1,721,076 research outputs found
From the detection of monitoring anomalies to the probabilistic forecast of the evolution of volcanic unrest: an entropy-based approach
No full textOwing to the current lack of plausible and exhaustive physical pre-eruptive models, often volcanologists rely on the observation of monitoring anomalies to track the evolution of volcanic unrest episodes. Taking advantage from the work made in the development of Bayesian Event Trees (BET), here we formalize an entropy-based model to translate the observation of anomalies into probability of a specific volcanic event of interest. The model is quite general and it could be used as a stand-alone eruption forecasting tool or to set up conditional probabilities for methodologies like the BET and of the Bayesian Belief Network (BBN). The proposed model has some important features worth noting: (i) it is rooted in a coherent logic, which gives a physical sense to the heuristic information of volcanologists in terms of entropy; (ii) it is fully transparent and can be established in advance of a crisis, making the results reproducible and revisable, providing a transparent audit trail that reduces the overall degree of subjectivity in communication with civil authorities; (iii) it can be embedded in a unified probabilistic framework, which provides an univocal taxonomy of different kinds of uncertainty affecting the forecast and handles these uncertainties in a formal way. Finally, for the sake of example, we apply the procedure to track the evolution of the 1982–1984 phase of unrest at Campi Flegrei.Published5OSV1: Verso la previsione dei fenomeni vulcanici pericolosiJCR Journa
Machine Learning for Tsunami Waves Forecasting Using Regression Trees
Full text linkAfter a seismic event, tsunami early warning systems (TEWSs) try to accurately forecast the maximum height of
incident waves at specific target points in front of the coast, so that early warnings can be launched on locations
where the impact of tsunami waves can be destructive to deliver aids in these locations in the immediate post-
event management. The uncertainty on the forecast can be quantified with ensembles of alternative scenarios.
Similarly, in probabilistic tsunami hazard analysis (PTHA) a large number of simulations is required to cover the
natural variability of the source process in each location. To improve the accuracy and computational efficiency
of tsunami forecasting methods, scientists have recently started to exploit machine learning techniques to
process pre-computed simulation data. However, the approaches proposed in literature, mainly based on neural
networks, suffer of high training time and limited model explainability. To overtake these issues, this paper
describes a machine learning approach based on regression trees to model and forecast tsunami evolutions. The
algorithm takes as input a set of simulations forming an ensemble that describes potential benefit regional impact
of tsunami source scenarios in a given source area, and it provides predictive models to forecast the tsunami
waves for other potential tsunami sources in the same area. The experimental evaluation, performed on the
2003 M6.8 Zemmouri-Boumerdes earthquake and tsunami simulation data, shows that regression trees achieve
high forecasting accuracy. Moreover, they provide domain experts with fully-explainable and interpretable
models, which are a valuable support for environmental scientists because they describe underlying rules and
patterns behind the models and allow for an explicit inspection of their functioning. This can enable a full and
trustable exploration of source uncertainty in tsunami early-warning and urgent computing scenarios, with large
ensembles of computationally light tsunami simulations
A parallel machine learning-based approach for tsunami waves forecasting using regression trees
No full textForecasting Tsunami Waves Using Regression Trees
No full textAfter a seismic event, tsunami early warning systems (TEWSs) try to accurately forecast the maximum height of incident waves at specific target points in front of the coast. The goal is to launch early warnings on locations where the impact of tsunami waves can be destructive, and to refine these forecasts in urgent computing mode in its immediate aftermath, to help organizing potential recovery operations. For improving the accuracy and computational efficiency of classic tsunami forecasting methods based on simulation models, scientists have recently started to exploit machine learning techniques to process pre-computed simulation data, in order to extract tsunami predictive models. However, the proposed approaches, mainly based on neural networks, suffer of high training time and limited model explainability. This paper describes a machine learning approach based on regression trees to model and forecast tsunami evolutions to overtake these issues. The experimental evaluation, performed on a real-world earthquake and tsunami simulation case study, shows that regression trees achieve high forecasting accuracy. Moreover, they provide domain experts with fully-explainable and interpretable models, which are a valuable support for environmental scientists because they describe underlying rules and patterns behind the models and allow for an explicit inspection of their functioning
S43A-2470: On the uncertainties of seismic parameters: a Bayesian framework for their estimation using Brune's model
No full textThe estimation of seismic parameters from ground-motion records is subject to many uncertainties, such as:
(i) parameterization, modeling procedures and underlying hypotheses, (ii) approximated input parameters, (iii) instrumental errors on records and their impact in data post-processing, (iv) procedures to estimate model’s parameters. However these uncertainties are rarely treated and propagated to the final results.
