GNGTS 2016 - Atti del 35° Convegno Nazionale

218 GNGTS 2016 S essione 1.2 From GPS velocities to strain-rate: methodology. We estimate a continuous 2D strain- rate field starting from scattered GPS velocities. To compare geophysical and GPS inferred stress we need to estimate geodetic strain-rates at the same locations (P i ) as boreholes and focal mechanisms. For this purpose it is necessary that a sufficiently large number of GPS stations are available to compute a horizontal gradient in the neighborhood of each P i point within a correlation distance d 0 of the order of some tens of kilometers. The concept of correlation distance for the computation of a strain rate is conveniently embodied into the algorithm of least squares collocation. Least squares collocation is an autoregressive algorithm where the velocity interpolated at a point P is expressed as a weighted average of the velocities measured at points i (i=1..n), with weight function decreasing with the distance from P to the i-th GPS site. To decrease the weight of spurious data, perhaps affected by local disturbances, we apply a low pass filter that results in smoothing the estimated velocity field. The horizontal gradient of the computed velocities at P i follows from their differentiation in the north and east directions. The 2D strain rate is the symmetrical part of the velocity gradient. The eigenvalues and eigenvectors of the strain rate tensor are computed by matrix diagonalization, yielding a maximum/minimum strain rate (typically the most extensional/ compressional, corresponding to a positive/negative horizontal derivative of the velocity field), and the azimuth ϑ of the most extensional strain rate. Hence ϑ corresponds to S h . We compute one correlation distance d 0 for each point P i , by analyzing the correlation profile of the local GPS site velocities. For each point we compute the shear strain rate at increasing values of the correlation distance, to find the value of d 0 which maximizes the shear strain rate. This ensures that all the correlated velocities contribute to the computation of the strain rate tensor. We find correlation distances ranging from 50 to 120 km with a mean value of 53 km. Results. We focus on the difference in azimuth of minimum horizontal stresses (geodetic Sh min - geophysical Sh min ) between each borehole or focal mechanism measurement, and the geodetically inferred direction. We also consider equivalence classes of seismic province Fig. 2 – Eigenvectors of the strain rate tensor inferred from GNSS velocities (green arrows) are interpolated to the center of those 87 Individual Seismogenic Sources of DISS 3.2.0 (brown rectangles) sufficiently covered by GNSS data. Extension is in blue and compression in red. Major tectonic lineaments are in orange.