S T G Raghukanth

Developing a ground motion model (GMM) for Fourier amplitude spectrum (FAS) is essential in seismology and engineering for generating response spectrum and synthetic time histories. Despite data-driven techniques being efficient in modeling complex relations, very few GMMs are developed for FAS using them. An efficient hybrid data-driven algorithm combining genetic algorithm and artificial neural network is implemented using the GeoNet database with 905 records from 77 events in the current work. The input parameters of the model are moment magnitude, Joyner–Boore distance, shear wave velocity, depth to the top of the rupture plane, fault, and tectonic flags. The developed FAS model is statistically tested to be robust and has good agreement with the recorded data and other available GMMs. The developed GMM to FAS has an overall correlation coefficient in the range of 0.8108–0.9298 and sigma in the range of 0.26–0.4 (in log10 units). Further, synthetic time histories are generated from the predicted FAS values and are consistent with various ground motion parameters and the response spectra.

The present work aims at exploring the application of nonlinear principal component analysis in dimensionality reduction and prediction of response spectra. The evaluation is performed based on log10 scaled response spectra at 91 spectral periods corresponding to 13552 records available in the NGA-West2 database. The non-linear principal component analysis performed on the data showed that 91 spectral periods can be addressed with just 3 principal components. Further, an artificial neural network (ANN) model is developed to predict these three principal components with magnitude, distance, shear wave velocity and focal mechanism as input. The inter- and intra-event residuals obtained for the response spectra predicted using the developed model are comparable with the existing ground motion prediction equations (GMPEs) from the same database. The developed model is also observed to capture all the prominent attenuation features of ground motions. Hence, the study indicates that the response spectra can be described with just three uncorrelated variables.

A novel methodology for obtaining a seismic zone map of India is demonstrated in this study, wherein a concrete theoretical framework is provided for deriving the zones and the respective zonal response spectra. The method involves time series clustering of uniform hazard response spectra (UHRS) that were obtained for the entire country on a 0.1° × 0.1° grid by performing probabilistic analysis corresponding to a 2475-year return period. The Euclidean distance between the UHRS values at all periods (27 data points between 0.01 s and 5 s) was taken as the similarity measure in an evolutionary particle swarm optimization algorithm. The analysis was conducted with a swarm population of 100 over 3000 iterations, and the mean UHRS of the resulting clusters was assumed as the cluster centre. Various quality/validity indices including the compactness measure, similarity measure, combined measure and Dunn Index were used to verify the results of the clustering. Based on these clusters, the entire country can be divided into seven zones, with a unique zonal spectrum for each zone.

Ground motion prediction equations (GMPEs) are developed using past strong-motion records to predict the effect of future events. Often, the records in the database are incomplete, not covering all possible input scenarios or not recorded at the site of interest for performing site-specific hazards. Such cases are handled by adjusting the GMPEs to suit the required site/region characteristics. Recent studies have shown that scaling the Fourier amplitude spectrum (FAS) rather than the pseudo-spectral acceleration (PSA) is physically justifiable. The present work develops a recurrent neural network to predict the PSA ordinates from the FAS and duration (D5-95) for the New Zealand region. The developed network has no potential underfit or overfit and has a strong correlation coefficient, R > 0.97, with total sigma values in the range of 0.1–0.13 (log10 units). If the predicted FAS and duration are used as inputs, its uncertainty must be included in the final sigma, which lies from 0.25 to 0.3 (log10 units). At low frequency, scaling of FAS and PSA values is identical. In contrast, scaling of higher-frequency FAS values affects the wide range of the PSA values, with a prominent effect initially observed at lower frequencies and later at higher frequencies.

This article addresses the effect of the rupture process on tsunami wave simulations by assessing the propagation of uncertainties from source to wave heights. Thirteen slip models available for the 2004 (Mw 9.2) Sumatra earthquake are utilized in the evaluation. First, quasi-static displacement of the ocean floor is estimated using Okada’s solutions. Further, the corresponding displacement time histories provided as an initial condition for tsunami simulations by modeling the region in Clawpack. The simulated results are compared against the four tidal-gauge data available in the east-coast of India and three altimeter recordings from satellites. The comparisons pointed to the sensitivity of simulated wave heights toward the input slip distribution and rupture process. Further, it is noted from the standard deviations estimated between the results of thirteen models that the value reduced from maximum slip (6.53 m) to displacement (2.60 m), which further reduces in the wave height estimates (1.70 m). Hence, this study suggests the need for proper quantification of the uncertainty propagation in tsunami hazard estimations.

