Imaging transient slip events and their interaction with slow earthquakes in southwest Japan using reanalyzed GEONET GPS time series

Geodetic and seismic studies over the past decade have revealed a wide range of slow earthquake phenomena including long- and short-term slow slip events (SSEs), non-volcanic tremor, low and very-low frequency earthquakes (LFEs, V-LFEs) in major subduction zones. Despite much progress, the physical mechanisms of SSEs and seismic slow earthquakes and how these phenomena relate to each other are still not clear. The Japanese GPS network (GEONET) with ~1450 stations provide a unique opportunity to study subduction zone dynamics and to understand these different faulting behaviors at various spatiotemporal scales. We extended our reanalysis of GPS position time series for the entire GEONET from 1996 to 2014 using JPL GIPSY/OASIS-II based GPS Network Processor and raw data provided by Geospatial Information Authority of Japan and Caltech. The reestimated JPL precise GPS orbits, GMF troposphere model, and single receiver phase ambiguity resolution strategy were used in the analysis. The resultant ~18 years position estimates show good consistency and reduced scattering over the entire time period compared to prior analysis. We perform a systematic time series analysis to identify and correct the offsets caused by earthquakes, instrument and other unknown sources. We image the spatiotemporal variation of slip transients and investigate how they interact with seismic slow earthquakes. Our application to recurrent long-term SSEs in Bungo Channel region shows these events share the similar spatial distribution with regard to interplate coupling model, consistent with their recurrent nature. The universal modulation of transient slip rate on downdip LFEs/tremors and their different spatial patterns and moment release suggest that they are not different manifestations of the same physical process. There is considerable difference in space-time details of these slip transients, indicating that these recurrent events are not identical. We find a clear difference in temporal correspondence between transient slip rate and up-dip V-LFEs, possibly reflecting different triggering mode from that of LFEs/tremors. The high quality of GPS position time series also help reveal much smaller events over the "quiet" inter-SSE period, showing that not all SSEs are accompanied by LFE/tremors.