Monitoring and Detecting Transient Deformation in Taiwan Using Dense GPS Arrays and the Network Strain Filter

Transient deformation, or slow slip, is the result of slip on a fault with rupture velocities much less than the shear wave velocity [Ohtani et al., 2010]. Studies have shown that transient events can release accumulated energy equivalent to up to Mw 6.3-7.2, implying that such deformation can make significant contributions to the moment release budget [Wallace and Beavan, 2010], and the occurrence of these events could trigger the generation of other earthquakes. Taiwan is an island situated at an active plate boundary with high seismic activity. With a convergence rate of 80 mm/yr between the Philippine Sea plate and the Eurasian plate [Yu et al., 1997], such a tectonic setting results in frequent earthquakes as well as slow slip events, making Taiwan an ideal place to study the relationship between earthquakes and transient deformation. In order to detect transient events from massive geodetic data, we intend to use the Network Strain Filter (NSF) published by Ohtani et al.[2010], a tool to detect transient deformation from large-scale geodetic networks, to look for signals in the continuously-recorded GPS time series in Taiwan. In the NSF, GPS time series data are expressed as a sum of steady site velocities, spatially coherent displacements representing transient deformation, seasonal variation, site-specific benchmark wobble, and reference frame errors. From the time series, it is possible to derive time-dependent maps of surface displacement, velocity, and strain rate, from which it is possible to detect transient deformation. Transient displacements are represented by a spatial wavelet basis with temporally varying coefficients that are estimated with Kalman filter. After the transient deformation patterns are identified, we will model the deformation as slip on a planar fault in an elastic homogeneous half space [Okada, 1985] for each event, to verify that the signals could result from changes in slip-rate along the faults in different regions. With the model results, the seismic moment (M0) could then be calculated as defined in Brune [1968]. Combining existing earthquake data and the accumulated magnitude of transient events can therefore provide a more precise estimate of strain accumulation of the region and further leads to a better understanding of the earthquake potential estimation in Taiwan.