Joint Inversion Method Of Magnetic Anomaly And Gradient Multi-parameters Data And3D Human-computer Interactive Modelling Technique

In the magnetic exploration, gradient measurements’ advantages over total fieldmeasurements are as follows: they can reflect abnormal characteristics of shallowsubsurface zones and improve the identification of superimposed abnormal bodies.Therefore, a joint inversion method specific to total field and corresponding gradienttensor has gradually become a hotspot research. But little effort has been dedicated tostudying how to theoretically explain why gradient measurements can have anadvantage, and how to maximize the use of gradient advantages in the inversion withmulti-parameter constraints. In this paper, multi-parameter inversion using magnetictotal field and gradient is the research object, and the superiority of joint inversionwith gradient data has been evaluated in theory. So the above-mentioned questionshave been answered. Meanwhile, it have put the techniques of human-computerinteraction modeling and3D visualization as research tools, aiming at researching thejoint inversion method with constraints of magnetic anomalies and gradient anomaliesand developing practical software.1. Based on theoretical formula of the forward of magnetic fields, the features ofmagnetic gradient data and root of the "superiority" of gradient measurements overmagnetic total field are analyzed. Meanwhile, for field sources with different depthand different size and measuring grid with different scale, this paper had pointed outthe complementary between magnetic total field and magnetic gradient: these dataperformed characteristics of the field from different aspects. To obtain total variationof field and thus reduce the multiplicities of geophysical interpretation, theinformation of multi-parameters should be cooperative considered in inversion. So thespecific technical measures of multi-parameters inversion were determined finally.2. Modeling analysis proves that effects of gradient data in multi-parametersinversion should be evaluated by two sides: for large scaled study objects, gradientdata can assist total-field data to optimize quantitative inversion precision; for smallscaled detected objects, gradient data can supplement lack of information in total field and thus enhance the detecting resolution. The second effect is often more valuable inactual exploration3. The limitations of using gradient data in multi-parameters inversion witherrors has been analyzed from simulation of different types of instrument noises.4. Based on analysis of differences between measured gradient and convertedgradient in the measuring and calculating process, it has been confirmed that asanother expression form of total field, converted gradient can not take the place ofmeasured gradient and not be used in multi-parameters joint inversion. Thus thenecessity of the gradient measurement has been proved.5. Aiming at solving the problems of big data management and inversion processcontrol from multi-parameters measurement, the system for forward and inversionwith multi-parameter magnetic fields by human-computer interaction have beendeveloped, in which program design technology were used. In terms of modelstructure, a topology polyhedron data structure have been designed. On model editing,this paper have deduced coordinate mapping formula between2D logical coordinateand3D geographic coordinates. Based on them, functionality on visualizationinteractive editing for models were realized. In inverting the shape of geologicalbodies, real-time forward and inversion were achieved by multi-threaded paralleltechnology.6. This paper made an exploratory attempt on combining the inversion forphysical properties and for geological body shape, aiming at making use ofadvantages of properties inversion and shape inversion and covering insufficients ontwo of them simultaneously. Based on the basic principle of iso-surface reconstructionand mesh simplification in computational geometry, conversion technique betweenphysical property models and shape models can be realized. Therefore, both inversioncan learn from each other and restrict mutually is possible.