Adding geodetic strain rate data to a seismogenic context

In several seismically active areas deformation processes at depth must generate deformation at the surface, and the measurement of such surface deformation is an important boundary condition for models of the evolution of interacting blocks before, during and after earthquakes. The network of some 160 permanent GPS stations disseminated in Europe under the European Permanent Network of EUREF, with additional densification stations in particular areas such as the north east of Italy, provides a valuable contribution to the estimate of the average surface strain rate. The expected strain rate is of the order of 20-40 nanostrain per year, corresponding to a velocity change of a few mm/year over distances of some hundreds of kilometers. Consequently, we must require accuracy in the velocities of fractions of mm/year, and full control of systematic errors which may mask tectonic signals. The procedures for the systematic processing of SINEX files, representing the densified network, are reviewed here with the intent of meeting and possibly exceeding such specifications. A method for determining the noise in time series of coordinates, and of obtaining a reliable estimate of the accuracy in the estimated station velocities is described. In particular, it is shown that, on average, at least three years of continuous tracking of a permanent GPS station are required for a reliable estimate of its velocity. Then the problem of calculation of the velocity field and its horizontal gradient is addressed. We focus on the algorithm of weighted least squares collocation as a technique of minimum variance to interpolate velocities and strain rates. We present the large scale velocity flow across most of continental Europe, after subtraction of a rigid rotation approximating the generalised NE drift of Eurasia, showing a variety of intraplate and interplate processes. Finally we review the frictional model of Anderson to describe fault interaction and stress release, and present analytical expressions for recurrence times of fault instabilities. This simple framework enables a number of key problems to be identified to make proper use of the geodetically inferred strain rate data. Taking the seismicity in Friuli as a test bed, we discuss requirements on the knowledge of fault geometries, local rheology, fault plane solutions, role of pore fluid pressure and historical seismicity which, in conjunction with the surface geodetic data, are necessary to attempt a more advanced modelling of the dynamic and potentially seismogenic processes at depth.