RECORDING TELESEISMIC EARTHQUAKES USING OCEAN-BOTTOM SEISMOGRAPHS AT MIDOCEAN RIDGES

Existing teleseismic data recorded by ocean-bottom seismographs (OBS) are sparse, but they are sufficient for analysis of earthquake detecti on thresholds under various background noise conditions. Long-period P , S, and surface waves are consistently recorded by OBS's for magnitud e 5.7 to 6+ events at ranges greater than 100 degrees. Both Love and R ayleigh waves are recorded for very large events, with high coherence in the 15- to 70-sec period range; high coherence in the 15- to 35-sec range is typical for events of magnitude 5.5 to 6+. Short-period body -wave arrivals (1 Hz), on the other hand, have only been clearly recor ded by OBS's in the North Atlantic, during calm-weather periods or, by OBS's in the Pacific, for very large events at ranges less than a few tens of degrees. Seismograms recorded at the East Pacific Rise (EPR) and at the Mid-Atlantic Ridge (MAR) illustrate the high signal coheren ce between instruments deployed in an array of OBS's. Recordings of th e pressure field as well as the vertical and horizontal displacement f ields are used to assess the capabilities of OBS sensors and the frequ ency range of high signal-to-noise arrivals. Contamination of long-per iod, body-wave arrivals by secondary phases, due to reverberation in t he water column, can significantly hinder investigation of relative tr avel-time anomalies across an OBS array on rough seafloor, particularl y at low signal-to-noise ratios. The nature of the reverberations is i llustrated in short-period data, and the basic physics behind the diff erences between the pressure and displacement signals is discussed. A bias of about 0.5 sec can be introduced to relative arrival times, wit h deeper stations appearing erroneously late, for an array where seafl oor depths vary by about 1 km. Reflectivity synthetics provide the bas is for designing optimum filters for removing the reverberation bias i n long-period, P-wave data from the Mid-Atlantic Ridge, 34 degrees S. The resulting relative travel-time anomaly is 0.4 to 0.6 sec with dela ys with distance from the axis on the east flank of the spreading cent er.