Abstract: After 2004 Sumatra earthquake, different authors obtain the fault slip models from different observation data, such as seismic waves and GPS data. However, the inversion of the co-seismic model from observed co-seismic displacements can not use far-field data as the constraint while adopting half-space medium, and then the inverted model is difficult in explaining far-field data. In this paper, based on the fault geometry model and GPS data provided by Hoechner et al., we inverted the slip model of 2004 Sumatra earthquake by using the inversion algorithm based on spherical dislocation theory. We found that the moment magnitude of the fault model is 9.24, and the maximum slip reach 30.4 m, calculated co-seismic displacements of our fault model agree well with far-field GPS observations after considering the effect of earth curvature. Then, in order to study the effect of 2004 Sumatra earthquake on crustal movement in Southern China, we select two periods of GPS velocity fields (2001—2004 and 2004—2007). From the differences of them, we can see Southern China block has a significant trend of southwest movement after this earthquake, which implied South China block has been affected by the viscoelastic relaxation of this earthquake significantly. Using the viscoelastic dislocation theory of Tanaka, we simulated the differences of two periods of GPS velocity fields in south China block based on our fault model. We found that the optimal viscosity of this area is about 2×10¹⁹ Pa·s when taking the effect of viscoelastic relaxation into account, the RMS (root mean square) of the differences reduced from 3.2 mm to 1.9 mm. In studying crust deformation before and after the 2004 Sumatra earthquake, the viscoelastic relaxation of this earthquake should be considered when based on south China block continuously.

Key words: Viscoelastic relaxation, The 2004 Sumatra earthquake, GPS horizontal velocity fields, South China block