40th Workshop of the International School of Geophysics Properties and Processes of Crustal Fault Zones Erice, Sicily (IT), May 2013 A source-controlled experiment at Mt. Etna to investigate the origin of wavefield polarization in fault zones G. Di Giulio 1, A. Rovelli 2, F. Cara 2, P. P. Bruno 3, M. Punzo 4 and F. Varriale 5 1) Istituto Nazionale di Geofisica e Vulcanologia, Sezione Roma 1, Sismologia e Tettonofisica,Via Arcivescovado 8, L'Aquila, Italy. 2) Istituto Nazionale di Geofisica e Vulcanologia, Sezione Roma 1, Sismologia e Tettonofisica, Via di Vigna Murata 605, Roma, Italy. 3) Istituto Nazionale di Geofisica e Vulcanologia, Osservatorio Vesuviano, Via Diocleziano, Napoli, Italy. 4) Consiglio Nazionale delle Ricerche, IAMC, Calata di Porta Massa, Napoli, Italy. 5) AMRA, Via Nuova Agnano 11, Napoli, Italy. 1. Abstract In the damage zone of the Pernicana Fault at Mt. Etna, Italy, ground motion is strongly polarized in the horizontal plane. A large number of ambient noise measurements performed in the area, provides H/V spectral ratios that exceed a factor of 10 around 1 Hz near the fault trace and decrease away from the fault. Across the fault, polarization azimuth varies approximately by 30°. In principle, both locally radiated energy due to fault creeping and site/propagation effect could be invoked to explain observations. Earthquake recordings in the fault zone show a similar behavior suggesting a fault-related site effect, as the polarization direction is independent of azimuth, distance and depth of different sources. In order to investigate the directional amplification mechanism, we performed a source-controlled experiment using a high-resolution Vibroseis machine (Ivi-MiniVib®). The machine was operating about fifty meters from the fault scarp, where the natural site polarization was oriented along a N150°/N330° direction. Ground motion was produced through a iron plate vibrating in the horizontal plane with fixed azimuths of motion, parallel and transversal to the observed site polarization. Three-component seismological stations were installed along profiles in the two orthogonal directions with a 50-m spacing. When shear excitation is parallel to the site polarization, the seismic signal propagates efficiently maintaining the same horizontal polarization of the source and is well recorded up to distances as large as 300 m. When shear excitation is orthogonal to the site polarization, the ground excitation looses the initial source polarization in less than 50 m. At larger distances, transmitted energy propagates with the natural site polarization independently of the source polarization. 2. Polarization in the Piano Pernicana Fault Zone 3. Vibroseis Experiment 4. Results The Pernicana Fault System (PFS) is a roughly EW oriented strike-slipe active fault that extends from the NE Rift of Mt Etna Volcano to the coastline with a lenght of about 18 km (Fig. 1). The PFS corresponds to the northern boundary of the unstable sector of the eastern flank of the volcanic edifice (Fig. 1). This flank shows a predominant left- lateral down-slope motion (Neri et al., 2004; Puglisi and Acocella, ). A B Fig. 1. A) (top) Geology of Mount Etna, redrawn from Acocella and Neri [2005]. (bottom) Lidar image showing the topography map of the PFS. Color is proportional to the elevation (from blue to red). The black rectangle indicates the area of Piano Pernicana where we concentrated the seismic experiments. B) Displacement vectors (black arrows) and height variations (colour scale) between July 2005 and June 2006, redrawn from Puglisi and Acocella (project DPC-INGV V4-Flank). We observed a directional resonance around 1 Hz of the horizontal components of the ground motion in the most damaged part of the PFS (Piano Pernicana). This strong polarization (Fig. 2) was observed in recordings of both ambient noise and local eartquakes suggesting that horizontal polarization is the effect of local fault properties (Di Giulio et al., 2009). Ground-motion records in the damage zone of the PFS also show strong variations of directional resonance in the horizontal plane, with an abrupt rotation of azimuth by about 30° across the fault, varying from N166° on the north side to N139° on the south (Fig. 2). Pischiutta et al. (2013) interpreted the observed directional resonance in terms of changes in the kinematic and deformation fields on the opposite sides of the fault. The northern side is primarily affected by the left-lateral strike-slip movement, whereas the southern side, that is subjected also to sliding, is under a dominant extensional stress regime. Brittle deformation models based on the observed kinematic field predict a near-orthogonal relation between the dominant fracture orientation and the azimuth of the observed directional resonance. Pischiutta et al. (2013) indicated that the directional amplification is related to the stiffness anisotropy of the fault damage zone, with larger seismic motions normal to the fractures. Fig. 2. (top) Contouring plots showing the geometric mean of horizontal-to-vertical spectral ratios (HVSR) as a function of frequency (x axis) and direction of motion (y axis) for seismic noise recordings (from Di Giulio et al., 2009). At two sites (#5 and #10) the individual curves for each rotated horizontal-to-vertical spectral ratio are also shown. The rose diagrams indicate the results from polarization analysis. (bottom) Figure showing the distribution of the polarization of the ambient noise along two transects (AA' and BB'): we adopted a colour scale normalized to unity (in red and blue to the north and south of the fault, respectively. Redrawn from Pischiutta et al., 2013). We performed a seismic experiment in the PFS (at Piano Pernicana), using a mobile shaker truck as active source and deploying two linear arrays of three-components seismic stations (Fig. 3). We deployed the seismic stations at different distances from the source (from 10 to 300 mt) in two orthogonal directions: parallel and orthogonal to the observed 1-Hz polarization of ground motion. The shaker truck produced harmonic shear vibrations in the horizontal plane from 5 to 40 Hz as well as sweep signals. This active source was polarized orthogonally to the