My research focuses on studying the earthquake cycle deformation of the continental thrust faults and megathrust faults at subduction zones. I use GNSS (Global Navigation Satellite System), seismic, and geological data to investigate fault coupling, frictional properties, fault zone rheology, and slow-moving landslides. In particular, exploring fault rupture scenarios prior to the occurrence of megathrust earthquake is crucial for urban hazard mitigation and tsunami predictions. Seafloor geodetic measurements are crucial to study the spatial extent of the locked zone and transient slip events on the subduction megathrust. My research group and I have investigated different seafloor geodetic techniques including GNSS-acoustic observations and ocean-bottom absolute pressure gauge, aiming at establish a geodetic observatory network in the Taiwan plate boundary zone for understating the basic science and hazard of megathrust earthquakes.
I examine interseismic coupling of the Manila subduction zone and fault activity in the Luzon area using a block model constrained by GPS data collected from 1998 to 2015. Inferred coupling ratio is 0.34–0.48 at latitudes 15–19°N along the Manila Trench. The accumulated strain along the Manila subduction zone at latitudes 15–19°N could be balanced by earthquakes with composite magnitudes of Mw 8.8–9.2, assuming recurrence intervals of 500–1000 years. GPS observations are consistent with full locking of the majority of active faults in Luzon to a depth of 20 km. Inferred moments of large inland earthquakes in Luzon fall in the range of Mw 6.9–7.6 assuming a recurrence interval of 100 years. I found both the southernmost Ryukyu Trench and Manila subduction zone are subject to threats of Mw 7–9 tsunami earthquakes generated by shallow rupture. However, a better characterization of fault coupling is hampered by the poor spatial resolution near the trench axis. Seafloor geodetic measurements are crucial to study the spatial extent of the locked zone and transient slip events on the subduction megathrust.
Furthermore, my student, I, and scientists from the Earth Observatory of Singapore use the stress perturbation imparted by the 2016 earthquake sequence in Kumamoto, Japan, studying the rheology of lithosphere. We also used a similar technique to constrain the rheological properties beneath the Taiwan orogenic belt using the stress perturbation following the 1999 Chi-Chi earthquake and fourteen-year postseismic geodetic observations. The evolution of stress and strain rate in the lower crust is best explained by a power-law Burgers rheology with rapid increases in effective viscosities from ~1017 to ~1019 Pa s within a year. The short-term modulation of the lower-crustal strength during the seismic cycle may alter the energy budget of mountain building. Incorporating the laboratory data and associated uncertainties, inferred thermal gradients suggest an eastward increase from 19.5±2.5°C/km in the Coastal Plain to 32±3°C/km in the Central Range. Geodetic observations may bridge the gap between laboratory and lithospheric scales to investigate crustal rheology and tectonic evolution.
Rock uplift on the Earth surface is a key observation for studies of tectonics and geodynamic processes. Taiwan mountain belt is subject to a rapid uplift rate of 20 mm/yr as revealed by GNSS and leveling measurements in previous studies. The extremely high rates are about two to four times larger than the geological vertical rates inferred from different techniques and time scales. Probing discrepancies between geodetic and geological data is a fundamental question in order to understand the spatial and temporal deformation patterns associated with various tectonic processes. Our study shows the geodetic vertical rates measured over the decadal-scale may not necessary reflect interseismic velocity but rather contain signals from surface processes, groundwater discharge, long-lasting postseismic deformation, and slow-moving landslides.
Furthermore, I have found landslide-induced long-term, short-term, and seasonal motions in the Central Range of Taiwan. The association of landslide motions with high precipitation indicates that near-surface groundwater flow plays an important role in the initiation and acceleration of sliding processes. Studying fault slip during landslides may provide important clues to the common underlying physical processes between landslides and tectonic faulting. In addition, the seismic and aseismic signals emerge from landslides have a higher signal-to-noise ratio compared to those from deep faulting. If precursory signals do exist, there is a better chance to detect seismic and geodetic signatures prior to and during sliding processes. Modeling landslide motions would advance our understanding on the physical processes associated with landslides triggering mechanisms and natural faulting at greater depths.
I have analyzed borehole strainmeter data collected by Sacks-Evertson borehole strainmeters in eastern Taiwan and made great progress in this field by extending the investigation with a quantitative analysis in order to isolate tectonic source effects. We show that the volumetric strain responses to barometric pressure and groundwater level are -1 to -3 nε/hPa and -0.3 to -1nε/hPa, respectively, consistent with theoretical estimates obtained using the Hooke’s law. Large precipitation causes significant contractional strain changes followed by a slow recovery with the amplitude of about -5.1 nε/hPa.