Revealing of surface deformations induced by geodynamic processes in the Kuril island arc from GNSS data

Kuril subduction zone is one of the most active continental margins due to the high plate convergence rate. Latest oceanographical, seismological and geological studies show a block structure of the Kuril island arc. In 2006-2008 Kuril GNSS network was installed along the island arc to provide information on the dynamics of the continental margin. Proper geodetic characterization of surface deformations in Kuril region is necessary for studies of regional geodynamical processes associated with seismic cycles and the evolution of the subduction zone. Since Kuril network has some disadvantages such as small amount of continuous stations (cGNSS) and its near-linear arrangement, special attention must be paid to correct processing of the GNSS data to exclude miscalculations that can affect further modeling of regional geodynamical processes.

We use regression analysis of time series of cGNSS stations displacements to distinguish components which are related to: 1) long-term accumulation of elastic stresses (secular velocity); 2) almost instant release of substantial part of accumulated stresses during main shock (coseismic offsets); 3) transient processes following large subduction eartquakes. The main advantages of the proposed regression analysis algorithm are: 1) an automatic process for detecting coseismic displacements, based on direct modeling of surface deformations using a dislocation model, 2) an automatic process for identifying transient processes; 3) taking into account the realistic GNSS noise model in calculating errors.

Since most of the GNSS stations were deployed only after large 2006-2007 Simushir earthquakes their time series were affected by intense and long-term postseismic transient processes such as afterslip and viscoelastic relaxation in the upper mantle. We use our direct models of these postseismic processes to construct residual time series, which allows us to estimate magnitudes of seasonal periodic signal and to calculate realistic errors.

We use correlation-based clustering algorithm to identify the influence of block structure of island arc on observed deformation patterns during interseismic, coseismic and postseismic stages of the seismic cycle. We also check our processing of GNSS data by constructing model of slip distribution in the source of 2006 Simushir earthquake on the basis of our estimates of coseismic offsets and by comparing our model with previous ones obtained on the basis of satellite geodetic data. The performed analysis of continuous GNSS observations shows that different parts of Kuril island arc are at different stages of seismic cycle.