E. Del Pezzo

Automatic Classification of Seismic Signals at Mt. Vesuvius Volcano, Italy, Using Neural Networks

We present a new strategy for reliable automatic classification of local seismic signals and volcano-tectonic earthquakes (vt). The method is based on a supervised neural network in which a new approach for feature extraction from short period seismic signals is applied. To reduce the number of records required for the analysis we set up a specialized neural classifier, able to distinguish two classes of signals, for each of the selected stations. The neural network architecture is a multilayer perceptron (mlp) with a single hidden layer. Spectral features of the signals and the parameterized attributes of their waveform have been used as input for this network. Feature extraction is done by using both the linear predictor coding technique for computing the spectrograms, and a function of the amplitude for characterizing waveforms. Compared to strategies that use only spectral signatures, the inclusion of properly normalized amplitude features improves the performance of the classifiers, and allows the network to better generalize. To train the mlp network we compared the performance of the quasi-Newton algorithm with the scaled conjugate gradient method. We found that the scaled conjugate gradient approach is the faster of the two, with quite equally good performance. Our method was tested on a dataset recorded by four selected stations of the Mt. Vesuvius monitoring network, for the discrimination of low magnitude vt events and transient signals caused by either artificial (quarry blasts, underwater explosions) and natural (thunder) sources. In this test application we obtained 100% correct classification for one of the possible pairs of signal types (vt versus quarry blasts). Because this method was developed independently of this particular discrimination task, it can be applied to a broad range of other applications.

The role of hydrothermal fluids in triggering the July–August 2000 seismic swarm at Campi Flegrei, Italy: evidence from seismological and mesostructural data

Abstract This study presents a detailed analysis and interpretation of the seismicity that occurred on July 2–7 and August 22, 2000, during a ground uplift episode which started on March 2000 at Solfatara crater, Campi Flegrei. Earthquakes are located using a probabilistic grid-search procedure acting on a 3-D heterogeneous earth structure. The mainshock of the July swarm depicts a spectrum characterized by a few narrow peaks spanning the 1–5-Hz frequency band. For this event, we hypothesize a direct involvement of magmatic fluids in the source process. Conversely, the spectra of the August events are typical of shear failure. For these latter events, we evaluate the source properties from P- and S-wave displacement spectra. Results for the most energetic shocks (M d around 2) yield a source radius in the order of 100 m and stress drop around 10 bars, in agreement with most of the earthquakes that occurred during the 1982–1984 bradyseismic crises. For the August swarm we identify two clusters of similar earthquakes. Application of high-resolution relative location techniques to these events allows for the recognition of two parallel alignments trending NE–SW. The relationship among source dimension and relative location evidences overlapping of sources. This may be interpreted in terms of either a heterogeneous stress field or a lubrication process acting over the fault surface. For a selected subset of the August events, we also analyze the splitting of the shear waves: results are indicative of wave propagation through a densely fractured medium characterized by a distribution of cracks oriented NE–SW. The pattern of faulting suggested by relative locations and shear-wave splitting is not consistent with the surface trace of NW–SE striking faults. However, a detailed mesostructural analysis carried out over the Solfatara area indicated the occurrence of two main crack systems striking NW–SE and NE–SW. This latter system shows a strike consistent with that derived from seismic evidence. Results from a stress analysis of the crack systems indicate that a fluid overpressure within the NW–SE-striking faults is able to form NE–SW cracks. We found that the pressure of fluids P f required to activate the NW–SE faults is less than σ Hmin , while the P f value required to open the NE–SW cracks is higher than σ Hmax . Our main conclusions are: (a) the Solfatara area is affected by two orthogonal fracture systems, and the fluid pathway during the 2000 crisis mainly occurred along the NNE–SSW/NE–SW-striking crack system; (b) the July seismicity is associated to the upward migration of a pressure front triggered by an excess of fluid pressure from a small-size magmatic intrusion; conversely, the August events are associated to the brittle readjustment of the inflated system occurring along some lubricated structures.

Array tracking of the volcanic tremor source at Deception Island, Antarctica

We have found experimental evidence which shows that the volcanic tremor recorded at Deception Island (South Shetland Islands, Antarctica) is a superposition in time of overlapping hybrid events. We studied data from a small aperture seismic array. Data analysis for tremor and hybrids included: (1) spectral analysis; (2) apparent slowness and back-azimuth determination by using the zero-lag cross-correlation method; and (3) polarization analysis. Both types of events share these common features: (a) two dominant spectral bands at frequencies 1–3 Hz (the most energetic) and 4–8 Hz; (b) several coherent phases with the same back-azimuth to the source and apparent slowness along the whole signal; (c) in the high frequency band, the apparent slowness is very low (around 0.17 s/km), indicating the propagation of body waves; (d) in the low frequency band, the apparent slowness is high (around 1.6 s/km), consistent with the presence of surface waves; and (e) clear P-wave onset followed by a complex pattern of Rayleigh waves. Therefore, both types of events are strongly related because they share the same source region, the same wave-propagation properties, and the same wave composition. Moreover, several arrivals, that resemble a single hybrid event, have been found along the tremor signals. Due to these reasons, we hypothesize that volcanic tremor of Deception Island is a superposition of hybrid type events. The source of both types could be the interaction between thaw water and hot materials in a shallow aquifer.

