Fig 1 - uploaded by Mario Ruiz
Content may be subject to copyright.
Source publication
Geophysics experts are interested in understanding the behavior of volcanoes and forecasting possible eruptions by monitoring and detecting the increment on volcano-seismic activity, with the aim of safeguarding human lives and material losses. This paper presents an automatic volcanic event detection and classification system, which considers feat...
Context in source publication
Context 1
... study is referred to Cotopaxi, an active volcano which is part of the so-called Ring of Fire, located at latitude 0 • 41'05" S and longitude 78 • 25'54.8" W in the Andean mountain region of Ecuador. On this volcano, a monitoring system has been previously deployed by the Instituto Geofísico de la Escuela Politécnica Nacional (IGEPN) (see Fig. 1), which currently has installed: (a) six short period (SP) seismological stations (PITA, NAS2, VC1, REF, CAMI, and TAM), four of them with vertical-axis sensors and two of them with three-axis sensors, and all of them with a frequency response range of 1-50 Hz; (b) six broadband (BB) stations (VC2, REF, NAS, TAM, MORU, and VCES), with ...
Similar publications
Very few diseases are as deadly as Breast cancer
and Lung cancer that causes a high percentage of deaths every
year. Data mining approach works quite effectively in diagnosing
cancer. This has been possible mainly because of the potential
of classification systems that has been improved in a great
arrangement in helping medical experts. In this res...
Neovascularization is a new vessel in the retina beside the artery-venous. Neovascularization can appear on the optic disk and the entire surface of the retina. The retina categorized in Proliferative Diabetic Retinopathy (PDR) if it has neovascularization. PDR is a severe Diabetic Retinopathy (DR). An image classification system between normal and...
Citations
... Because of this, some of the assumptions made by these models cannot be verified, which has an impact on the precision of forecasts. Support vector machine (SVM), Naïve Bayes (NB), and random forest (RF) are examples of cutting-edge ML technologies that can outperform statistical models [21], [22], [23], [24], [25]. Many metrics can be used to evaluate the different models as in [26]. ...
The need for an earthquake early-warning system (EEWS) is unavoidable in order to save lives. In terms of managing earthquake disasters and achieving effective risk mitigation, the quick identification of the earthquake’s intensity is a valuable factor. In light of this, the on-site intensity measurement can be transmitted over an Internet of Things (IoT) network. In this regard, a machine learning (ML) strategy based on numerous linear and non-linear models is proposed in this study for a quick determination of earthquake intensity after two seconds from the P-wave onset. We call this model an on-site two-second ML model-based earthquake intensity determination (2S-ML-EIOS). The utilized dataset INSTANCE for this model is observed by the number of 386 stations from the Italian national seismic network. Our model has been trained on 50,000 occurrences (150 thousand of 2s-three-component seismic windows). The model has the ability to deal with limited features of the waveform traces leading to reliable estimation of the earthquake intensity. The suggested model has a 98.59% accuracy rate in predicting earthquake intensity. The suggested 2S-ML-EIOS model can be used with a centralized IoT system to promptly send the alarm, and the IoT system will then instruct the affected administration to take the appropriate action. The 2S-ML-EIOS results are contrasted with those from the traditional manual solution approach, which corresponds to the ideal solution mean. Based on the extreme gradient boosting (XGB) model, the 2S-ML-EIOS can achieve the best intensity determination, and this improved performance demonstrates the methodology’s efficacy for EEWS.
... The discrimination between natural and artificial seismic events is an important problem as several studies have demonstrated [19], [20], [21]. Accurate discrimination can provide the decision-makers with reliable approaches for disaster management to delineate the spatial distribution of a probable EQ and its influences on population [22], [23], [24]. ...
... As a result, some assumptions in these models cannot be established, which, in turn, affects the accuracy of predictions. With the breakthrough of ML technology, e.g., support vector machine (SVM) [23], [35], [36], Naïve Bayes (NB) [33], and random forest (RF) [37], it can achieve better performance compared to the statistical models. Although many efforts in the literature context have been exerted for discriminating the EQs and QBs, a reliable and adaptive solution is still desired. ...
Efficient handling and planning for the urban regions’ sustainable development require a vast range of up-to-date and thematic information. Besides, obtaining an uncontaminated catalog of seismic activity is desirable to study the earthquake (EQ) clusters spatial allocation, which is a key role in mitigating seismic hazards and alleviating EQ losses by enhancing the assessment of seismic hazards. This paper considers the northeastern part of Egypt where the seismicity catalog is contaminated by quarry blasts (QBs) operated throughout the mapped area. Consequently, it is desirable to discriminate these QBs from the EQs for genuine seismicity and hazard analysis. Accordingly, we provide an efficient machine learning (ML) model for decontaminating the seismicity database so that EQ clusters can be properly delineated by relying on 870 events (EQs and QBs) observed by only one seismic station called “
GLL
”, a member of the Egyptian National Seismic Network (ENSN). The model focuses on magnitudes < 3 that have high uncertainty of being EQs or QBs and take a long time for analysis. The approach examines several linear and non-linear ML models and finally selects the best model with only two features leading to the optimal classification between the EQs and QBs. The optimization process is accomplished throughout two stages. The obtained results prove that the proposed scheme achieves 100% discrimination between the EQs and QBs relying on the extreme gradient boosting (XGB) model.
