Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TimesNet-Gen: Deep Learning-based Site Specific Strong Motion Generation

Baris Yilmaz, Bevan Deniz Cilgin, