Simultaneous Inversion of Multi-Channel IP Data with Structural Simliarities

Time-domain spectral induced polarization data can be inverted by using the concept of four dimensional (4-D) inversion method. In this approach, the entire multi-channel data set is simultaneously inverted, where the apparent chargeability and model chargeability at entire time-channels are composed into a single data and model vectors in the space and channel domains. Regularizations not only in the space domain but also in the channel domain is applied to reflect the characteristic of chargeability curves. Furthermore, we introduce the concept of structural similarity between resistivity and chargeability models by using minimal cross-gradient constraints. Developed inversion algorithm was successfully applied to several simple subsurface models to obtain reasonable chargeability distribution. Through the structural simlarity constraint, more reasonable IP data distribution can be obtained since noisy IP data can resort to resistivity distribution which more robust and stable due to higher S/N ratio.