The 1998-1999 seismic series at Deception Island volcano, Antarctica

Abstract During the 1998–1999 Antarctic summer the pattern of seismic activity at Deception Island volcano changed significantly. The change was characterized by the occurrence of an intense swarm of volcano–tectonic (VT) earthquakes. More than 2000 VT earthquakes with S–P times smaller than 4 s were recorded in the period January–February 1999. Pure volcanic events were also detected; especially long-period (LP) events, volcanic tremor and some hybrid events. Seismic monitoring was performed using two short-period small-aperture arrays, among other instruments. Based on their signal-to-noise ratios we selected 863 VT earthquakes, 350 LP events and tremor episodes, and 9 hybrid events for analysis. We estimated apparent slowness and back-azimuth for all events using the Zero Lag Cross-Correlation array technique. Combining this information with S–P times and other indirect evidence, we identified two different source regions. LP seismicity is located less than 1–1.5 km southwest of the Fumarole array site. These events are likely to have a hydrothermal origin. VT earthquakes and hybrid events are located at depths of 0.3–10 km in an area under the bay of Deception Island. The area extends from the Fumarole array to the northeast with epicentral distances that range from 0.5 to 12 km. Most hypocenters are clustered in a small volume of around 8 km3. The sources of the LP seismicity and the VT earthquakes are spatially distinct, which indicates that they are not produced by the same mechanisms. Moment magnitude analyses of the VT earthquakes provide an average magnitude of 0.5 and very low average stress drop, around 1 bar. A study of first motion of the P-waves suggests that the events in this small source region should have a variety of source mechanisms. This is supported by the existence of families of events with the same waveforms. The occurrence of repeating fracture processes with low stress drop and small fault dimensions can be explained by the lubrication of pre-existing zones of weakness by pressurized fluids. The most probable hypothesis that explains the generation of this seismic series at Deception Island is: a seismic series caused by the stress generated by the uplift of the source area due to a magmatic injection in depth. We favor this hypothesis since it is compatible with the majority of the characteristics of the seismicity and explains the spatial and temporal behavior of the series.

Qc of three component seismograms of volcanic microearthquakes at Campi Flegrei volcanic area — Southern Italy

Digital recordings of three component microearthquake codas from shallow seismic events in the volcanic region of Campi Flegrei — Southern Italy — were used with an automatic technique to calculate the attenuation factorQc (codaQ) in the hypothesis of singleS toS backscattering.Results show the same value ofQ for each of the three components. This result is interpreted as due to isotropicS wave radiation pattern.A check of the coda method was performed using a single station method based on simple assumptions on the direct SH wave spectrum. Single stationQ was averaged over the stations and over the earthquakes. Results show that the two methods lead to comparable results.A frequency dependence quite different from that evaluated in active tectonic regions was found for coda attenuation, comparable to other volcanic areas throughout the world. This is interpreted as due to the presence of magma that affects anelasticity and scattering.

Shallow structure of Mt. Vesuvius Volcano, Italy, from seismic array analysis

Data from a portable dense seismic array deployed on Mt. Vesuvius in May 1994, during a 2D seismic tomography experiment, are analyzed in the present paper. The array consisted of two groups of short period geophones, 4.5 hz natural frequency, formed by 16 and 25 vertical components (plus two horizontal components), distributed along an arc like shape along the summit crater. Stacks of later arrivals, interpreted as reflected phases, provide a significant constraint to a boundary layer located in the depth range 1.5–2.2 km beneath the summit crater with average velocity V=1.8–2.2 km/s, interpreted as the top of the limestone basement. The correlation methods applied on microtremor records allowed to infer the shallow velocity structure, up to 400 m, beneath the crater rim.