This article focuses on estimating the seismic recurrence parameters of India and adjoining regions based on a comprehensive catalogue assimilated from various sources. The study region encompasses latitude 0 ∘ N–40 ∘ N and longitude 65 ∘ E–100 ∘ E. The updated catalogue for the region contained 69519 events, including 28770 mainshocks. The updated catalogue was employed in the estimation of recurrence characteristics of the region. Here, zonal and spatial smoothening-based approaches were employed to estimate seismicity characteristics on a grid of 0.1∘× 0.1∘. The active regions like the Himalayas, North-Eastern India, Andaman, and Koyna-Warna regions were observed to have relatively lesser b values indicating the occurrence of larger magnitude events and higher values for activity rate N(4), indicating a more frequent occurrence of an earthquake. The reported values can be further used in seismic hazard estimations for the region

In the current study, ground motion models (GMMs) are derived using the European Strong Motion (ESM) database for pseudo-spectral acceleration (PSA), peak ground acceleration (PGA), peak ground velocity (PGV), peak ground displacement (PGD), cumulative absolute velocity (CAV), arias intensity (Ia), and significant duration. In addition to addressing random effects associated with ground motion regression, such as inter-event, inter-site, inter-locality, and inter-region variabilities, the current study also aims at reducing the standard deviations (STDs) of the GMMs through development of a hybrid non-parametric GMM. The hybrid model is derived through an ensemble-weighted method of five non-parametric machine learning models: shallow neural network, deep neural network (DNN), gated recurrent unit (GRU), support vector, and random forest (RF) regression techniques; with weights based on model performances. The resulting hybrid model, which also accounts for epistemic uncertainty, is compared against other regional models and is found superior for all output variables. The inter-event, inter-site, inter-locality, and inter-region deviations, and total ergodic sigma of PSA for the ensemble model lies between 0.3164–0.4478, 0.4156–0.5339, 0.1449–0.3687, 0.0819–0.2421, and 0.668–0.8545, respectively. The coefficient of determination (R2) between predicted and recorded values lies between 0.8435–0.9114 for all the output variables.

This study presents broadband ground motions for the Indo-Gangetic basin, a large sedimentary basin in India, for potential future great (Mw 8.5) Himalayan earthquakes. We use a recently developed 3D earth structure model of the basin as an input to simulate low-frequency ground motion (0–0.5 Hz). These ground motions are further combined with high-frequency scattering waveforms by using a hybrid approach, thus yielding broadband ground motions (0–10 Hz). We calibrate the 3D model and scattering parameters by comparing the simulated ground motions against available recorded data for two past earthquakes in Himalaya. Our approach accounts for the physics of interaction between the scattered seismic waves with deep basin sediments. Our results indicate that the ground motion intensities exhibit frequency-dependent amplification at various basin depths. We also observe that in the event of a great earthquake, the ground motion intensities are larger at deep basin sites near the source and exhibit an attenuating trend over distance similar to the ground motion models. The extreme ground motion simulations performed in our study reveal that the national building codes may not provide safe recommendations at deep basin sites, especially in the near field region. The period-dependent vertical-to-horizontal spectral ratio deviates from the code-recommended constant 2/3 at least up to 6 s at these sites.

The present study aims at developing a prediction model for ground motion intensity measures using the artificial neural network (ANN) technique for active shallow crustal earthquakes in India. The database for the study consists of 659 ground motion records collected from 138 earthquakes recorded by various seismic networks in the study region. Owing to the lack of near-field data, we have added 116 records from seven earthquakes over a distance < 30 km and M > 6 from the NGA database. The developed model predicts 21 ground motion parameters (GMPs) in both horizontal and vertical directions, with input predictor variables of magnitude (M), hypocentral distance (R), site condition (S), and flag for the region (f). A multi-layer perceptron (MLP), with a total of 276 unknowns, constitutes the architecture of the model. The residuals associated with the GMPs are analyzed in detail to aid in hazard calculations. In addition, a comparison of the developed model with global relations is performed. Further, the model is demonstrated by performing seismic hazard analysis for GMPs for 2% and 10% probability of exceedance in 50 years. The ANN model is a first version and has to be improved as more strong motion data becomes available for the region. The developed ground motion model must be combined along with other global models in seismic hazard analysis.

The present study aims at developing a ground motion model (GMM) for the 5% damped horizontal spectral acceleration, using regression analysis of strong motion records available for the Western Himalayan region. In addition to developing a model using just the regional data, the study also explores three different methods to derive a GMM that can circumvent the limitation of near field data shortage in the Himalayas. The alternatives explored in this study include calibrating a global model to the regional dataset; deriving a GMM by appending a dense near field foreign dataset to that of the regional data; and deriving a near source correction factor to the regional model. These models are applicable for shallow crustal earthquakes of magnitudes between Mw 4.0–7.9 and depth up to 45 km over distances up to 960 km. The efficacy of each of these models is established through comparison with the recorded data and with other regional GMMs. Moreover, the best model among the four proposed GMMs is verified through derivation of rankings based on quantitative analysis of residuals that were obtained between the observations and the respective estimates.