An Algorithm For Waveform Inversion Of Crosswell Radar Data

An algorithm for waveform inversion of crosswell radar data has been developed and implemented. The algorithm is based on an objective function that has two terms: One term measures<br>the misfit between the calculated data and the observed data, both of which are in the frequency domain. The other term measures the deviation between the current model and the starting model. Because the objective function is non-linear, it is minimized iteratively using conjugate direction and line search methods. The waveform inversion is tested with a model that is homogeneous, except for a small heterogeneity in the center. The estimated models have prominent anomalies that correspond to the heterogeneity in the original model and also have side-lobes that are adjacent to the prominent anomalies. Both the vertical and the horizontal resolutions improve as the wavelength decreases and are always better than the corresponding resolutions obtained with traveltime tomography. The vertical resolution is always better than the horizontal resolution.