A double seismic antenna experiment at teide Volcano: existence of local seismicity and lack of evidences of Volcanic tremor

Data analyzed in the present work correspond to a 40 days field experiment carried out in Teide Volcano (Canary Islands, Spain) with two short-period small-aperture dense seismic antennas in 1994. The objective of this experiment was to detect, analyze and locate the local seismicity. We analyzed also the background seismic noise to investigate the possible presence of volcanic tremor. From a set of 76 events, we selected 21 of them in base of their good signal-to-noise ratio and their possibility to locate their seismic source by using the seismic antennas. A visual classification based on the S‐P time and seismogram shape has permitted to establish three groups of events: local seismicity (S‐P time between 3 and 5 s), very local earthquakes (S‐P time smaller than 3 s) and artificial explosions. These earthquakes have been located by applying the Zero Lag CrossCorrelation technique and the inverse ray-tracing procedure. Those earthquakes that were recorded simultaneously by both seismic antennas were also located by intersecting both back-azimuths. The analysis of the seismicity has revealed that the amount of seismicity in Teide Volcano is moderate. This seismicity could be distributed in three main areas: inside the Caldera Edifice (below the Teide‐Pico Viejo complex), in the eastern border of the Caldera Edifice and offshore of the island. At present, this activity is the only indicator of the volcano dynamics. The analysis of the back-ground seismic noise has revealed that at frequencies lower than 2 Hz, the Oceanic Load signal is predominant over other signals, even over local earthquakes with a magnitude of 2.0. Due to this, although if in the Teide area were present a weak volcanic tremor, or other volcanic signals with predominant peaks below 2 Hz, to observe them would be a very difficult task. q 2000 Elsevier Science B.V. All rights reserved.

Slowness Anomalies from Two Dense Seismic Arrays at Deception Island Volcano, Antarctica

In this article, we analyze the data collected by two short-period seismic arrays deployed at Deception volcano, Southern Shetland Islands, Antarctica. The field survey was conducted during the 1998-1999 austral summer and was aimed at a quantitative assessment of the complex wave fields associated with the magmatic and hydrothermal activity of the volcano. The two arrays had apertures of 320 m and 240 m and were separated by a distance of about 3 km. During the experiment, the arrays recorded several regional earthquakes related to the dynamics of the Brans- field Strait and adjoining areas and local volcano-tectonic earthquakes. Seismograms of earthquakes recorded at regional distances reveal a marked difference in the ap- parent velocities measured at the two array sites. We investigate the causes and implications of these anomalies by first comparing the effectiveness of estimating the horizontal slowness vector using three different techniques: the multiple signal classification (MUSIC) approach, the zero-lag cross correlation (ZLC) method, and plane-wave fitting to P-wave arrival times. While each technique provides the same horizontal slowness vector as the most likely estimates, the plane-wave fitting is associated with the most robust definition of measurement uncertainties. We then investigate the dispersive properties of Rayleigh waves in the 1-8 Hz frequency band at both arrays and invert the two dispersion curves for a shallow velocity structure. The results indicate a marked difference in the seismic velocities for the shallower 200 m beneath the two sites. This may be reconciled with the observed wave vector anomalies by assuming the existence of a sharp lateral velocity heterogeneity, the effect of which would be to bend downward rays impinging at the northernmost array. The reliability of this hypothesis is verified by computing finite-difference wave fronts in a 2D heterogeneous medium. Based on the morpho-structural char- acteristics of the volcano, the inferred velocity discontinuity maybe associated with the ring-fracture system bordering the collapsed caldera structure that extends over the inner part of the island.

Attenuation of short period seismic waves at Etna as compared to other volcanic areas

CodaQ for Etna volcano is frequency dependent and theQ frequency pattern and the numerical values ranging from about 100 at 1 Hz to about 300 at 18 Hz are similar to the values obtained for other volcanoes: Campi Flegrei, Aeolian Islands and Hawaii. Moreover the frequency pattern and the numerical values of coda quality factor, for most of the seismically active zones of Italy are very different from those of the volcanic zones.Several studies of the location of magma chambers show the presence of magma pockets beneath Lipari and Vulcano Islands of the Aeolian archipelago and an anomalous low velocity body beneath Etna. These evidences suggest that a possible interpretation of the characteristic frequency pattern ofQ on volcanic areas is that the presence of magma can modify the scattering environment and consequently the codaQ estimates.

Separation of intrinsic and scattering Q for volcanic areas: a comparison between Etna and Campi Flegrei

Abstract We studied the seismic attenuation in the short period range (1–24 Hz) at the Campi Flegrei volcanic area, using the method of Frankel (1991) to separate intrinsic and scattering attenuation parameters. The results, compared to those obtained at Etna volcano, show that scattering phenomena play a main role in the attenuation mechanism of the seismic waves for these two volcanic areas. The comparison of the present results with those obtained in non volcanic areas, allow us to re-consider the importance of the molten materials (magma) as a main cause of the seismic attenuation in the active volcanic areas.