2D Magnetotelluric Inversion Using Hybrid Stabilizing Functionals: Exponential Minimum Support and Smoothness

Except for producing smoothing models, magnetotelluric in-version also requires focused inversion, which can image sharp boundaries and clear interfaces of electrical structures. The pro-posed minimum support (MS) and minimum gradient support (MGS) have good focusing ability; however, their focusing ef-fect is highly dependent on the selection of the focusing param-eter. For finding a more stable way to achieve a well-focused image, we have developed an exponential minimum support (EMS) stabilizing functional, which combines the properties of the exponential stabilizer and MS, and determined the enhanced focusing ability and smaller dependence on the magnitude of the focusing parameter. Moreover, we construct a model constraint with the hybrid stabilizing functionals EMS and the smoothest (SM) and fulfill the optimization based on Occam's inversion scheme. Through theoretical analysis and model tests, we determine the selection schemes of the regularization parameter, weighting parameter between EMS and SM, and focusing coefficient in EMS to ensure the stability and reliability of this method. Afterward, we perform the inversions of the synthetic and field data. By comparing the inversion results of the two synthetic wedge models using hybrid EMS and SM and those obtained using other stabilizers, we determine that the hybrid method is capable of imaging sharp boundaries and is stable for a wide range of focusing parameter values. The inversion result of the COPROD2 field data set clearly images the high-conductivity anomalies at depths of 10-60 km and also depicts the highly conductive features within the upper mantle at depths greater than 100 km. The model tests and field data application verify the stability and capability of this method. Thus, this novel regularization approach provides a tool for sharp boundary inversion and a new reference model for deep structural interpretation.