... Traditional artificial neural networks usually connect different levels in a fully connected manner, which is easy to cause parameter redundancy. Therefore, the network needs to rely on a very large amount of data to train these parameters [7][8]. CNN chooses a large number of local connections to reduce the parameter scale of the network. ...
... By formula(8), the signal coefficients are mapped from the time-scale plane to the new time-frequency plane. This step is called synchronous compression. ...
Seismic detection technology has been widely used in safety detection of engineering construction abroad. Although it has just started in the field of engineering in our country, its role is becoming more and more important. Through computer technology, micro-seismic detection can provide accurate data for the construction safety detection of large-scale projects, which has important practical significance for the rapid and effective identification of micro-seismic signals. Based on this, the purpose of this article is to study the feature extraction and classification of microseismic signals based on neural games. This article first summarizes the development status of microseismic monitoring technology. Using traditional convolutional neural networks for analysis, a multi-scale feature fusion network is proposed on the basis of convolutional neural networks and big data, the multi-scale feature fusion network is used to research and analyze microseismic feature extraction and classification. This article systematically explains The principle of microseismic signal acquisition and the construction of multi-scale feature fusion network. And use big data, comparative analysis method, observation method and other research methods to study the theme of this article. Experimental research shows that the db7 wavelet base has little effect on the Megatron signal.
... En la Región Andina, las principales fuentes sísmicas son producto de la actividad volcánica, ver Lara Cueva et al. (2016), así como de las numerosas fallas geológicas existentes, ver Eguez et al. (2003). En particular, en la provincia del Azuay existen cuatro fallas geológicas (Paute, Girón, Gualaceo y Tarqui) y un historial de al menos cuatro sismos importantes. ...
La detección temprana de eventos sísmicos permite reducir daños materiales, el número de personas afectadas e incluso salvar vidas. En particular, la actividad sísmica en Ecuador es alta, dado que se encuentra en el denominado Cinturón de Fuego del Pacífico. En tal contexto, el presente artículo tiene como objetivo comparar algoritmos para la detección automática de eventos sísmicos. Dicha comparación se realiza con respecto a la funcionalidad y configuración de los parámetros requeridos para cada algoritmo. Además, la implementación se lleva a cabo sobre una plataforma computacional tipo SBC (Single Board Computer) con la finalidad de obtener una herramienta portable, escalable, económica y de bajo costo computacional. Los métodos comparados son: Classic STA/LTA, Recursive STA/LTA, Delayed STA/LTA, Z-detector, Baer and Kradolfer picker y AR-AIC (Autoregressive-Akaike-Information-Criterion-picker). Para la evaluación y comparación se desarrollan múltiples experimentos empleando registros sísmicos reales proporcionados por la Red Sísmica del Austro (RSA), disponibles como fuente de entrada a los algoritmos. Como resultado se obtiene que el algoritmo Classic STA/LTA presenta el mejor rendimiento, ya que del total de eventos reales (58), solo un evento no fue detectado. Además, se consiguen 6 falsos negativos, logrando un 98,2% de precisión. Cabe recalcar que el software utilizado para la comparación de algoritmos de detección de eventos sísmicos está disponible de forma libre.
... They have proved to be successful tools (as a second opinion) for analyzing data in various fields of study, including volcanic seismology. Some examples of MLC applications in the volcano seismic event classification context have been developed from supervised learning models such as artificial neural networks [3], [4], deep neural networks [5], [6], support vector machine (SVM) [7], [8], random forest [9] decision trees [10], Hidden Markov Model (HMM) [11], [12], evolutionary algorithms [13], [14] and Gaussian mixture models (GMM) [15] to other approaches based on unsupervised learning [16], [17] and semi-supervised learning [18]. ...
With the growing ability to collect large volumes of volcano seismic data, the detection and labeling process of these records is increasingly challenging. Clearly, analyzing all available data through manual inspection is no longer a viable option. Supervised Machine Learning models might be considered to automatize the analysis of data acquired by in situ monitoring stations. However, the direct application of such algorithms is defiant, given the high complexity of waveforms, and the scarce and often imbalanced amount of labeled data. In light of this and motivated by the wide success that Generative Adversarial Networks (GANs) have seen at generating images, we present ESeismic-GAN, a GAN model to generate the magnitude frequency response of volcanic events. Our experiments demonstrate that ESeismic-GAN learns to generate the frequency components that characterize long-period and volcano-tectonic events from Cotopaxi volcano. We evaluate the performance of ESeismic-GAN during the training stage using Frechet Distance and later on we reconstruct the signals into time-domain to be finally evaluated with Frechet Inception Distance.
