An automated regionalization of a reduced domain of the Earth's surface, without a-priori seismotectonic information, and the corresponding anelastic structures are achieved by means of a sequential application of a principal component analysis, a clustering procedure and the stochastic inversion algorithm to anelastic attenuation coefficients of Rayleigh waves. Our database is formed by path-averaged attenuation coefficients derived from fundamental mode Rayleigh waves crossing the Iberian Peninsula. The wavetrains were recorded at the broad-band stations installed some years ago in the Iberian Peninsula for the ILIHA project. Before the sequential application of the algorithms, the area covered by the seismic paths was characterized by 22 maps of local attenuation coefficients corresponding to the 10–120 s period range. All these maps were obtained by applying Yanovskaya's formulation for laterally heterogeneous media to the set of path-averaged attenuation coefficients mentioned above. After that, we divided the Iberian Peninsula into six homogeneous regions in terms of the local attenuation coefficients for 22 different periods and by means of the principal component analysis and the clustering algorithm. We then obtained, by stochastic inversion, the respective anelastic structure down to a depth of 225 km for the six homogeneous regions. The coefficients vary from 1.0 to 2.0×10-3 km-1 and the Q-1ß structures, ranging from 110×10-3 to 40×10-3, suggest for five of the regions an asthenosphere with its upper and lower boundaries close to 80 and 180 km depth, respectively. A shallow beginning of the asthenosphere is detected for the sixth homogeneous region. This range is very similar to the depths deduced from previous tomographic studies of the Iberian Peninsula based on Rayleigh wave phase and group velocities. Finally, correlations obtained between the regions and some seismotectonic characteristics of the Peninsula are discussed. It is noteworthy that we cannot associate each of the six homogeneous regions with a single Hercynian, Alpine or Neogene domain.
The Canary Archipelago, near the African plate between 27° and 30° N and 13° and 19° W, is included in the oldest region (> 135 m.y.) of the eastern Atlantic Ocean. The tectonic seismicity is moderate, with low magnitudes, and it is almost totally related to an active fault located between the islands of Tenerife and Gran Canaria. The purpose of this paper is to provide a clear idea of the anelasticity of coda waves in the region. In order to compute master dispersion curves as a function of the quality factor, Q, we have applied inversion methods to obtain analytical magnification curves for all the seismographic stations used. fp-t data, corresponding to the coda wave dispersion of 52 local and regional seismic events recorded at 6 seismographic stations, have been obtained. The fit of the data to the master curves determined yields estimates of coda-Q values and their frequency dependence. Most of the Q0 values determined, at 1 Hz, vary between 100 and 300, showing a strong frequency dependence in the range 0.2–2 Hz. The characteristic v value (degree of frequency dependence of Q) in the region, given by the law Q = Q0 fv is v = 0.9. By applying regionalization techniques an iso-Q0 map, for a reference frequency of 1 Hz, has been obtained for the Canary Archipelago. This map shows important lateral variations in Q0, reflecting the tectonics and the volcanism of the region. The lowest Q0 values are found along the trend of the fault and beneath the islands.
Propagation of 0.2–2.5 s short-period Rayleigh-waves across a part of northwestern Iberia is investigated. Analysis of a seismic profile line with recorded seismograms up to distances of 70 km from the shot-point, provides group-velocities and attenuation coefficients of the fundamental mode. First higher-mode group-velocity has been obtained for only one of the studied cases. The studied region has been divided into two subrogions, the northwestern and the southeastern side from the shot-point. Inversion procedures applied to the velocity and attenuation data yield the shear velocity and the Q−1β models up to a depth of about 1.5 km for the NW, and 2.4 for the SE side. From the shear-wave obtained models corresponding to the NW and to the SE sides, it is possible to infer the existence of some degree of lateral variation of S velocity. The NW side presents velocities varying between 1.8 km · s−1 (at the surface) and 3.3 km · s−1 (at 1.2 km depth). The velocities in the SE model, for the same depths, are 2.2 and 3.4 km · s−1, respectively. Using only the fist six traces of the SE side, it is possible to infer a low-velocity region located between 0.4 km and 0.7 km depth. It seems that the velocities are higher in the complex geological region of the SE side than in the NW one, which is in agreement with P velocity models found by others. From the tentative anelastic attenuation coefficients, Q−1β models have been inferred for both sides of the seismic refraction profile. Although the obtained Q−1β models are somehow different in shape, the maximum Q−1β value is about 0.05 in both cases.
A tentative tomography of Iberia based on broadband surface wave data is performed for the first time. In this paper we have complemented a previous study of the northern half of the Iberian Peninsula with a study of the southern half of the peninsula based on a detailed analysis of Rayleigh wave dispersion. Now we present the first shear velocity contour maps at different subcrustal depths. Quality digital records provided by the NARS array have been used. Standard techniques, as multiple filtering and time-variable filtering, are employed to efficiently remove higher mode interference and improve isolation of the fundamental mode Rayleigh wave from the seismograms, and thus to calculate the interstation Rayleigh wave group velocity. Frequency-domain Wiener deconvolution is used to determine the interstation phase velocity. We perform simultaneous inversion of phase and group velocity dispersion data by means of the stochastic inverse operator. The inversion results previously reported for the northern Iberian region together with the results obtained here are used for 3-D mapping of shear wave velocity over the Iberian Peninsula. Lateral variation in velocity at five depth intervals chosen in agreement with the theoretical earth models determined by joint inversion, is displayed for the lithosphere and asthenosphere of Iberia.
The coda-Q method has been applied to a set of earthquakes located in the Mediterranean region, between the Iberian Peninsula and the Balearic Islands (the Valencia trough), and recorded at the short-period seismographic station of Ebro (EBR), located in Roquetes (Tarragona), to obtain the anelastic quality factor Q of coda waves. The applied method uses a set of master curves taking into account the frequency-dependence of the coda-Q values. To determine the master curves used to obtain the coda-Q values, non-linear inversion techniques have been applied to the short-period instruments. The coda-Q values and the frequency-dependence obtained for the region seem to indicate that the southern part of the Valencia trough tends to attenuate the seismic energy in the same manner as the Betic System, and the northern part seems to present higher Q values than the southern one. This fact may indicate different Q structures in both regions. A good correlation is observed among the obtained coda-Q values, their frequency-dependence, and known coda-Q values and Lg wave attenuation in the coastal region of the Mediterranean Sea.