Search

We develop a simple tomographic approach exploiting the decay rate of coda waves to map the absorption properties of the crust in a region delimited approximately by the Rhine Graben to the North, the Apennines to the South, the Massif Central to the West and the Dinarides to the East. Our dataset comprises 40 000 coda records of about 2000 weak to moderate crustal earthquakes, with magnitude ranging from 2.8 to 6 and recorded by broad-band, accelerometric and short-period stations. After proper choice of a coda window minimizing the effects of variable epicentral distances, we measure the coda quality factor Q c in five non-overlapping frequency windows covering the 1–32 Hz band for all available source station pairs. These measurements are subsequently converted into maps of absorption quality factor ( Q i ) using a linearized, approximate relation between Q c and Q i . In practice the following procedure is applied in each frequency band: (1) we divide the target region into 40 × 40 km cells; (2) for each source-station pair, we assign the measured Q c value to each pixel intercepted by the direct ray path; (3) the results are averaged over all paths and subsequently smoothed with a 3 × 3 pixels moving window. Our approach is consistent with the high sensitivity of Q c to the value of Q i between source and station. Our tomographic approach reveals strong lateral variations of absorption with length scales ranging from 100 km to 1000 km. At low frequency ( ∼ 1 Hz ), the correlation with the surface geology is clear, Cenozoic and Mesozoic sedimentary basins (resp. crystalline massifs) being recognized as high (resp. low)-absorption regions. Furthermore the Q i map delineates finer geological features such as the Ivrea Body, the Rhône Valley, or felsic intrusions in the central Alps. At high-frequency (>16 Hz), only the thickest Cenozoic sedimentary deposits show up as high-attenuation regions and a north/south dichotomy is apparent in the absorption structure. The limit between low-attenuation regions to the North and high-attenuation region to the South correlates geographically with the location of the Periadriatic Lineament (PL), a major late-alpine crustal- to lithospheric-scale structure. Furthermore, the attenuation structure seems to prolong the PL to the West along a line marked by large historical earthquakes. The Apennines orogenic belts exhibit a distinct frequency behavior, with high attenuation at low-frequency and low-attenuation at high-frequency. Low-frequency absorption may likely be explained by the relatively thick cover of Cenozoic sedimentary materials, as well as by shallow geothermal activity. We hypothesize that the frequency dependence of the attenuation structure, in particular in the Apennines, is caused by a change of the wavefield composition which accentuates the sensitivity of the coda to the deeper parts of the medium as the frequency increases. [ABSTRACT FROM AUTHOR]

Coda quality factor (coda-Q) is used to estimate the attenuation and scattering properties of the Earth (Aki & Chouet 1975). So far focus has been on earthquake data at frequencies above 1 Hz, as the high noise level in the first and second microseismic peak, and possibly lower levels of scattering, hinder stable measurements at lower frequencies. In this work, we measure and map coda-Q in the period bands 2.5 s - 5 s and 5 s - 10 s in the greater Alpine region using noise cross-correlations using permanent seismic stations and data from the temporary AlpArray experiment. We demonstrate that it is possible to measure coda-Q using noise cross-correlations, in spite of a relatively high noise level of the correlations. When severe quality selection criteria are applied, the resulting measurement is independent of azimuth and distance so there is no sign of influence of the directivity of the noise field. At longer periods than 10 s, the dispersion of the data is too high to obtain reliable maps. Mapping shows persistent low-coda-Q anomalies in the Po plain and north-east of the Alps and high-coda-Q anomalies north of the Adriatic Sea and in southeastern France, however the shape of the correlations leads to a long coda window that doesn't allow us to discriminate between early and late coda, so we cannot distinguish between intrinsic and scattering Q. We additionally compared cross-correlations stacked over one year of data, with those of four years of data, for the stations pairs where such a long times series was available. This comparison demonstrates that the convergence of the coda-Q measurement is slow, but any bias is systematically towards higher coda-Q. Therefore, high coda-Q values may be estimated, but the geographical distribution between high and low coda-Q areas is respected. [ABSTRACT FROM AUTHOR]