A minimal-code framework for 3D magnetotelluric inversion integrating fuzzy c-means clustering

Magnetotelluric (MT) inversion plays a critical role in the exploration of deep mineral resources. The resistivity models produced by conventional MT inversion algorithms are smooth. Integrating fuzzy c-means (FCM) clustering into MT inversion allows for sharp resistivity contrasts. However, existing methods face significant challenges due to the complexity of the problem and high computational cost. To address these issues, we propose using a dual-loop mechanism to create a minimal-code framework that integrates FCM clustering into 3D MT inversion (MinCode-FCM-MT). The dual-loop mechanism couples an inner-loop iterative inversion with outer-loop clustering correction. Through this coupling, the FCM clustering is executed during outer-loop cycle on resistivity models. This framework allows for flexible modification of key variables, enabling precise control over the inversion process and improving the delineation clarity of anomalous boundaries. Synthetic examples demonstrate that the MinCode-FCM-MT inversion effectively resolved anomaly boundaries more accurately than conventional MT inversion. Finally, inversion of the MT data collected over the Beiya deposit on the southeastern margin of the Tibetan Plateau further confirms the practicality of the new framework. The resistivity model found using MinCode-FCM-MT provided critical support for the precise imaging of geological unit boundaries and identification of anomalous electrical resistivity structures. • Proposed 3D MT inversion framework with FCM clustering using a dual-loop mechanism. • The framework enhances the clarity of the resistivity anomaly boundary. • The framework delineated the conductive anomaly beneath the Beiya deposit.