| With the development of satellite,airborne and ground multiscale data measured techniques,and the increasement of the measured data precision,the traditional potential field interpretation methods fails to obtain satisfactory results with low precision and more priori information.The new geological target position and physical property interpretation methods need to be established.In this thesis,we primarily evaluate the influence of survey design parameters on the target detection ability,and the effect of different data type and multilevel data measured at different height surface on the target interpretation by the data analysis tools.Based on the analysis results,we propose several target horizontal position,buried depth and physical property inversion methods using the multilevel gravity and magnetic gradient tensor data.The new techniques have better performance on the geological target interpretation and provide more accurate essential data and basis for geophysical modelling.Firstly,we focus on the geophysical data acquisition procedure and use the data analysis tools Depth Resolution Plot(DRP)and Discrete Picard Plot(DPP)to analyze the survey design parameters(line spacing,line layout position,flying height,instrument precision)effect on the interpretation results.The analysis is useful for inversion methods establishment and the survey parameters design.Additionally,we make a comparative analysis of the target detection ability between gravity or magnetic anomaly and the gradient components,and evaluate the multilevel measured data and the single observation surface anomaly detection performance.Space domain and frequency domain data analysis techniques show potential field gradient anomaly has better resolution in detecting shallow geological bodies,and the gravity and magnetic anomaly contain more long-wavelength information.The multilevel data analysis results indicate that multilevel data provide more large-amplitude singular values for inversion.Through the quantification of survey design parameters,measured potential field type and data distribution on target detection ability,new methods developed based on the high precision multilevel potential gradient tensor data have more advantages.We excavate the useful target information contained in measured data and take advantage of different heights observed gradient data,then develop automatic inversion methods for buried sources parameters with low computed error.The source parameters contain the horizontal position,buried depth and physical property distribution.Therefore,the mainly research theme of this thesis is about new methods to determine this three parameters.We develop edge detection methods for horizontal position calculated based on the potential field gradient tensor,depth estimation method and imaging method based on the multilevel data,physical property inversion method constrained by the geometric parameters inversion results.The edge detection methods are widely used in geological target horizontal position interpretation.The traditional methods can generate unclear edge position and larger error for deep geological structures.In this thesis,we propose several edge detection methods based on potential field tensor data.We firstly analyze the commonly used curvature gradient tensor eigenvalue edge detector on deep and shallow bodies caused anomalies.The result shows that the method fails to detecting edges of different amplitude anomaly bodies.In order to improve this problem,a new directional gradient tensor eigenvalues method is established,and we use the vertical derivatives of the potential field to balance deep and shallow bodies anomalies,the synthetic and real measured data application results show that the new method can outline different amplitude anomalies simultaneously,but it has poor horizontal resolution.We introduce the structure tensor in image processing field to the potential field interpretation to improve the problem.The new method of the directional horizontal derivatives based on structure tensor obtains more clearly edge position.The application on the synthetic and measured potential field data demonstrates the new method effectiveness.Furthermore,in order to improve the offset problem of the traditional method on deep geological bodies edge detection,a new edge detector based on three dimensional structure tensor is developed.Because of using more tensor components,the edges detected by the new method are more accurately,the real measured data processing results demonstrate the superiority of the new method,and the new method detects more reliable detailed structures.Compared with other geophysical methods,gravity and magnetic data has poor depth resolution,but the fast and reliable depth estimation is also the important task of potential field data interpretation.The traditional depth estimation methods need more priori information and cause poor precision estimated results with complicated computation procedure.This paper modifies the traditional Tilt-depth method,analytical signal method and normalized total gradient method.The Tilt-depth method has poor horizontal resolution,and we propose improved strategy for depth estimation to obtain high depth estimated precision.In order to achieve different source type geological bodies depth information,we develop generalized Tilt-depth method based on multilevel data,the synthetic and measured data are used to test the method performance.The traditional analytical signal method need structural index priori information or high-order derivatives of potential field,we develop multilevel analytical signals method by relating the depth parameter with three observed surface heights,the application of the method on the synthetic and real data indicates the method can suppress the noise and obtain reliable source parameters.The normalized total gradient method is the normalization of analytical signals at different downward continuation depths.In order to estimate the source horizontal position and depth extent simultaneously,this paper defines the normalized edge detection method based on directional total derivative filter.The synthetic and measured data test results demonstrate the maximum position of the new method 3D results indicate the buried bodies position.The edge detection,depth estimation and imaging method can be used for source position inversion,and they belong to geometrical parameter inversion category.The physical property inversion is used for physical property parameter inversion.The geometrical parameters inversion and physical property parameters inversion constitute the complete potential field interpretation.In this thesis,we construct constrained condition based on the geometrical parameters inversion results and used for physical property inversion.We choose the focused inversion method to solve the inversion question.The synthetic simple model result indicates focused inversion obtains source physical property distribution with steep boundary,and achieves large error level for the synthetic model with deep and shallow sources.In order to improve the inversion resolution,we constrain the horizontal and depth extent information to the physical property inversion by introducing depth weighting function and horizontal gradient weighting function.The new proposed self-constrained physical property inversion method can obtain reliable physical property distribution without enough other geophysical and geological priori information by constraining the estimated geometrical parameters to the inversion process.The developed gravity and magnetic data geological target position and physical property inversion methods in this thesis aim to excavate the source information contained in the measured data,and provide more detailed information of the source spatial positon and physical property.The synthetic and real measured data application results demonstrate the new methods have good application value. |
PDF Full Text Request