Three-dimensional magnetotelluric inversion using non-linear conjugate gradients

We have formulated a 3-D inverse solution for the magnetotelluric (MT) prob lem using the non-linear conjugate gradient method. Finite difference metho ds are used to compute predicted data efficiently and objective functional gradients. Only six forward modelling applications per frequency are typica lly required to produce the model update at each iteration. This efficiency is achieved by incorporating a simple line search procedure that calls for a sufficient reduction in the objective functional, instead of an exact de termination of its minimum along a given descent direction. Additional effi ciencies in the scheme are sought by incorporating preconditioning to accel erate solution convergence. Even with these efficiencies, the solution's re alism and complexity are still limited by the speed and memory of serial pr ocessors. To overcome this barrier, the scheme has been implemented on a pa rallel computing platform where tens to thousands of processors operate on the problem simultaneously. The inversion scheme is tested by inverting dat a produced with a forward modelling code algorithmically different from tha t employed in the inversion algorithm. This check provides independent veri fication of the scheme since the two forward modelling algorithms are prone to different types of numerical error.