Efficient Waveform Tomography with Low Frequency, Wide-Angle Data and Sparse Survey Geometries

We present a study of the ability of efficient waveform tomography to recover 2-D velocity models from a synthetic surface experiment. We show that tomographic, frequency domain waveform inversion using only a subset of the available frequencies can give results equivalent to waveform inversion with many frequencies, for a low starting frequency, using the 2003 CCSS dataset and crustal velocity model. Aliasing effects due to sparse survey geometries begin to affect the result only if sources are spaced with an interval larger than or equal to 3 x Nyquist, provided receiver spacings are unaliased. We investigate this effect further using increased starting frequencies, and find that waveform tomography can produce results consistently superior to traveltime tomography at frequencies up to approximately 3.0 Hz for this dataset.