The existence of remanent magnetizations poses a great challenge for interpreting magnetic data. Researchers have developed many different strategies to deal with the remanence problem. Recently, an inversion-based method, termed magnetization clustering inversion (MCI), was developed by combining classical Tikhonov regularized inversion scheme with an unsupervised machine learning technique, and was successfully applied to both synthetic and field data sets. However, the MCI method, in its current implementation, requires users to specify the values of two weighting parameters, one for the standard smoothness regularization term, and the other for the newly introduced clustering term. We have developed an efficient method for automatically searching for the optimal weighting parameter values. Our search algorithm is guaranteed to converge within three search stages. We have also successfully applied the automated MCI algorithm to a field data set from Carajás Mineral Province in Brazil.
Presentation Date: Tuesday, October 16, 2018
Start Time: 8:30:00 AM
Location: 213B (Anaheim Convention Center)
Presentation Type: Oral