GNGTS 2019 - Atti del 38° Convegno Nazionale

572 GNGTS 2019 S essione 3.1 scheme devised by De Siena et al. (2017) to obtain absorption maps using the 2-D analytic kernels defined first by Del Pezzo et al. (2016) for each source and receiver pair. The scheme is valid in a diffusive, highly heterogeneous medium; it allows to map the effective sensitivity of the Q c parameters in space and build the rows of the inversion matrix after normalization for the total sensitivity in space. A first-order Tikhonov inversion has been performed to find the inverse coda quality factor at each node of the grid, which has a regular step of 0.5° by 0.5° of latitude and longitude, and at each frequency. The scheme allows to perform checkerboard tests, which discriminate the areas used for the final interpretation (Fig. 2). Results. Before interpreting the results and comparing them with those obtained with standard tomographic techniques, we discuss the biases affecting coda wave imaging in this region. This kernel-based inversion scheme has been tested using local seismicity, while in this study the epicentral distances are greater than 400 km. The effects will bias the estimation of coda attenuation are: (1) the presence of phases of waves reflected from mantle, (2) the break of the uniform half-space approximation caused by the thin oceanic crust. Our discussion thus grounds on the hypothesis that Q c –1 maps are especially sensitive to Moho depth. At present, the evaluation of these effects is still qualitatively. The obtained results (Fig. 3) are consistent with the main attenuation difference expected between oceanic and continental crust and coincides with outer front of the thinned Tyrrhenian- Apennine crust. High attenuation anomaly in central Italy could be related to the upwelling of fluids, probably caused by the slab break off (Wortel and Spakman, 2000). Between the Sicilian and Calabrian coast, we retrieve negative values of Q c –1 , which are not physically possible, and that could represent the effects of an interface with a strong impedance contrast, reflecting energy across the coda-intervals. Fig. 2 - The checkerboard test: the inversion (right) of syntetic data (left) shows the effective resolution of our map at central frequency 1Hz.