... Similarly, for the classification, several MLCs have solved the volcano seismic events classification problem from different learning perspectives, ranging from supervised techniques such as artificial neural networks [17]- [19], deep neural networks [20]- [22], support vector machine (SVM) [8], [12], [23], random forest [13], [24] decision trees [25], [26], Hidden Markov Model (HMM) [27], [28], evolutionary algorithms [29], [30] and Gaussian mixture models (GMM) [31] to other techniques using unsupervised [32], [33] and semi-supervised learning [9]. Nevertheless, the approaches mentioned above incur a high algorithm complexity due to the feature calculation, selection, and classification steps required to carry out the solutions, which is impractical for real-time scenarios. ...
... Opnq ANN [13] 637`17 no no 95 Opn 4 q RF [13] 637`17 no no 93 Opnq linear SVM [25] 914`5 no no 97 Opn 3 q GMM [31] 667`2 no no 94 ...
This work proposes a new approach based on a suit combination of mathematical morphology and similarity criteria techniques to classify long-period and volcano-tectonic seismic events of the Cotopaxi volcano. The proposed method explores the seismic signal domain to compute a new feature space based on the edges map of the seismic events pattern represented in the gray-level spectrogram images, which is used to feed a set of similarity-based classifiers. The
$L_{2}$
-norm was selected as the best metric to be implemented by the proposed method. In terms of classification performance, the
$L_{2}$
-norm was statistically superior in the D1 data set (seismic events with overlapped signals of nonvolcanic origin) and similar in the D2 data set (events without overlapped signals) with respect to the other metrics, reaching accuracy mean scores of 93.34% and 96.88%, respectively. These results demonstrated that the computed edges map feature space is a better environment for separating both seismic events compared with the original gray-level space. Regarding the execution time, total time (TT) and time per-sample (TS) did not exceed 0.388 and 0.002 s during the training stage, respectively. During the testing stage, a TS of no more than 0.012 s was achieved. Finally, its execution time is faster, and the algorithm complexity is lower compared with the state-of-the-art methods, which makes it a practical and beneficial scheme to implement for real-time seismic events’ classification.
... In their feature extraction stages, most of the previous works deal with a well-known set of features (quantities derived from inputs (Bishop et al., 2006)) related to the microearthquakes in both time and frequency domains (Cárdenas-Peña et al., 2013;Álvarez et al., 2012). Moreover, there are also studies using together features from time, frequency, and scale domains (Soto et al., 2018;Lara-Cueva et al., 2016b;Lara-Cueva et al., 2015;Duque et al., 2020), and recently, the intensity statistics, shape, and texture features computed from the seismic event pattern represented in the grey-level spectrogram image (Pérez et al., 2020a). ...
... SVM is often sensitive to class imbalance, and for this reason, we first determined the training percentage for our database, which maximized the system performance, this value was set to 60%. We then set the free parameter ν to 0.05 by making an iterative process in order to maximize the system performance (Lara-Cueva et al., 2016b). ...
Adequate detection and classification of seismic events are crucial for understanding the internal status of a Volcano. Machine learning-based classifiers use different features from the time, frequency, and scale domains related to seismic events. Regarding power spectrum-based features, several methods can be used to compute such features. However, the more suitable method for analyzing volcanic activity is undetermined. This paper presents a study about the main frequency bands, which allows maximizing the performance metrics of an automated classifier for long-period (LP) and volcano-tectonic (VT) events based on parametric (Yule-Walker and Burg) and non-parametric (Welch and Multitaper) power spectrum density estimation methods. Feature selection using embedded (pruning) and wrapper (recursive feature elimination) methods was applied to select the main frequencies that maximize the balanced error rate of suitable classification algorithms, such as decision trees (DT) and support vector machines (SVM). Bootstrapping was used to estimate a confidence interval for the frequencies of the microearthquakes. An amplitude threshold difference of at least 3 dB was used to guarantee that possible frequency features that characterize each type of event do not overlap between classes. The method who achieved the worst overall performance was not considered by the voting strategy. A Dataset from Cotopaxi volcano was used to test the proposed classification schema. The best results show for DT classifier a total of 10 key frequencies, while for SVM classifier 39 key frequencies grouped in three main frequency bands, as main features to distinguish LP events from VT earthquakes. The best classification results were achieved by the Welch method with the DT and by the Multitaper method with the SVM classifiers. Furthermore, the study confirms that there is a frequency band above 40 Hz, which seems like a critical feature for the detection and classification of stages.
