structure. This 2.5-D model is composed of magnetic boundaries with arbitrary strike and magnetic intensity. Two numerical formulas combine to express this model. One of them is a two-dimensional magnetic structure expression for a realistically shaped magnetic layer, and the other is a magnetization contrast expression for magnetic boundaries of variable strike. We use a Genetic Algorithm as the computational technique that supports optimum solutions for magnetization, magnetic strike, and boundary location. In practice, calculations are more accurate in the space domain instead of the more conventional frequency domain because it better preserves the short wavelength components and the true geometry between magnetic sources and observation points even for uneven survey track lines. The above leads to high resolution in the inferred magnetization without the need of upward continuation, which is particularly useful for inverting near-bottom survey data. The code is designed to use smaller storage and less computational time. Its application to synthetic data illustrates the power of resolution and precision in interpreting the fine scale processes of mid-ocean ridge accretion.
Key words: Genetic Algorithm, geomagnetic vector data, near-bottom survey, 2.5 D magnetic structure.


Corresponding author E-mail: myamamoto@geomar.de


[Full text] (PDF 1.9 MB)