CHINESE JOURNAL OF GEOPHYSICS Vol.57, No.6, 2014, pp: 800–808 THE IMPROVED ANISOTROPY NORMALIZED VARIANCE FOR DETECTING NON-VERTICAL MAGNETIZATION ANOMALIES ZHANG Heng-Lei 1,2,4 , Y. R. Marangoni 2 , ZUO Ren-Guang 3 , HU Xiang-Yun 1,4* 1 Institute of Geophysics and Geomatics, China University of Geoscience, Wuhan 430074, China 2 Geophysics Department, Sao Paulo University, Sao Paulo, Brazil 3 State Key Laboratory of Geological Process and Mineral Resources, China University of Geoscience, Wuhan 430074, China 4 Hubei Subsurface Multi-scale Imaging Key Laboratory, Institute of Geophysics and Geomatics, China University of Geosciences, Wuhan 430074, China Abstract For the magnetic field with unknown magnetization direction, it is necessary to study the edge detection for the magnetic source of oblique magnetization directly. In this study, we propose a new edge detection called anisotropy normalized variance based on magnetic gradient tensor (ANV MGT). First we use an anisotropy scale to improve the core function of ANV to show the effect of the anisotropy Gaussian function. Then the magnetic gradient tensor mode is used in the ANV method to process the anomalies with oblique magnetization. We construct a numerical model of complex magnetic anomalies with an oblique magnetization, which is used to test the edge detection method. The model test shows that the proposed method, ANV MGT, can detect the edge of the magnetic source with non-vertical magnetization direction. It also shows that the proposed method is better than application of ANV on the 3D analytic signal amplitude. In the application to real data from western China, a perfect magnetite ore location is determined by performing ANV MGT for the non-reduction to the pole data. Key words Magnetic exploration; Anisotropy normalized variance; Gradient tensor; Edge detection 1 INTRODUCTION The fine processing for gravity and magnetic anomalies can quickly obtain the information about the structure, lithological change, and other underground geological information in potential field study, which is important in geological and geophysical interpretation. In recent years, a lot of research have focused on the edge detection (Lahti and Karinen, 2010; Cooper and Cowan, 2011; ZHANG Heng-Lei, et al., 2011; ZHANG Heng-Lei, et al., 2012; MA Guo-Qing, et al., 2012; Santos et al., 2012; WANG Yan-Guo et al., 2013; Cooper, 2013; Zhou et al., 2013). For the edge detection methods for magnetic anomalies, most of them are based on the vertical magnetization field, which implies the reduction to the pole (RTP) is necessary. While when the magnetization direction is unknown, the RTP field cannot objectively reflect the vertical magnetization field characteristics. In addition, the instability of the RTP at low latitude also restricts the application. Since Nabighian (1972) pointed out that the analytical signal amplitude (ASA) can eliminate the influence of oblique magnetization of the 2D magnetic anomaly, the 3D ASA has been widely applied, although it does not completely eliminate the effects from 3D source, while at low latitudes or unknown magnetization direction study areas, the 3D ASA is commonly used to reduce the oblique magnetization. Based on the 2D body magnetic anomaly equations, Verduzco et al. (2004) stated that the horizontal derivative of Tilt gradient is independent of the magnetization direction, and it has been applied in the 3D case. In recent years, with the improvement of the magnetic gradient tensor measurement system, the appropriate treatment methods and interpretation techniques developed rapidly (Clark et al., 2009; Beiki, 2010; Beiki and Pedersen, 2010; Beiki et al., 2011; Beiki et al., 2012; Oru¸ c et al., 2013). Previous studies have indicated that we can reduce the magnetization direction E-mail: sky0610@163.com *Corresponding author: xyhu@cug.edu.cn