Résumé
In places where sedimentation and erosion compete at fast rates, part of the record of past earthquakes on faults may be buried, hence hidden, in the first few metres below the surface. We developed a novel form of palaeoseismology, of geophysical type, based on the use of a dense pseudo-3-D Ground Penetrating Radar (GPR) survey to investigate such possible buried earthquake traces, on a long, fast-slipping strike-slip fault (Hope fault, New Zealand), at a site (Terako) where marked alluvial conditions prevail. We first used LiDAR data to analyse the ground surface morphology of the 2 km2 site at the greatest resolution. Nineteen morphological markers were observed, mainly alluvial terrace risers and small stream channels that are all dextrally offset by the fault by amounts ranging between 3 and 200 m. The measurements document about 10 past earthquake slip events with a mean coseismic slip of 3.3 +/- 1 m, with the most recent earthquake event having a slip of 3 +/- 0.5 m. We then investigated a detailed area of the site (400 x 600 m2) with pseudo-3-D GPR. We measured 56, similar to 400 m long, 510 m spaced GPR profiles (250 MHz), parallel to the fault and evenly distributed on either side. The analysis revealed the existence of a palaeosurface buried at about 3 m depth, corresponding to the top of alluvial terraces of different ages. That buried surface is incised by a dense network of stream channels that are all dextrally offset by the fault. We measured 48 lateral offsets in the buried channel network, more than twice than at the surface. These offsets range between 6 and 108 m, as observed at the surface, yet provide a more continuous record of the fault slip. The similarity of the successive slip increments suggests a slip per event averaging 4.4 +/- 1 m, fairly similar to that estimated from surface data. From the total surface and buried 67 offset collection, we infer that a minimum of 30 large earthquakes have broken the Hope fault at the Terako site in the last about 67 kyr, with an average coseismic slip of 3.2 +/- 1 m, a minimum average recurrence time of about 200 yr, and a magnitude of at least Mw 7.07.4. Our study therefore confirms that part of the record of past earthquakes may indeed reside in the first few metres below the surface, where it may be explored with geophysical, GPR-based palaeoseismology. Developing such a new palaeoseismological tool should provide rich information that may complement surface observations and help to document the past earthquakes on faults.