two corresponding array layouts (Fig. 3). Fig. 3. (top) Picture of the shaker truck (Vibroseis machine) used as source for the seismic experiment. (bottom) Map showing the two linear array layouts used in the experiment. The two linear arrays were deployed along direction N240° and N330°, and the source was polarized orthogonally to the layouts orientation. The seismic stations were equipped with 24-bit Reftek 130 digitizers and Lennartz velocimetric sensors (eigenfrequency 0.2 Hz). Array Layout N240 Array Layout N330 North Pol N330 PolN240 Fig. 4. Left-hand side: time-frequency analysis on raw data. Right-hand side: recordings (band-pass filtered around 5-Hz) of two seismic stations of the array layout N330. The stations are 10 and 300 m distant from the Vibroseis source. The 5-Hz signal is evident also for the station 300 m distant from the source. We analyzed the 5-Hz signal produced by the Vibroseis machine because this frequency is the closest to ''naturally polarized'' frequency of 1-Hz, ie observed on site by analysis of ground motion recordings (Di Giulio et al., 2009; Pischiutta et al., 2013). The 5-Hz frequency was excited several times during the experiment and was well- recorded by the stations of the the two linear arrays. Fig. 4 shows an example of recordings of the closest and farthest station to the Vibroseis source (10 and 300 mt, respectively). The particle motion of the signal in the horizontal plane shows that the source polarization produced by the Vibroseis machine is well consistent along the stations when the shear excitation is parallel to the site polarization (Figure 5). When shear excitation is orthogonal to the site polarization, the ground excitation looses the initial source polarization in about 50 m (Figure 5). Then, the polarization of signal is similar to the one observed on ambient vibrations. Fig. 5. Particle motion analysis in the horizontal (NS and EW) plane. We consider a window of seismic noise applying a low-pass filter at 2 Hz (columns A and D), the same noise window with a band-pass filter around 5 Hz (columns B and E), and a window including the harmonic 5 Hz signal produced by Vibroseis after a band-pass filter around 5 Hz (columns C and F). Columns A,B, and C refer to the N240 oriented array; coloumns D, E and F refer to the N330 oriented array. The numbers on the left indicate the distance of the seismic stations from the Vibroseis source. A B C D E F Noise lp 2 HzNoise bp 5 HzSignal bp 5 Hz Noise lp 2 Hz Noise bp 5 Hz Signal bp 5 Hz Array layout N240, Source Polarization N330 Array layout N330, Source Polarization N m 50 m 100 m 150 m 200 m 250 m 300 m 10 m 50 m 100 m 150 m 200 m 250 m 300 m -63 m Fig. 6. (top) Energy ratio of the radial and transverse components assuming as reference energy the one of the nearest seismic station (distant 10 m from the Vibroseis machine). The ratio was computed on a time-window (band-pass filtered at 5 Hz) assuming both noise and signal produced by Vibroseis. The red symbols indicate the radial horizontal component (radial to the predominant site polarization). The green symbols indicate the transverse horizontal component (transverse to the predominant site polarization). Array layout N330, Source Polarization N240Array layout N240, Source Polarization N330 Noise bp 5 Hz Signal bp 5 Hz 5. Conclusion Recent studies evidenced a strong polarization of the ground motion in the most damaged zone of the Pernicana Fault System (at Piano Pernicana). In this study we describe an experiment in the same area using a Vibroseis machine as active source. The results show that the shear polarization induced into the ground is lost in less than 50 m when the excitation is polarized with an azimuth different than the natural site polarization. At distance larger than 50 m, the polarization is consistent with the natural one. Whereas the stations of the N240° oriented array show a similar polarization, the stations of the N330° oriented array show a more chaotic polarization at distances greater than 150 m from the Vibroseis source, suggesting that these stations are probably out of the most damaged fault zone. These experimental results suggest a shallow origin of the wavefield polarization in fault zones, probably due to scattering and mode conversion related to heterogeneities of the predominantly oriented fractures. References Acocella, V. and M. Neri, Structural features of an active strike-slip fault on the sliding flank of Etna (Italy), J. Struct. Geol., 27(2), 343–355. Di Giulio, G., Cara, F., Rovelli, A., Lombardo, G. and Rigano, R., Evidences for strong directional resonances in intensely deformed zones of the Pernicana fault, Mount Etna, Italy, J. Geophys. Res., 114, doi: /2009JB Neri, M., Acocella, V. and Behncke, B., The role of the Pernicana Fault System in the spreading of Mt. Etna (Italy) during the 2002–2003 eruption, Bull. Volcanol., 66, 417–430. Pischiutta, M., Rovelli, A., Salvini, F., Di Giulio G., and Y. Ben-Zion, Directional resonance across the Pernicana Fault, Mt Etna, in relation to brittle deformation fields, Geophys. J. Int., 193 (2), doi: /gji/ggt031. Puglisi G. and V. Acocella, Hazards related to the flank dynamics at the Mt. Etna, , (Project DPC-INGV V4 -Flank – in: Comp. Radial Distance 300m Comp. Transverse Distance 300m Comp. Z Distance 300m Comp. Transverse Distance 10m Comp. Radial Distance 10m Comp. Z Distance 10m Time counts radial transverse radial Noise bp 5 Hz Signal bp 5 Hz We compared the energy of the horizontal components taking into account both noise and the Vibroseis signal band-pass filtered at 5-Hz and normalized to the energy of the station closest to the source (Fig. 6). As horizontal components, we considered the directions radial and transverse to the predominant site polarization (see Fig. 3). The distribution of energy as a function of distance from the source (Fig. 6) shows a strong decay of the signal within 100 m. Stations of N240- oriented array show, at larger distances from the source, more energy for the radial (natural polarization) than for the transverse component. The N330- oriented array shows, for stations at distance greater than 150 m from the source, a similar level of energy for radial and transverse components.