Özet:

With occurrence of an earthquake, an instant displacement on the Earth’s surface (so-called co-seismic displacement) and subsequently a post-seismic relaxation could be experienced. The post-seismic relaxation is a process that is a stress relaxation in the crust’s low viscosity layers and in the upper mantle. Effects of this time-dependent non-linear transient events on Earth’s surface can be monitored by Global Positioning System (GPS) and can be modelled deterministically through an exponential function using GPS time series. The post-seismic relaxation, which is not (can not be) modelled, affects other parameters in the mathematical model. In this study, the post-seismic relaxation effects on GPS time series error characteristics, velocity, and co-seismic displacements were examined by analyzing pre- and post-earthquake observations series both jointly and separately. The post-seismic relaxation time were optimized by Nelder-Mead simplex algorithm, and GPS error time series were analyzed using these optimized relaxation times and additional deterministic parameters with stochastic model combinations of white noise (WN)+flicker noise (FN), WN+FN+random-walk noise (RWN), WN+power-law noise (PL). The time series jointly analyzed can be characterized by RWN or PL in which spectral index is about to -1.25. On the contrary, all the separately analyzed time series is characterized by the WN+FN model combination. Both quality measures and visual inspections of GPS error time series demonstrates that the separately analyzing is a correct approach. Accordingly, the joint analysis can result in biased velocity estimates up to 0.52 mm/year, overestimation of velocity uncertainties up to %94 and co-seismic displacements differences up to 8 cm.

Anahtar Kelimeler: