From the 2010s on, pattern classification has proven an effective method for flagging alerts of volcano unrest before eruptive activity at Mt. Etna, Italy. The analysis has been applied online to ...volcanic tremor data, and has supported the surveillance activity of the volcano that provides timely information to Civil Protection and other authorities. However, after declaring an alert, no one knows how long the volcano unrest will last and if a climactic eruptive activity will actually begin. These are critical aspects when considering the effects of a prolonged state of alert. An example of longstanding unrest is related to the Christmas Eve eruption in 2018, which was heralded by several months of almost continuous Strombolian activity. Here, we discuss the usage of thresholds to detect conditions leading to paroxysmal activity, and the challenges associated with defining such thresholds, leveraging a dataset of 52 episodes of lava fountains occurring in 2021. We were able to identify conservative settings regarding the thresholds, allowing for an early warning of impending paroxysm in almost all cases (circa 85% for the first 4 months in 2021, and over 90% for the whole year). The chosen thresholds also proved useful to predict that a paroxysmal activity was about to end. Such information provides reliable numbers for volcanologists for their assessments, based on visual information, which may not be available in bad weather or cloudy conditions.
Early-warning assessment of a volcanic unrest requires that accurate information from monitoring is continuously gathered before volcanic activity starts. Seismic data are an optimal source of such ...information, overcoming safety problems due to dangerous conditions for field surveys or cloud cover that may hinder visibility. We designed a multi-station warning system based on the classification of patterns of the background seismic radiation, so-called volcanic tremor, by using Self-Organizing Maps (SOM) and fuzzy clustering. The classifier automatically detects patterns that are typical footprints of volcanic unrest. The issuance of the SOM colors on DEM allows their geographical visualization according to the stations of detection; this spatial location makes it possible to infer areas potentially impacted by eruptive phenomena. Tested at Mt. Etna (Italy), the classifier forecasted in hindsight patterns associated with fast-rising magma (typical of lava fountains) as well as a relatively long lead time of the outburst (lava flows from eruptive fractures). Receiver Operating Characteristics (ROC) curves gave an Area Under the Curve (AUC) ∼0.8 indicative of a good detection accuracy that cannot be achieved from a mere random choice.
Die Buchreihe Frühe Neuzeit – begründet 1987 von Jörg Jochen Berns, Gotthard Frühsorge, Klaus Garber, Wilhelm Kühlmann und Jan-Dirk Müller – dient der Grundlagenforschung in Editionen, Monographien ...und Sammelbänden. Dabei strebt sie nicht die großräumige Überschau an, die vorschnelle Synthese oder prätentiöse Konstruktion, sondern nimmt den Umweg über die Arbeit am Detail und die Erkundung verschütteter Traditionszusammenhänge.
SUMMARY
Infrasound monitoring plays an important role in the framework of the surveillance of Mt. Etna, Europe’s largest active volcano. Compared to seismic monitoring, which is particularly ...effective for buried sources, infrasound signals mirror the activity of shallow sources like Strombolian explosions or degassing. The interpretation of infrasound signals is difficult to the untrained eye, as we have to account for volcanic and non-volcanic sources. The problem of handling large and complex data sets can be tackled with machine learning, namely pattern recognition techniques. Here, we focus on so-called ‘Unsupervised Learning’, where we identify groups of patterns being similar to each other. The degree of similarity is based on a metric measuring the distance among the features of the patterns. This work aims at the identification of typical regimes of infrasound radiation and their relation to the state of volcanic activity at Mt. Etna. For this goal, we defined features describing any infrasound pattern. These features were obtained using wavelet transform. We applied ‘Self-Organizing Maps’ (SOM) to the features projecting them to a 2-D representation space—the ‘map’. An intriguing aspect of SOM resides in the fact that the position of the patterns on the map can be expressed by a colour code, in a manner that similar patterns are assigned a similar colour code. This simplified representation of multivariate patterns allows to follow the development of their characteristics with time efficiently. During a training phase we considered a reference data set, which encompassed a large variety of scenarios. We identified typical groups of patterns which correspond to a specific regime of activity, being representative of the state of the volcano or noise conditions. These groups form areas on the 2-D maps. In a second step, we considered a test data set, which was not used during the training phase. Applying the same pre-processing as for the training data, we blindly assigned the test patterns to the regimes found before, identifying the one whose colour code is most similar to the one calculated to the test pattern. We are thus able to assess the validity of the prediction. The classification scheme presented provides a reliable assessment of the state of activity and adds useful and supplementary details to the results of the real-time automatic system in operation at Istituto Nazionale di Geofisica e Vulcanologia—Osservarorio Etneo. This is of particular importance when no visible information of the volcanic activity is available either for unfavourable meteorological conditions or during night time.
Ground motion prediction equations (GMPEs) have been derived for peak ground acceleration (PGA), velocity (PGV), and 5 % damped spectral acceleration (PSA) at frequencies between 0.1 and 10 Hz for ...the volcanic area of Mt. Etna. The dataset consists of 91 earthquakes with epicentral distances between 0.5 and 100 km. Given the specific characteristics of the area, we divided our data set into two groups: shallow events (SE, focal depth <5 km), and deep events (DE, focal depth >5 km). The range of magnitude covered by the SE and the DE is 3.0 ≤
M
L
≤ 4.3 and 3.0 ≤
M
L
≤ 4.8, respectively. Signals of DE typically have more high frequencies than those of SE. These differences are clearly reflected in the empirical GMPEs of the two event groups. Empirical GMPEs were estimated considering several functional forms: Sabetta and Pugliese (Bull Seism Soc Am 77:1491–1513,
1987
) (SP87), Ambraseys et al. (Earth Eng Struct Dyn 25:371–400,
1996
) (AMB96), and Boore and Atkinson (Earth Spectra 24:99–138,
2008
) (BA2008). From ANOVA, we learn that most of the errors in our GMPEs can be attributed to unmodeled site effects, whereas errors related to event parameters are limited. For DE, BA2008 outperforms the simpler models SP87 or AMB96. For SE, the simple SP87 is preferable considering the Bayesian Information Criterion since it proves more stable with respect to confidence and gives very similar or even lower prediction errors during cross-validation than the BA2008 model. We compared our results to relationships derived for Italy (ITA10, Bindi et al. Bull Earth Eng 99:2471–2488,
2011
). For SE, the main differences are observed for distances greater than about 5 km for both horizontal and vertical PGAs. Conversely, for DE the ITA10 heavily overestimates the peak ground parameters for short distances.
Die Buchreihe Frühe Neuzeit - begründet 1987 von Jörg Jochen Berns, Gotthard Frühsorge, Klaus Garber, Wilhelm Kühlmann und Jan-Dirk Müller - dient der Grundlagenforschung in Editionen, Monographien ...und Sammelbänden. Dabei strebt sie nicht die großräumige Überschau an, die vorschnelle Synthese oder prätentiöse Konstruktion, sondern nimmt den Umweg über die Arbeit am Detail und die Erkundung verschütteter Traditionszusammenhänge.
Source and Qp parameters were estimated from the inversion of first arrival P waveform durations of about 300 microearthquakes recorded at a digital seismic network operating in southeastern Sicily. ...The average risetime and pulse width at each station do not show large differences, allowing us to exclude significant differential attenuation site effects. A first Qp estimate was obtained by applying the classical risetime method, under the assumption of a point‐like source time function. In order to investigate the effect of directivity due to the finiteness of seismic sources, new nonlinear relationships, based on a circular crack model rupturing at a constant velocity, were numerically built. These relationships were used to formulate a nonlinear inverse method for retrieving source (radius, dip, and strike of the circular crack) and Qp parameters from the inversion of risetime and pulse width data. The application of the method produced a better fit of the observed data and a Qp value higher than that obtained by applying the risetime method. The discrepancy between the different Q estimates may be due to a trade‐off among source dimension and Qp, as we inferred from a test on a subset of low‐magnitude events (Ml ≤ 2.5). A good agreement with independent estimates of fault plane solutions, as inferred from P polarities and S polarizations, was found. The estimated stress drops are generally very low (0.1–10 bars). This suggests that the background seismic activity in southeastern Sicily is related to fault segments and/or weakened zones where great stress accumulations are hindered.
We present the application of a classification method based on Kohonen maps and fuzzy clustering to geochemical analyses of volcanic products erupted on Mt. Etna from 1995 to 2005. Based on 13 major ...and trace elements, the classification allows a new way to visualize distinct compositional features of magma both considering long period as well as single eruptive events, such as in 2001 and 2002–03 flank eruptions. Products of the various vents do not necessarily form homogeneous groups, but show clear trends of chemical evolution with time. Using a convenient color code, the graphical visualization of the results in just a single picture allows the rapid identification of the compositional features of each sample and their comparison with all the products analyzed in the 10-year-long time span. This single picture accounts for the mutual interactions of the 13 components avoiding shortcomings of classical low-dimensional plots where components relevant for the discrimination have to be found in a priori study of many diagrams. On the basis of the synoptic information provided by pattern classification, we identify links between the products of different eruptive vents which deliver a reliable picture of a multifaceted plumbing system, in agreement with geochemical and geophysical evidence reported in literature. The analysis of the 13-dimensional data set using the Kohonen maps and fuzzy clustering simultaneously turned out to be straightforward and easy. Accordingly, the results of this application will be useful also as a contextual data set for new data in future ongoing eruptive episodes.
Ground motion scenarios for Mt. Etna are created using synthetic simulations with the program EXSIM. A large data set of weak motion records is exploited to identify important input parameters which ...govern the modeling of wave propagation effects, such as Q-values, high frequency cut-off and geometrical spreading. These parameters are used in the simulation of ground motion for earthquakes causing severe damage in the area. Two seismotectonic regimes are distinguished. Volcano-tectonic events, though being of limited magnitude (M
max
ca. 5), cause strong ground shaking for their shallow foci. Being rather frequent, these events represent a considerable threat to cities and villages on the flanks of the volcano. A second regime is related to earthquakes with foci in the crust, at depths of 10–30 km, and magnitudes ranging from 6 to 7. In our synthetic scenarios, we chose two examples of volcano-tectonic events, i.e. the October 29, 2002, Bongiardo event (I = VIII) and the May 8, 1914, Linera earthquake (I = IX–X). A further scenario regards the February 20, 1818 event, considered representative for stronger earthquakes with foci in the crust. We were able to reproduce the essential features of the macroseismic field, in particular accounting for the possibility of strong site effects. We learned that stress drop estimated for weak motion events is probably too low to explain the intensity of ground motion during stronger earthquakes. This corresponds to findings reported in the literature claiming an increase of stress drop with earthquake size.
We collected drone data to quantify the kinematics at extensional fractures and normal faults, integrated this information with seismological data to reconstruct the stress field, and critically ...compared the results with previous fieldwork to assess the best practice. As a key site, we analyzed a sector of the northeast rift of Mt Etna, an area affected by continuous ground deformation linked to gravity sliding of the volcano's eastern flank and dike injections. The studied sector is characterized also by the existence of eruptive craters and fissures and lava flows. This work shows that this rift segment is affected by a series of NNE- to NE-striking, parallel extensional fractures characterized by an opening mode along an average N105.7∘ vector. The stress field is characterized by a σHmin trending northwest–southeast. Normal faults strike parallel to the extensional fractures. The extensional strain obtained by cumulating the net offset at extensional fractures with the fault heave gives a stretching ratio of 1.003 in the northeastern part of the study area and 1.005 in the southwestern part. Given a maximum age of 1614 CE for the offset lavas, we obtained an extension rate of 1.9 cm yr−1 for the last 406 years. This value is consistent with the slip along the Pernicana Fault system, confirming that the NE rift structures accommodate the sliding of the eastern flank of the volcano.