Resolution of earthquake source characteristics is limited by structural complexities in the source region. This study explores the effects of dipping near-source bathymetry and structure on long-period teleseismic body waveforms from shallow subduction zone earthquakes. Ray theory synthetics calculated for shallow double-couple sources and velocity structures incorporating planar dipping interfaces suggest that P waveforms are most sensitive to water depth and sea bottom orientation. For sources beneath a substantial water layer the seafloor dip of the inner trench slope can give rise to prominent later arrivals and waveform complexity not found on synthetics calculated for horizontal bathymetry. Sediment thickness and the dip of the sediment-crust and crust-mantle interfaces generally have lesser effects. The high-amplitude arrivals result largely from changes in the source takeoff angles of the reverberations. These arrivals present difficulties for source depth an time function determination from P waveforms and may cause spurious complexity in the derived pattern of earthquake moment release. SH waveforms show little effect of sea bottom orientations or local structure for shallowly dipping thrust faulting events, suggesting that SH waveform analysis can stabilize source parameter studies with respect to unknown or incompletely parameterized bathymetry and structure.

Effects of nonhorizontal structure are observed in the P waveforms from shallow Kurile Trench events (Ms 5.8--7.2), which show prominent later arrivals not observed on SH waveforms. Synthetics calculated for a dipping water-crust interface provide a better fit to these arrivals than those calculated for a horizontal interface, suggesting that the P waveforms show systematic effects of seafloor dip. The synthetics are highly sensitive to the assumed orientation of the seafloor; the data are fit best for seafloor orientations consistent with the dip of the inner trench slope. The distorting effect of the water layer gives rise to systematic differences in P wave complexity between shallow events near the seismic front, which occur beneath a thick layer and show numerous later arrivals, and deeper, arcward events which occur beneath a thin water layer and show simple waveforms. Large reverberations in the P waveforms of the October 20, 1963 and June 10, 1975 Kurile tsunami earthquakes have previously been used as evidence for a long source duration and a change in focal mechanism during rupture; however, SH waveforms suggest simple source time functions with durations of 25 and 43 s. Inversion of P and SH waveforms and amplitudes yield shallowly dipping thrust mechanisms, with no conclusive evidence for a previously proposed steepening of the fault dip during rupture.

The large P wave reverberations cannot be fit using a point source and a horizontal seafloor; detailed modeling of the 1963 event as an upward propagating rupture extending through low-velocity sediments to the ocean bottom combined with the effects of the dipping seafloor allow the reverberation amplitudes to be matched. The anomalous tsunami excitation appears to be related to the location of the sources at shallow depths within low velocity sediments. ¿ American Geophysical Union 1989