... The results of each classifier were measured in terms of Accuracy (A), Precision (P), Sensitivity or Recall (R), Specificity (S), and Balanced Error Rate (BER), which are defined as follows [13]: ...
... Several MLCs have been reported in the literature for seismic event classification. Variations of the support vector machine (SVM) method seem to be the most widely used type of classifier, e.g., SVM with linear kernel [19], [20] and multiclass SVM [20]- [22]. Other less popular methods that have been performed satisfactorily are artificial neural networks (ANNs) [22], [23], decision trees [19], [24], hidden Markov models [25], evolutionary algorithms [26], [27], and, more recently, Gaussian mixture model [28]. ...
... Variations of the support vector machine (SVM) method seem to be the most widely used type of classifier, e.g., SVM with linear kernel [19], [20] and multiclass SVM [20]- [22]. Other less popular methods that have been performed satisfactorily are artificial neural networks (ANNs) [22], [23], decision trees [19], [24], hidden Markov models [25], evolutionary algorithms [26], [27], and, more recently, Gaussian mixture model [28]. Thus, five MLCs with different taxonomies (depending on their functionality) were considered in this article to perform a fair evaluation of the proposed descriptor output. ...
This article proposes a new volcano seismic signal descriptor for improving the area under the receiver operating characteristic curve (AUC) in the classification of long-period (LP) and volcano-tectonic (VT) seismic events. It aims to describe a volcanic seismic event from a different and novel point of view that involves image processing techniques instead of classical seismic signal processing strategies, such as frequency or scale analysis. The proposed descriptor allows exploring the seismic signal space for obtaining the determination of the event patterns and, subsequently, the extraction of intensity-, shape-, and texture-based features into a numeric vectorial output for supplying a set of selected machine learning classifiers with different taxonomies. The descriptor was validated on a seismic signal database collected at the Cotopaxi volcano, containing a total of 637 events, including LP, VT, and other types of seismic events (e.g., rockfall or icequakes). An accuracy value of 96% was obtained in the determination of the event patterns using the signal database, while the values of 0.95 and 0.96 were obtained for the AUC when using a feedforward backpropagation artificial neural network classifier on two experimental data sets, containing feature vectors representing signal with and without event overlapping, respectively. The obtained results demonstrate that the proposed descriptor is capable of providing adequate seismic signal representations in a different feature space and that its output provides competitive results in the classification of volcanic seismic events.
... In order to monitor, prevent and mitigate the risks related to these volcanic hazards, technological tools are increasingly needed. Thus, a wide variety of systems has been developed in the last decades to automatically detect volcanic events based on their seismic signals [1,[8][9][10][11]. Many of the proposed systems use machine learning techniques based on supervised training to create compelling detection models [12,14]. ...
... Since seismic-event catalogs with labels for each event are needed by supervised training schemes (i.e., training, validation, and test data-sets), this work studies the dependency of machine learning techniques on the type of approach used to label volcanic-seismic signals. In particular, we have used seismic signals from the Cotopaxi volcano, located in Ecuador, which have labels provided by analysts of the Ecuadorian Geophysical Institute (IGEPN) that show the beginning and the end of each volcanic earthquake [8,9]. In addition, we have implemented a traditional detector knows as STA/LTA (short-term average/long-term average) to automate the process of labeling [7]. ...
... The raw seismic signals from the Cotopaxi volcano, located in Ecuador, were provided by the IGEPN in several files formatted according to the standard SAC. Each file contains 20 min of the seismic signal sampled to 100 Hz and belonging to the vertical-axis of station VC2 during the year 2012 [8,9]. The station VC2 is a broadband seismograph that is part of the monitoring network installed by the IGEPN, presenting a good frequency response between 0.01 and 50 Hz. ...
Several systems have been developed in the last years to automatically detect volcanic events based on their seismic signals. Many of those systems use supervised machine learning algorithms in order to create the detection models. However, the supervised training of these machine learning techniques requires labeled-signal catalogs (i.e., training, validation and test data-sets) that in many cases are difficult to obtain. In fact, existing labeling schemes can consume a lot of time and resources without guarantying that the final detection model is accurate enough. Moreover, every labeling technique can produce a different set of events, without being defined so far which technique is the best for volcanic-event detection. Hence, this work proves that the labeling scheme used to create training sets definitely impacts the performance of seismic-event detectors. This is demonstrated by comparing two techniques for labeling seismic signals before to train a system for automated detection of volcanic events. The first technique is automatic and computationally efficient, while the second one is a handmade and time-consuming process carried out by expert analysts. Results show that none of the labeling techniques is completely trustworthy. As a matter of fact, our main result reveals that an improved detection accuracy is obtained when machine learning classifiers are trained with the conjunction of diverse labeling techniques.
