Computationally efficient EM modelling using Shur decomposition applied to large scale airborne applications

Inversion of airborne electromagnetics problems can be extremely challenging due to the large number of transmitter locations. Differential-equation based solutions offer compelling advantages over integral equation and stitched 1D inversions, but come at a high computational cost and can struggle with the inclusion of the air layer. We present a new modelling scheme that is well-posed in the air, is extremely memory efficient. The algorithm is numerically attractive. Furthermore, preconditioners can be used to which allow for very quick convergence of iterative solvers. These preconditioners can be used across transmitter locations, source frequency, and conductivity model—making the algorithm well suited for AEM forward modelling and inversion on parallel architecture.