Three-dimensional inversion of two-dimensional magnetotelluric data

摘要

In large-scale investigation of electrical structure, magnetotelluric (MT) survey is usually configured along profile orthogonal to regional strike direction. Therefore, two-dimensional (2D) inversion is now commonly used for the convenience of measurement and interpretation. However,our research suggests that 2D MT inversion considering only off-diagonal elements of impedance tensor can not provide off-profile structure which may sometimes imply the variation of resistivity parameter along the direction that parallels strike direction.In this paper, we process and invert 2D MT data set by 3D inversion program. We firstly eliminate the correlated noise caused by static effect employing moving window filtering in pseudo-section assisted by artificially recognizing the source of static effect on geology map and satellite image.After de-noising, MT data set becomes redundant for single reconstruction of resistivity model, so we split the whole data set into two distinct subsets and search their respective models independently. Covering the same model parameters in a particular area by repeating inversions, this is what we called "multiple coverage" of MT inversion which furnishes interpreters with more models and objective evaluation of reliability for the inversion.Our inversion results indicate that 3D MT inversion of 2D data is able to map the 3D distribution of resistivity on and off the profile; multiple coverage of MT inversion searches the best fitting models along distinct routes in model space and converges at the models close to each other; it also reveals the uncertainty of resistivity parameters in the area data set covers poorly.