Application of deep learning-based neural networks using theoretical seismograms as training data for locating earthquakes in the Hakone volcanic region, Japan

In the present study, we propose a new approach for determining earthquake hypocentral parameters. This approach integrates computed theoretical seismograms and deep machine learning. The theoretical seismograms are generated through a realistic three-dimensional Earth model, and are then used to create spatial images of seismic wave propagation at the Earth’s surface. These snapshots are subsequently utilized as a training data set for a convolutional neural network. Neural networks for determining hypocentral parameters such as the epicenter, depth, occurrence time, and magnitude are established using the temporal evolution of the snapshots. These networks are applied to seismograms from the seismic observation network in the Hakone volcanic region in Japan to demonstrate the suitability of the proposed approach for locating earthquakes. We demonstrate that the determination accuracy of hypocentral parameters can be improved by including theoretical seismograms for different earthquake locations and sizes, in the learning data set for the deep machine learning. Using the proposed method, the hypocentral parameters are automatically determined within seconds after detecting an event. This method can potentially serve in monitoring earthquake activity in active volcanic areas such as the Hakone region.

Application of Deep Learning-Based Neural Networks Using Theoretical Seismograms as Training Data for Locating Earthquakes in the Hakone Volcanic Region, Japan

In the present study, we propose a new approach for determining earthquake hypocentral parameters. This approach integrates computed theoretical seismograms and deep machine learning. The theoretical seismograms are generated through a realistic three-dimensional Earth model, and are then used to create spatial images of seismic wave propagation at the Earth’s surface. These snapshots are subsequently utilized as a training dataset for a convolutional neural network. Neural networks for determining hypocentral parameters such as the epicenter, depth, occurrence time, and magnitude are established using the temporal evolution of the snapshots. These networks are applied to seismograms from the seismic observation network in the Hakone volcanic region in Japan to demonstrate the suitability of the proposed approach for locating earthquakes. We demonstrate that the determination accuracy of hypocentral parameters can be improved by including theoretical seismograms for different earthquake locations and sizes, in the learning dataset for the deep machine learning. Using the proposed method, the hypocentral parameters are automatically determined within seconds after detecting an event. This method can potentially serve in monitoring earthquake activity in active volcanic areas such as the Hakone region.

Complex seismic activity and local stress perturbation of the Pyeongnam basin, northern Korean Peninsula

<div> <div> <div> <p>The western region of the Pyeongnam Basin has relatively higher e​arthquake activity than the rest of the Korean Peninsula. We analyzed 48 earthquakes in the area, with a magnitude (M<sub>L</sub>) of 2.0 or more, from January 2009 to June 2019. The hypocentral parameters were re-determined using an iterative algorithm that repeats the calculation until the residual error between the observed and calculated arrival time of a seismic phase at each station is minimized. Using the hypocenters and the optimal 1-D velocity model derived from this process, the focal mechanisms were determined using the first-motion polarities of body waves. Many earthquakes are associated with left-lateral strike-slip faults, with a strike in the NW-SE direction and a normal faulting component. A stress inversion was performed using data of the pressure and tensional axes from the focal mechanisms. The maximum principal stress in the study area acts in the NW-SE direction with high angles of plunge and differs from the maximum horizontal principal stress in the rest of the Korean Peninsula. This stress perturbation is caused by the detachment of a small local stress from the regional stress field due to the presence of weak faults with low shear strength that develop in the sedimentation environment of the Pyeongnam Basin.</p> </div> </div> </div>

Instrumental seismic catalogue of Mt. Etna earthquakes (Sicily, Italy): ten years (2000-2010) of instrumental recordings

<p>Instrumental seismic catalogues are an essential tool for the zonation of the territory and the production of seismic hazard maps. They are also a valuable instrument for detailed seismological studies regarding active volcanoes and, above all, for interpreting the magma dynamics and the evolution of eruptive phenomena. In this paper, we show the first instrumental earthquake catalogue of Mt. Etna, for the period 2000-2010, with the purpose of producing a homogeneous dataset of 10 years of seismological observations. During this period, 16,845 earthquakes have been recorded by the seismic network run by the Istituto Nazionale di Geofisica and Vulcanologia, Osservatorio Etneo, in Catania. A total of 6,330 events, corresponding to approximately 40% of all earthquakes recorded, were located by using a one-dimensional VP velocity model. The magnitude completeness of the catalogue is equal to about 1.5 for the whole period, except for some short periods in 2001 and 2002-2003 and at the end of 2009. The reliability of the data collected is supported by the good values of the main hypocentral parameters through the time. The spatial distribution of seismicity allowed the highlighting of several seismogenetic areas characterized by different seismic rates and focal depths. This seismic catalogue represents a fundamental tool for several research aiming to a better understanding of the behavior of an active volcano such as Mt. Etna.</p><div> </div>