In order to get accurate source parameter estimates from single station information, the whole waveform has to be analyzed. In this process, knowledge of the local velocity structure is essential. To investigate the local structure directly, a one-dimensional receiver function study of four teleseismic events was performed. A receiver function study isolates the local structure response from all other factors which interact to form the observed seismogram Langston (1979). The information in these so-called receiver functions is then translated into a simple model of local velocity structure by using an inversion method developed by Ammon et al. 1990. The initial crustal model for this analysis was adapted from Soviet Deep Seismic Sounding surveys.

The receiver function data showed large tangential components, with the radial components showing large scattered phases directly after direct P. The scattered phases appear to be composed of Rayleigh waves, based on particle motion. In order to examine possible topographic scattering indicated by the Rayleigh phase, synthetic seismograms were constructed using the T-matrix method. This method uses the extinction theorem to derive two matrices that relate the incident waves to the scattered waves, and to the waves on the surface Clouser (1992); Lakhtakia et al. (1984). The resulting scattered phases, however, were not large enough to explain the data. Therefore, the observed data have to be caused mainly by crustal scattering. The complexity of the receiver functions suggests that the crustal structure near Garm has significant lateral heterogeneities. The receiver function method, however, gets accurate results for homogeneous layers, a fact which makes the finding of an accurate crustal model from the teleseismic data impossible. Since it was not possible to determine the crustal structure directly, I used the initial model that was given by the DSS surveys in the following location study.

11/17/1997