Research On The Methods Used To Process And Inverse The High Precision Magnetic Data

In this thesis, some methods that used to process and inverse the high precision magnetic data are presented. High precision magnetic survey has been carried out for many years, but the methods that used to process and inverse the data have not been improved. In this thesis, the methods that used to separate and pick up the magnetic anomalies, and the methods used to find the minimal norm solution are especially introduced. First, previous methods that used to process the magnetic data, especially the algorithms that used to compute the continued data and the derivative data, and the factors that affect the resolution of these algorithms are all discussed in detail. Then the principle how wavelet analysis can be used to filter is interpreted. As anomalies separation and pickup have the same essential as filter, so wavelet can be used to anomalies separation and pickup. At last two kinds of inversion approaches are presented: first search for the shape of the source structures "knowing" the contrast as well as certain constrains on their geometry; the second searches for the contrast distribution. At the same time two inversion algorithm: Levenberg-Marquardt and general inversion are used respectively to these two kinds of approaches and each of their merits are analyzed. All processing and inversion algorithms have been explained in detail in this thesis. After all methods applied to numerical model and field data, some useful advice and conclusion can be given. The achievement of the thesis can be described as follow:1.Principle and algorithm of previous magnetic data processing methods have been presented in detail and every algorithm has been programmed. Tested on synthetic example, the results show that the spatial domain data processing methods presented in the thesis have higher resolution than the previous wave number domain methods.2.Wavelet has been discussed, this is a new data processing method and the test on synthetic examples and field data show that it is a good latent method.3.Two inversion approaches have been presented and discussed in detail, including the possible further work and each of the two approaches' merits.