Predicting Magnetization Direction Using Convolutional Neural Networks
Abstract
Magnetic data have been widely used for understanding the formation, evolution and structure of various geological features including but not limited to seamounts, oceanic plateaus, mineral deposit systems and many other igneous intrusions. Proper interpretation of magnetic data requires an accurate knowledge of total magnetization directions of the source bodies in an area of study. In this study, we investigated the use of Convolutional Neural Network (CNN), a data driven approach, to automatically predict the magnetization inclination and declination of a magnetic source body. CNN has achieved great success in other applications such as computer vision, but has not been attempted in the realm of magnetics. We simulated 16,380 magnetic data maps with different magnetization directions from a synthetic source body, all subject to the same background field. Two CNNs were trained separately, one for predicting inclinations and the other for predicting declination. We achieved 98% and 100% testing accuracy in declination and inclination predictions, respectively. In addition, we investigated the effect of having different parameters such as ambient magnetic field strength, survey height, magnetic susceptibility, source body shape, and source body location, on the predication accuracy We also applied the procedures to a set of field data from Black Hill Norite in Australia, and achieved satisfactory results. Our study shows that CNN holds great promise for automatically predicting magnetization directions based on magnetic data maps.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFMGP42A..09N
- Keywords:
-
- 1599 General or miscellaneous;
- GEOMAGNETISM AND PALEOMAGNETISM