The Siling Co Lake is the largest endorheic lake in Central Tibet. Altimetric measures, combined with lake contours, show that in 1972-1999 its water level remained stable, while it increased by about 1.0 m/yr in the period [2000][2001][2002][2003][2004][2005][2006]. The increased rate gradually stepped down to 0.2 m/yr in 2007-2011. The ground motion associated with the water load increase is studied by interferometric synthetic aperture radar (InSAR) using 107 ERS and Envisat SAR images during the period 1992-2011. The deformation amplitude closely follows the lake level temporal evolution, except that subsidence continues in 2008-2011, while the lake level stagnated. This temporal evolution suggests a non elastic relaxation process taking place at a decade timescale. Phase delay maps are used to constrain possible layered viscoelastic rheological models. An elastic model could partly explain the observed subsidence rate if elastic moduli are about twice lower than those extracted from V P /V S profiles. The surface deformation pattern is also extracted by projecting the phase delay maps against the best fit model temporal behavior. It shows that deep relaxation in the asthenosphere is negligible at the decade timescale and favors the existence of a ductile channel in the deep crust above a more rigid mantle. Overall, the best fit model includes a ductile lower crust, with a viscosity of 1-3 × 10 18 Pa s between 25 and 35 km and the Moho (at 65 km), overlying a rigid mantle.

International audienc

After large earthquakes, parts of the fault continue to slip for days to months during the afterslip phase, a behaviour documented for many earthquakes. Yet, little is known about the early stage, i.e., from minutes to hours after the mainshock. Its detailed study requires continuous high-rate position time series close to the fault, and advanced signal processing to accurately extract the surface displacements. Here, we use refined kinematic precise point positioning processing to document the early postseismic deformation for three earthquakes along the South American subduction zone (2010 Mw8.8 Maule, Chile; 2015 Mw8.3 Illapel, Chile; 2016 Mw7.6 Pedernales, Ecuador). First, we show that early afterslip generates significant surface displacement as early as a few tens of minutes after the earthquake. Our analysis of the time series indicates that, over the first 36 hours, more than half of the displacement occurs within the first 12 hours, a time window often disregarded with daily positioning. Thus, estimates of coseismic offsets can be biased by more than 10% if early postseismic displacements are acknowledged as coseismic ones. Finally, these results highlight the difficulty to accurately evaluate the different contribution to the seismic cycle budget and thus the associated hazard on faults.

No abstract