Using the Quake-Catcher Network (QCN) to derive source parameters and the site attenuation term, kappa (κ), using aftershocks of the 2010 Darfield, New Zealand earthquake

We utilize a dense network of Quake-Catcher Network (QCN) MEMs accelerometers to investigate source parameters and the shallow site attenuation parameter, kappa (κ), for aftershocks of the 3 September 2010 Mw7.1 Darfield earthquake in Christchurch, NZ. Approximately 190 QCN accelerometers captured over 180 aftershocks ≥ Mw4.0 from 9 September 2010 to 31 July 2011. Sensors were deployed in local residences as part of the QCN Rapid Aftershock Mobilization Project (RAMP), collecting vast amounts of data at dense spatial scales. The low cost, 14-bit QCN sensors perform within ANSS Class C sensor standards (Evans et al., 2013), and, the time series and response spectra of the sensors compare favorably to the strong-motion 24-bit NZ GeoNet sensors (Cochran et al., 2011). To find κ, we measure deviations from the ω-2 fall-off on the acceleration amplitude spectrum of Fourier-transformed S-wave windows containing 80% of the S-wave energy. We use both manual and automated methods to fit the slope of the fall-off (i.e., κ) following Anderson and Hough (1984). A known issue with this method is that κ should be measured above the corner frequency (f0) to avoid bias from source effects. Studies have recently reported larger than average stress drops for these aftershocks (e.g., Kaiser and Oth, 2013), which may yield significant variation from the theoretically determined f0. Here, we aim to find the site attenuation, κ, by simultaneously solving for f0 and the seismic moment (M0) for each station and event. For robust results, we employ several methods to find the source and site parameters. Initially, we use a linearized least-squares fitting routine for each event-station pair (e.g., Anderson and Humphrey, 1991). This method does not require a single M0 for an event recorded at multiple stations, resulting in disagreements across M0 and f0 for any given event. Consequently, we also employ a more physically meaningful approach that calculates a single M0 and f0 for a given event using a linearized general inversion scheme (e.g., Sarker and Abers, 1998; Stachnik et al., 2004). Due to a strong trade-off between f0 and κ, we lastly try the nonlinear least-squares Gauss-Newton algorithm, which obtains a constant M0 and produces a more reasonable f0 and well-fitted κ. Initial results show κ estimates range from 0.01 to 0.1 sec and our calculated moment magnitudes (Mw) agree with the USGS NEIC catalog. Another goal of this study is to determine if the QCN data can be confidently used to find source parameters. With the vast amount of strong-motion data collected, QCN offers an ideal dataset to determine source parameters from spectral fitting; particularly in Christchurch, where smaller datasets may contain a proportionally higher number of recordings that are biased by local effects, including site amplification and nonlinear response like liquefaction. For a single event, preliminary findings show that QCN sensors yield higher M0 values than GeoNet stations, thus prompting further investigation.