Setting the stage for the 2008 Mw 6 earthquake at the Gofar transform fault, Pacific Ocean: slow slip, repeating earthquakes, interseismic and co-seismic activity from OBS data

Fast slipping oceanic transform faults show a quasi-periodic occurrence of large (M>5.5) earthquakes. For example, at the Gofar transform fault in the East Pacific Ocean, slipping at a high rate of ~14 cm/yr, some segments rupture every 5-6 years in a Mw~6 earthquake while other segments remain quite in the global record. Based on the regularity of the seismic cycles, US American researchers deployed an ocean bottom seismograph (OBS) network to capture the predicted 2008 Mw 6.0 event. Indeed, the event was record on 12^th September 2008, providing a unique seismological dataset of stations spaced at 10-20 km and OBS operating for 2 months to 13 months with 12 OBS out of 16 OBS covering the full observation period and recording three component data. Previous studies analysed the seismic behaviour (McGuire et al., Nature Geoscience, 2012; Gong and Fun, G-cubed, 2022), focusing solely on micro-seismicity. Here, we re-analysed the archived dataset down-loaded from the EarthScope Consortium (www.iris.edu) and (i) used a machine-learning technique developed to study large datasets of OBS data (PICK-Blue, Bornstein et al., EPS, 2024) to reveal seismicity pattern covering the interseismic phase before the earthquake, the co-seismic and post-seismic phase. In addition, (ii) we searched for similar earthquakes rupturing periodically the same patch of the rupture zone and found two classes of events: repeating at very short time intervals and hence within one day, which we call “bursts”, and events repeating within more than seven days, which we call “repeaters”. Such earthquakes are generally used to reveal seismic creep. Last, (iii) we search for time-dependent features in the continuous recordings and found evidence for spontaneous velocity-changes and gradual healing which we interpret in terms of slow slip events. We found that most bursts and repeaters occurred throughout the year at the segment eastward of the 10 km long mainshock area, while most other segments show little evidence for repeaters, except the segment to the west of the mainshock showing repeaters in April and May 2008. In addition, we observed slow slip in the mainshock area and at the two segments towards the east, while the other segments showed no evidence for prominent velocity changes within the fault zone. We conclude that the occurrences of creep on adjacent segments and slow slip loaded the later mainshock area over several months, subsequently issuing the mainshock. Most striking, the Gofar transforms shows contrasting seismogenic behaviour at adjacent segments: one accommodating plate motion by creep while the other issues large earthquakes.