| Inversion is one of the most important parts in gravity and magnetic data processing. However, inherent non-uniqueness of the inverted solution becomes a major obstacle to the field applications. Joint inversion has overcome the limitation of either method, fully excavates data information, and has the potential to obtain more reliable results, which makes it a highlight in the current research. Considering the practical requirements, the various use of constraints and joint inversion methodology of gravity and magnetic data were studied. In this paper, the following subjects were analyzed in detail:First, the establishment of the objective function was conducted for the single potential field data inversion, and generalized least-square algorithm was derived. We added the depth weightings directly by weighting the model terms. Meanwhile, a transform function is introduced to introduce the bounds information. The convergence of this algorithm is stable and efficient, which laid the foundation for the further development of joint inversion algorithm.Second, based on the structure coupling, cross-gradient joint inversion algorithm with multi-constraint conditions was developed. The objective function was built by combining physical property parameters and the corresponding data sets, and the structure coupling equality constraints was incorporated within the iterative framework. We proposed a so-called structure coupling factor to solve the structural disturbance, which significantly improves the inversion efficiency. In additional, depth-weightings and bound constraints are also used along with the cross-gradients coupled method to enhance the inversion quality. Furthermore, local cross-gradient coupling joint inversion algorithm was studied to meet the need of different practical cases. Synthetic examples show that the algorithm converges well, the model structures achieve high consistence. The main advantage for the algorithm is the capacity of distinguishing the anomalies with different characteristics. The cross-plot of physical properties show a convergence feature for the joint inversion situation, which is much different from that of separate inversion.Third, the influence of the remanent magnetization in joint inversion of gravity and magnetic data was analyzed for the first time. Because of the interactive information transmission mechanism of the joint inversion, when strong interference exists, the inverted results will be affected if ignoring the remanence. The normalized magnetic source strength, which is weakly sensitive to the direction of magnetization, was introduced in the joint inversion to conquer the unwanted effects of the remanent magnetization. Synthetic tests show that this method works better than the conventional joint inversion and the single inversion of normalized magnetic source strength. It indicates the importance of adopting this method to get better results when remanent magnetization exists.Fourth, based on petro-physical coupling, a study on guided multi-physical coupling joint inversion was studied. Membership function and physical property classes were introduced in this method. The inversion of petro-physical parameters along with their membership and classes were computed at the same time, guiding the iterative model parameters to fit the reality. Synthetic tests show that the constraints can be obtained with edge-preserved feature in the inverted model. The focusing characteristics of the model is much better than the physical bounds constraints. Also, the property anomaly recognition capacity is stronger than the single constraint. The cross-plot of the coupling method yields physical scatters near the corresponding class centers, which proves the focussing feature of this method.Fifth, a field data set obtained in the middle and lower reaches of the Yangtze River was used to carry out joint inversion analysis, which further verifies the effectiveness and the practicality of the algorithm described in this paper. |
PDF Full Text Request