A Novel Method for Predicting Local Site Amplification Factors Using 1-D Convolutional Neural Networks

The analysis of site seismic amplification characteristics is one of the important tasks of seismic safety evaluation. Owing to the high computational cost and complex implementation of numerical simulations, significant differences exist in the prediction of seismic ground motion amplification in engineering problems. In this paper, a novel prediction method for the amplification characteristics of local sites was proposed, using a state-of-the-art convolutional neural network (CNN) combined with real-time seismic signals. The amplification factors were computed by the standard spectral ratio method according to the observed records of seven stations in the Lower Hutt Valley, New Zealand. Based on the geological exploration data from the seven stations and the geological hazard information of the Lower Hutt Valley, eight parameters related to the seismic information were presumed to influence the amplification characteristics of the local site. The CNN method was used to establish the relationship between the amplification factors of local sites and the eight parameters, and the training samples and testing samples were generated through the observed and geological data other than the estimated values. To analyze the CNN prediction ability for amplification factors on unrecorded domains, two CNN models were established for comparison. One CNN model used about 80% of the data from 44 seismic events of the seven stations for training and the remaining data for testing. The other CNN model used the data of six stations to train and the remaining station’s data to test the CNN. The results showed that the CNN method based on the observation data can provide a powerful tool for predicting the amplification factors of local sites both for recorded positions and for unrecorded positions, while the traditional standard spectral ratio method only predicts the amplification factors for recorded positions. The comparison of the two CNN models showed that both can effectively predict the amplification factors of local ground motion without records, and the accuracy and stability of predictions can meet the requirements. With increasing seismic records, the CNN method becomes practical and effective for prediction purposes in earthquake engineering.

Near-Source Simulation of Strong Ground Motion in Amatrice Downtown Including Site Effects

On 24 August 2016, a Mw 6.0 earthquake started a damaging seismic sequence in central Italy. The historical center of Amatrice village reached the XI degree (MCS scale) but the high vulnerability alone could not explain the heavy damage. Unfortunately, at the time of the earthquake only AMT station, 200 m away from the downtown, recorded the mainshock, whereas tens of temporary stations were installed afterwards. We propose a method to simulate the ground motion affecting Amatrice, using the FFT amplitude recorded at AMT, which has been modified by the standard spectral ratio (SSR) computed at 14 seismic stations in downtown. We tested the procedure by comparing simulations and recordings of two later mainshocks (Mw 5.9 and Mw 6.5), underlining advantages and limits of the technique. The strong motion variability of simulations was related to the proximity of the seismic source, accounted for by the ground motion at AMT, and to the peculiar site effects, described by the transfer function at the sites. The largest amplification characterized the stations close to the NE hill edge and produced simulated values of intensity measures clearly above one standard deviation of the GMM expected for Italy, up to 1.6 g for PGA.

Combining recordings of earthquake ground-motion and ambient vibration analysis to estimate site response variability in the city of Lucerne, Switzerland

<p>Estimation of site effects is an essential part of local seismic hazard and risk assessment, especially in densely populated urban areas. The goal of this study is to assess the site response variability in the city of Lucerne (Central Switzerland), located in a basin filled with unconsolidated deposits. Even though it is a low-to-moderate seismicity area, the long-term seismic risk cannot be neglected, in particular, because the region was struck by strong earthquakes in the past (i.e. Mw 5.9 in 1601).</p><p>To determine the spatial distribution of the soil response in the test area, we combined earthquake and ambient noise recordings using the Hybrid Standard Spectral Ratio method (SSRh) introduced by Perron et al. (2018). In the first step, we installed a temporary seismic network to record ground-motion from low-magnitude or distant earthquakes. At selected urban sites inside the sedimentary basin, the dataset was used to estimate the amplification factors with respect to a rock site using the Standard Spectral Ratio approach (SSR - Borcherdt, 1970). Then, a survey including several dozens of densely distributed single-station ambient noise measurements was performed which enabled us to estimate the basin response variability relative to the seismic stations of the temporary seismic network. Finally, we corrected the noise-based evaluation using the SSR amplification functions. To verify the useability of the presented technique in the Lucerne area, we applied the SSRh method also to the temporary stations, the resulting amplification functions largely coincide with the SSR curves. However, the daily variability of the noise wavefield due to human activities can slightly affect the results. We will also discuss the influence of the station distribution and density of the temporary network deployment.</p><p>The amplification model for the Lucerne area estimated using the SSRh method shows consistency with geological data. The results indicate that seismic waves can be amplified up to 10 times in some parts of the basin compared to the rock site. The highest amplification factors are observed for frequencies between 0.8 and 2Hz. This means a local significant increase in seismic hazard.</p><p>The presented work is a part of a detailed site response analysis study for the Lucerne area, considering 2D and 3D site effects and potential non-linear soil behaviour. This PhD project is performed in the framework of the Horizon 2020 ITN funded project URBASIS-EU, which focuses on seismic hazard and risk in urban areas.</p><p>REFERENCES</p><p>Borcherdt, R.D., 1970. Effects of local geology on ground motion near San Francisco Bay. Bull. Seismol. Soc. Am. 60, 29–61.</p><p>Perron, V., Gélis, C., Froment, B., Hollender, F., Bard, P.-Y., Cultrera, G., Cushing, E.M., 2018. Can broad-band earthquake site responses be predicted by the ambient noise spectral ratio? Insight from observations at two sedimentary basins. Geophys. J. Int. 215, 1442–1454.</p>

Methodology to identify the reference rock sites in regions of medium-to-high seismicity: an application in Central Italy

SUMMARY To evaluate the site response using both empirical approaches (e.g. standard spectral ratio, ground motion models (GMMs), generalized inversion techniques, etc.) and numerical 1-D/2-D analyses, the definition of the reference motion, that is the ground motion recorded at stations unaffected by site-effects due to topographic, stratigraphic or basin effects, is needed. The main objective of this work is to define a robust strategy to identify the seismic stations that can be considered as reference rock sites, using six proxies for the site response: three proxies are related to the analysis of geophysical and seismological data (the repeatable site term from the residual analysis, the resonance frequencies from horizontal-to-vertical spectral ratios on noise or earthquake signals, the average shear wave velocity in the first 30 m); the remaining ones concern geomorphological and installation features (outcropping rocks or stiff soils, flat topography and absence of interaction with structures). We introduce a weighting scheme to take into account the availability and the quality of the site information, as well as the fulfillment of the criterion associated to each proxy. We also introduce a hierarchical index, to take into account the relevance of the proposed proxies in the description of the site effects, and an acceptance threshold for reference rock sites identification. The procedure is applied on a very large data set, composed by accelerometric and velocimetric waveforms, recorded in Central Italy in the period 2008–2018. This data set is composed by more than 30 000 waveforms relative to 450 earthquakes in the magnitude range 3.2–6.5 and recorded by more than 450 stations. A total of 36 out of 133 candidate stations are identified as reference sites: the majority of them are installed on rock with flat topography, but this condition is not sufficient to guarantee the absence of amplifications, especially at high frequencies. Seismological analyses are necessary to exclude stations affected by resonances. We test the impact of using these sites by calibrating a GMMs. The results show that for reference rock sites the median predictions are reduced down to about 45 per cent at short periods in comparison to the generic rock motions.