The magnetotelluric (MT) method has been widelyemployed in the exploration of hydrocarbon and mineralresources.Traditional linear iterative inversion method can determine the electrical resistivity of the Earths subsurface fromMT data rapidly, but it relies on the gradient of the forwardoperator and its result dependents on the initial model extremely. In order to avoid the disadvantages of traditional linearinversion, a novel non-linear geophysical inversion algorithmis proposed for the MT data based on improved differentialevolution. The proposed algorithm is applied to invert thesynthetic MT data of 1D layered geo-electrical models. Theconsistent results are obtained in the noise-free cases. WhenGauss noises of 10% and 20% are added to the syntheticdata, the results of inversions remain fairly good. Numericalexperiment results demonstrate that the improved inversionalgorithm has the advantages of independent of initial model,capable of global exploration, and anti-noise capability. It makesMT data inversion more effective.
展开▼
