Iterative Complexity Addition: A Method to Estimate Uncertainty in Geophysical Inversions

Abstract

Iterative Complexity Addition (ICA) is a method to estimate the uncertainty obtained from the inversion of geophysical data. The method utilizes a grid search with gradient descent for each set of model parameters to invert 1D magnetotelluric (MT) data. The algorithm finds a set of minimally complex local optima that meet the error criteria of the measurement, implying a possible solution to the inversion. The set of solutions generated can be analyzed for variance in the model parameters to generate statistical estimates of their values. These statistics reflect the likelihood of existence of each set of model parameters that can be used to enhance decision-making for subsurface resource development. The method design is here described with a discussion on algorithm strengths, weaknesses, and potential improvements.