Research On Crossgradient Joint Inversion Of Multiple Geophysical Data And Its Application

Crossgradient joint inversion method does not depend on the petrophysicalrelationship, and assumes the structural similarity between each geophysical model, soit can be applied to the joint interpretation of any different geophysical explorationmethods. The crossgradient joint inversion method implements structural coupling ofmultiple geophysical models. When the petrophysical relationship is unkown, it canimprove the reliability of the inversion results, reduce the non-uniqueness of theinversion, and reconcile contradictions between interpretation results of the differentmethods. Joint inversion, like a separate inversion method, need to add someregularization and the effect of joint inversion have close relattionships with theeffectiveness of the separate inversion methods.This paper mainly developed joint inversion method of gravity, magnetic, electricaland seismic data. In order to enhance the efficiency of the joint inversion method, thispaper improves Occam’s inversion method. Occam’s inversion method has advantagesthat has reliability and ability to consider various prior informations, but becausecalculating of the Lagrange multiplier takes a long time, many researchers haveproposed several direct selection methods. This paper improves the calculation methodof Lagrange multipliers in two–dimensional magnetotelluric Occam’s inversion. In thesearch progress, this method adds omparison of data fitting and its target level.Improved method, in the phase of smoothing model of the inversion, can eliminatesuperfluous forward calculations. This improvement has no effect on the inversionresults, but it can accelerate inversion and more accord with Occam’s idea. According tothe model experiment and field data processing, the improved method can reduce thenumber of forward calculation by20%~50%.The joint inversion method, when the observation data have different theircoverage, it is usually required to extract the data corresponding to the overlapped area,or extend some models. This paper proposes and implements crossgradient joint inversion method with sub-region crossgradient constraints, then tests a synthetic modeland demonstrates the effectiveness of the algorithm. The method of sub-regioncrossgradient constraints not only well recover structural similar model within theoverlapped region, but also at the boundary with the non-overlapped area overlappedarea can still obtain smoothly varied models.This paper proposes a joint inversion of multiple crossgradient joint inversionmethod based on the data space method. The data space method in the memory storageis more effective than the model space method, because in general cases the dataparameter numbers of the inversion problem is far less than the model parameternumbers. This algorithm uses a rectangular element with four triangular subelements,the algorithm is able to handle terrain and easier to measure crossgradient function andprocess the model covariance matrix in the data space method. This paper also proposesthe default values of weight factors of model gradients in crossgradient constraints item,analysis the effects of different weights through synthetic examples.The developed inversion algorithm is applied to Jinchang prospecting and thecomprehensive geological and geophysical interpretation of Benxi-Ji’an deep geologicalsurvey of gravity, magnetic and magnetotelluric method. The results show that all of themodel space joint inversion and the data space joint inversion can recover more similargeophysical models than separate inversions in their structures, and conducive to betterdetermine geological informations such as the nature of geological bodies andidentifying deep geological structure of the study area, and the data space method ismore effective than the model space in the memory requirments.