井地磁异常模量联合反演

Joint inversion of surface and borehole magnetic amplitude data

三维反演是磁测数据定量解释的重要方法,在金属矿勘探中扮演着重要的角色.但是在实际矿区的应用中,传统的磁总场异常反演方法依然存在两个问题:一是地面磁异常反演的深度分辨率较低,深部场源体的成像效果差;二是金属矿中可能包含强剩磁,反演结果可能是完全错误的.尽管前人对上述两个问题分别进行了广泛的研究,但尚未尝试同时解决这两个问题.本文在前人研究的基础上,提出了一种井地磁异常模量联合反演方法,该方法需要的控制参数少,无需加入额外的地质信息,且可用于多场源复杂磁异常的反演,具有较强的适用性.本文方法首先将地面和井中磁异常转化为模量数据,然后利用基于核函数或距离的加权函数将井地模量数据结合起来,使得该方法适用于联合反演.我们利用井地多种异常参量进行反演的模型试验表明,在强剩磁存在时,本文方法的效果优于其他方法,在减少剩磁影响的同时,也改善了深部成像效果,具有良好的应用前景.

3D magnetic inversion for susceptibility distribution is a powerful tool in quantitative interpretation of magnetic data in mineral exploration.However,the inversion and interpretation of such data are faced with two problems.One problem is the poor imaging results of deep sources when only surface data are inverted.The other is the unknown total magnetization directions of sources when strong remanence exists.To deal with these problems simultaneously,we propose a method through the joint inversion of surface and borehole magnetic amplitude data.In this method,we first transform both surface and borehole magnetic data to magnetic amplitude data that are less sensitive to the directions of total magnetization,and then preform a joint inversion of the whole amplitude data to generate a 3D susceptibility distribution.The amplitude inversion algorithm uses Tikhonov regularization and imposes a positivity constraint on the effective susceptibility defined as the ratio of magnetization magnitude over the geomagnetic field strength.In addition,a distance-based or sensitivity-based weighting function is used to make the algorithm applicable to joint data sets.To solve this positivity-constrained inversion problem efficiently,an appropriate optimization method must be chosen.We first use an interior-point method to incorporate the positivity constraint into the total objective function,and then minimize the objective function via a Gauss-Newton method due to the nonlinearity introduced by the positivity constraint and the amplitude data.To further improve the efficiency of the inversion algorithm,we use a conjugate gradient method to carry out the fast matrix-vector multiplication during the minimization.To verify the utility of the proposed method,we invert the synthetic data using different inversion methods.The results show that the proposed method reduces the inherently nonuniqueness of magnetic inversion and also improves the imaging quality of deep sources in the presence of strong remanence compared with the other methods.