For example, on one side, density of rocks, velocity model, geometrical spreading, radiation pattern are just some of the common parameters needed to estimate the main seismic parameters of an earthquake and are generally used as average values. On the other side, uncertainties derived from the acquisition system and processing of the data are often neglected. Nevertheless, in many cases these uncertainties may be particularly important, as for example in the analysis of historical
earthquakes, where both instrumental response and treatment of analog records intrinsically imply non negligible sources of uncertainty.
Here, we present a new Bayesian procedure to estimate seismic parameters that allows: (i) to obtain a robust estimation of the Brune’s model parameters (Brune 1970, 1971) and relatives uncertainties, (ii) to account for the uncertainty related to the Earth model, and (iii) to propagate such uncertainties on the estimation of seismological parameters (seismic moment, moment magnitude, radius of the circular source zone and static stress drop). It is important to highlight that this study does not want to discuss the validity or the physical significance of the Brune’s model, but it is focused on the details of how to fit it on a dataset in order to evaluate the seismological parameters, accounting and properly propagating a rather large range of uncertainties. These capabilities of the proposed procedure are finally demonstrated through an illustrative application analyzing seismic records from historical events.UnpublishedSan Francisco, USA.3.1. Fisica dei terremotiope
Assessing volumes of tephra fallout deposits: a simplified method for data scarcity cases
No full textA new method for assessing volumes of tephra deposits based on only two thickness data is presented. It is based on the assumptions of elliptical shape for isopachs, a statistical characterization of their eccentricity, and an empirical relationship between their deposit thinning length scale and volumes. The method can be applied if the pair of thickness data are suf- ficiently distant from the volcano source, with a minimum distance ratio larger than 2. The method was tested against about 40 published volumes, from both equatorial belt and mid-latitude volcanoes. The results are statistically consistent with the published results, demonstrating the usefulness of the method. When applied in forward, the model allowed us to calculate the volume for some important tephra layers in the Mediterranean tephrostratigraphy, providing, for the first time, an assess- ment of the size of these eruptions or layers
S43A-2470: On the uncertainties of seismic parameters: a Bayesian framework for their estimation using Brune's model
No full textThe estimation of seismic parameters from ground-motion records is subject to many uncertainties, such as:
(i) parameterization, modeling procedures and underlying hypotheses, (ii) approximated input parameters, (iii) instrumental errors on records and their impact in data post-processing, (iv) procedures to estimate model’s parameters. However these uncertainties are rarely treated and propagated to the final results.
For example, on one side, density of rocks, velocity model, geometrical spreading, radiation pattern are just some of the common parameters needed to estimate the main seismic parameters of an earthquake and are generally used as average values. On the other side, uncertainties derived from the acquisition system and processing of the data are often neglected. Nevertheless, in many cases these uncertainties may be particularly important, as for example in the analysis of historical
earthquakes, where both instrumental response and treatment of analog records intrinsically imply non negligible sources of uncertainty.
Here, we present a new Bayesian procedure to estimate seismic parameters that allows: (i) to obtain a robust estimation of the Brune’s model parameters (Brune 1970, 1971) and relatives uncertainties, (ii) to account for the uncertainty related to the Earth model, and (iii) to propagate such uncertainties on the estimation of seismological parameters (seismic moment, moment magnitude, radius of the circular source zone and static stress drop). It is important to highlight that this study does not want to discuss the validity or the physical significance of the Brune’s model, but it is focused on the details of how to fit it on a dataset in order to evaluate the seismological parameters, accounting and properly propagating a rather large range of uncertainties. These capabilities of the proposed procedure are finally demonstrated through an illustrative application analyzing seismic records from historical events.UnpublishedSan Francisco, USA.3.1. Fisica dei terremotiope
A Testable Worldwide Earthquake Faulting Mechanism Model
No full textIn this article, we present a simple model to forecast global focal mechanisms. This model is based on a simple discrete counting distribution of the global centroid moment tensor catalog, and it also includes, using a Bayesian scheme, the a priori information from the Anderson theory of faulting. Our model is tested in hindcasting mode against independent data of global large earthquakes with Ms≥7. We obtained statistically significant good agreement between model and data using consistency test, demonstrating that this simple model can satisfactorily forecast focal mechanisms at the global scale. The defined testing procedure can be used to test the model in prospective mode against future events. These forecasts may inform short‐ to long‐term hazard quantifications that require a finite source characterization, as well as real‐time source inversion algorithms.Published3577–35853T. Fisica dei terremoti e Sorgente SismicaJCR Journa
- …
