Ground motion model for Peninsular India using an artificial neural network

Yellapragada, Meenakshi; Vemula, Sreenath; Alne, Akshay; Raghukanth, S. T. G.

doi:10.1177/87552930221144330

Cited by 12 publications

(7 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Including NESS data is particularly beneficial due to its repository of near-field ground motion records from large seismic events, thus widening the dataset's scope (Meenakshi et al, 2023;Podili and Raghukanth, 2023). We have found that the model predictions are comparable with other vertical GMMs without bias.…”

Section: Introductionmentioning

confidence: 79%

ANN-Based Ground Motion and Physics-Based Broadband Models for Vertical Spectra

Sharma,

Arya,

Gade

et al. 2024

Preprint

View full text Add to dashboard Cite

Vertical ground motion plays a crucial role in the design of critical structures, with recorded vertical ground motions often surpassing their horizontal counterparts in the near-field. This study proposes a new simplified Ground Motion Model (GMM) for vertical spectra by combining comprehensive datasets from the NESS and NGA-West2 databases. The proposed Artificial Neural Network (ANN) architecture-based model requires only 288 unknowns to predict spectral accelerations ($Sa$) at 33 distinct periods ranging from 0 to 4 seconds. Notably, this model inherently captures known physical phenomena with reduced variability using a minimum number of unknowns compared to the GMMs existing literature, thus offering a valuable addition to current hazard estimation frameworks. Furthermore, recognizing the necessity for physics-based simulations in vertical ground motion analysis, we introduce a novel physics-based broadband model for vertical spectra using ANN methodology. The proposed broadband model exhibits better robustness due to the comprehensiveness of the dataset utilized and the inclusion of source path and site characteristics at the input layer. Additionally, the model effectively captures the physical trends with minimal deviation. Further, a comparison of the model's output with physics-based simulations for the 2008 Mw 6.9 Iwate earthquake and recorded data showcase its ability to accurately capture the non-ergodic nature of ground motion. The proposed models serve as essential tools for physics-based broadband simulations and hazard assessments in active shallow crustal regions.

show abstract

Section: Introductionmentioning

confidence: 79%

ANN-Based Ground Motion and Physics-Based Broadband Models for Vertical Spectra

Sharma,

Arya,

Gade

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…In addition to EQ detection, other factors like PGA, peak ground velocity (PGV), peak ground displacement (PGD), and active tectonics play crucial roles in assessing seismic hazards [53][54][55][56]. Various studies have employed ML models to predict these parameters [57]. For instance, Yao et al [36] used DL models to evaluate PGA predictions for dynamic rupture scenarios.…”

Section: Related Workmentioning

confidence: 99%

Employing Machine Learning for Seismic Intensity Estimation Using a Single Station for Earthquake Early Warning

Abdalzaher,

Soliman,

Krichen

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

An earthquake early-warning system (EEWS) is an indispensable tool for mitigating loss of life caused by earthquakes. The ability to rapidly assess the severity of an earthquake is crucial for effectively managing earthquake disasters and implementing successful risk-reduction strategies. In this regard, the utilization of an Internet of Things (IoT) network enables the real-time transmission of on-site intensity measurements. This paper introduces a novel approach based on machine-learning (ML) techniques to accurately and promptly determine earthquake intensity by analyzing the seismic activity 2 s after the onset of the p-wave. The proposed model, referred to as 2S1C1S, leverages data from a single station and a single component to evaluate earthquake intensity. The dataset employed in this study, named “INSTANCE,” comprises data from the Italian National Seismic Network (INSN) via hundreds of stations. The model has been trained on a substantial dataset of 50,000 instances, which corresponds to 150,000 seismic windows of 2 s each, encompassing 3C. By effectively capturing key features from the waveform traces, the proposed model provides a reliable estimation of earthquake intensity, achieving an impressive accuracy rate of 99.05% in forecasting based on any single component from the 3C. The 2S1C1S model can be seamlessly integrated into a centralized IoT system, enabling the swift transmission of alerts to the relevant authorities for prompt response and action. Additionally, a comprehensive comparison is conducted between the results obtained from the 2S1C1S method and those derived from the conventional manual solution method, which is considered the benchmark. The experimental results demonstrate that the proposed 2S1C1S model, employing extreme gradient boosting (XGB), surpasses several ML benchmarks in accurately determining earthquake intensity, thus highlighting the effectiveness of this methodology for earthquake early-warning systems (EEWSs).

show abstract

“…In this section, the PSA is further compared with the global GMMs that are developed for the NGA-East project along with the results of NGA-East GMM tool provided by the PEER group and Meenakshi et al 35 Predictions of PSA from those global GMMs are obtained from the electronic appendices of the PEER project. The global GMMs are developed for rock site conditions Class A type with V s 30-3000 m/s.…”

Section: Comparison With Existing Modelsmentioning

confidence: 99%

“…Both GMMs are developed using ANN. Further, the residuals are decomposed into between-event, site-to-site, and within-event-single-site residuals using the linear mixed effects method by following a similar procedure provided by Khosravikia et al 36 and Meenakshi et al 35 Further, the developed networks are verified by checking whether the residuals depict any bias against the predictor variables such as M w , R rup , and V s 30. The performance analysis and the parametric study have also been presented to confirm the efficacy of the developed GMMs.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Ground motion models for Fourier amplitude spectra and response spectra using Machine learning techniques

Meenakshi,

Sreenath,

STG

2023

Earthq Engng Struct Dyn

Self Cite

View full text Add to dashboard Cite

One of the main objectives in engineering seismology or in seismic hazard studies is to estimate the possible ground motion for a given earthquake scenario. In the sparse data regions mostly the ground motion models (GMM) are developed using either seismological models or hybrid empirical approaches. However, if these GMMs do not accommodate the regional seismological attributes, a large uncertainty in ground motion estimates is possible. To overcome this concern, scaling 5% damped Pseudo Spectral Acceleration (PSA) from Fourier amplitude spectra (FAS) proves to be physically consistent as it can capture both spatial and temporal characteristics of the ground motion. Hence, the present study aims to develop a GMM for PSA using FAS and significant duration as the predictor variables. However, since there are few GMMs available for FAS, the current study also aims to develop a GMM for FAS using the earthquake parameters as the predictor variables for intraplate regions. This article employs an Artificial Neural Network (ANN) to develop both GMMs using the Next Generation Attenuation (NGA)‐East database for both horizontal (Effective Amplitude spectra for FAS and RotD50 component for PSA) and vertical components. To verify the performance of the developed models, the residuals analysis and parametric studies have been performed. The parametric study shows that the GMMs can capture the magnitude and distance scaling consistent with the observations. Further, the PSA GMM compared with the global GMMs and it is observed that the predictions lie well within the median of all the available models, proving the models’ effectiveness in estimating the ground motion predictions for future data. The developed model can be used only within the considered ranges of the predicted variables such as rupture distances between [19.05–1000] km, the Mw ranges from [3.12–5.74] with Vs30 in the range of [209–2000] m/s.

show abstract

Ground motion model for Peninsular India using an artificial neural network

Cited by 12 publications

References 45 publications

ANN-Based Ground Motion and Physics-Based Broadband Models for Vertical Spectra

ANN-Based Ground Motion and Physics-Based Broadband Models for Vertical Spectra

Employing Machine Learning for Seismic Intensity Estimation Using a Single Station for Earthquake Early Warning

Ground motion models for Fourier amplitude spectra and response spectra using Machine learning techniques

Contact Info

Product

Resources

About