Space geodetic observations and models of postseismic deformation due to the 2005 M7.6 Kashmir (Pakistan) earthquake

We use the L-band Advanced Land Observing Satellite (ALOS) and C-band Envisat interferometric synthetic aperture data and campaign GPS observations to study the postseismic deformation due to the 2005 magnitude 7.6 Kashmir (Pakistan) earthquake that occurred in the northwestern Himalaya. Envisat data are available from both the descending and ascending orbits and span a time period of similar to 4.5years immediately following the earthquake (2005-2010), with nearly monthly acquisitions. However, the Envisat data are highly decorrelated due to high topography and snow cover. ALOS data are available from the ascending orbit and span a time period of similar to 2.5years between 2007 and 2009, over which they remain reasonably well correlated. We derive the mean line-of-sight (LOS) postseismic velocity maps in the epicentral area of the Kashmir earthquake using persistent scatterer method for Envisat data and selective stacking for ALOS data. LOS velocities from all data sets indicate an uplift (decrease in radar range), primarily in the hanging wall of the earthquake rupture over the entire period of synthetic aperture radar observations (2005-2010). Models of poroelastic relaxation predict uplift of both the footwall and the hanging wall, while models of viscoelastic relaxation below the brittle-ductile transition predict subsidence (increase in radar range) in both the footwall and the hanging wall. Therefore, the observed pattern of surface velocities indicates that the early several years of postseismic deformation were dominated by afterslip on the fault plane, possibly with a minor contribution from poroelastic rebound. Kinematic inversions of interferometric synthetic aperture radar and GPS data confirm that the observed deformation is consistent with afterslip, primarily downdip of the seismic asperity. To place constraints on the effective viscosity of the ductile substrate in the study area, we subtract the surface deformation predicted by stress-driven afterslip model from the mean LOS velocities and compare the residuals to models of viscoelastic relaxation for a range of assumed viscosities. We show that in order to prevent surface subsidence, the effective viscosity has to be greater than 10(19)Pas. ations are negligible