Depth determination for shallow teleseismic earthquakes: Methods and results

Abstract

The depths of moderate‐sized shallow earthquakes can be routinely determined to an accuracy of several kilometers using teleseismic data alone, making it possible to determine the vertical positions of earthquakes within the lithosphere. This capability provides a valuable tool for plate tectonic studies since earthquake depths can be used to constrain lithospheric thermal and mechanical structure at plate boundaries and in intraplate regions. Two basic techniques are used for depth analyses. Surface wave spectra can be used to determine focal depths, as mode excitation reflects the behavior of the eigenfunctions with depth. The primary limitation on such studies stems from lateral heterogeneity of velocity structure. The waveforms of body waves are also diagnostic of focal depth, which controls the time separation between the direct arrival and near‐source surface reflections. These studies can be limited by a partial trade‐off between depth and source time function duration. The reliability of the body wave modeling approach, now in general use, is shown by excellent agreement between depths determined by different investigators using different algorithms. The depth determination is robust to typical uncertainties in focal mechanism and near‐source structure. The trade‐off with source time function duration generally does not produce major difficulties for small events (Ms < 6.5). For larger events, the trade‐off can be resolved using a approach in which, for a given depth, the time function yielding the best fit to a suite of seismograms is found by deconvolution. Deconvolution at a series of depths yields a misfit function, whose minimum indicates the focal depth. Tests with synthetic data show that the deconvolution process yields reliable depth estimates even given uncertainties in focal mechanism and near‐source structure. It successfully handles multiple sources, moderate‐sized finite sources, and horizontally propagating ruptures. Good results are obtained for vertically propagating ruptures until the fault area becomes so large that waveforms differ significantly from the point source approximation. A number of important tectonic results have emerged from studies using these depth determination methods. The maximum depth of oceanic intraplate seismicity increases with lithospheric age and is approximately bounded by the 750°C isotherm. This depth roughly equals the flexural thickness of the lithosphere but is much less than the depth to the low‐velocity zone indicated by surface wave dispersion. The maximum depth of continental intraplate seismicity is also temperature limited, but by a lower temperature. These observations are consistent with the rapid weakening of the appropriate rocks with depth predicted by standard temperature dependent rheologies. The maximum depth of oceanic transform seismicity corresponds to an isotherm of about 400°C. This temperature is lower than would be predicted from the intraplate results, and may indicate that transforms are either hotter or weaker than expected. The depths of ridge crest earthquakes large enough for teleseismic analysis are extremely shallow (0–8 km). The depth distribution of seismicity in continental normal faulting regions provides constraints on the nature of the extension process. At subduction zones, precise depth determinations can resolve the upper and lower flexural regions in the bending downgoing plate, precisely locate the plate interface, and map the vertical extent and position of faulting in large earthquakes.