Exploring Seismicity around the Southern Central Section of the Alpine Fault, New Zealand, using Distributed Acoustic Sensing

The South Island Seismology at the Speed of Light Experiment (SISSLE) was an investigation of seismicity around the Alpine Fault, near Haast, New Zealand (NZ) using Distributed Acoustic Sensing (DAS) in 2023. In this study, measurements of seismicity were acquired on a 30 km long optical fibre that crossed the Alpine Fault and co-located nodes over an approximately three-month period. An analysis of this dataset is reported here. Coincident detection with manual arrival picking were used to identify the timing of 110 local (< 50 km from the fibre) and regional (>= 50 km) earthquakes detected at the nodes. This timing was used to identify earthquakes in the DAS archive which were manually picked. A subset of 25 of the local DAS records were then re-picked using a semi-automated manually guided picker developed to improve pick consistency. Delayed arrivals were noted at specific locations along the fibre in all records examined. These guided picks (excluding those that appeared to be delayed) were fitted using a least-squares approach based on ray tracing solutions to the Hamilton equations and NZ's current velocity model to determine earthquake hypocentres. Fitted epicentres were comparable to the epicentres reported by GeoNet where available. Fit uncertainties were assessed by introducing Gaussian noise proportional to the acoustic pathlength into the arrival data. Delayed arrivals, computed from the difference between the observed and fitted arrivals, varied systematically along the fibre suggesting that low-velocity structures were present. Motivated by correlations between valley reconstructions based on solutions to Laplace’s equation and the location of the delayed arrivals, it was proposed that these structures may be sediment-filled basins. Several basin topologies were investigated using a forward modelling approach. This modelling demonstrated that the measurements were insensitive to near-surface off-fibre structure and that the median arrival delay, <Δt>, was approximately proportional to sediment depth z_sed, so that if the seismic velocity of the sediment is assumed to be constant v_sed then: z_sed=-v_sed×<Δt>. This simple interpretation appears to be consistent with previously reported gravity and seismic reflection data on lines approximately co-located with the DAS fibre. Using a combination of ray tracing and wavefield modelling, `late’ fault zone head waves, that follow the direct arrivals, were predicted for a single local earthquake. These late fault zone head waves were not observed experimentally suggesting that the reflectivity of the Alpine Fault at Haast is low. `Early’ emergent arrivals were observed for two regional earthquakes located near the Alpine Fault. Because of the fibre’s orientation perpendicular to the fault, it was not possible to establish that these emergent arrivals were fault zone head waves from the DAS data alone. If these are in fact fault zone head waves, then the average velocity contrast Δv_P/v_P between the hanging wall and footwall of the Alpine Fault over approximately 100 km to the south and north of Haast are in the range 8.3–12 % and 1.8–2.5 %, respectively.