Subsurface imaging using magnetotelluric data

A linear programming approach is developed to construct a pseudo‐impulse response for magnetotelluric (MT) data. The constructed time function is made up of discrete pulses whose amplitudes depend upon the electromagnetic reflection and transmission coefficients at various layer interfaces. The arrival time of an individual pulse corresponds to the time for a reference signal to travel a particular raypath from the surface to a reflector and back. The display of the impulse responses recovered from many stations produces an MT reflectivity section which is analogous to the image ray section regularly interpreted in reflection seismology. Application of linear programming inversion to one‐dimensional conductivity models shows the viability of the method and validates the physical interpretation of the pseudo‐impulse response function. Using a number of simple two‐dimensional geologic models, we show that a line of MT stations acquired perpendicular to strike produces a reflectivity section which is an image of the explored target. The interpretation of the MT image section follows the conventional guidelines used in reflection seismology; features such as traveltime pullup, primary and multiple reflections, and diffraction events are evident on the final section.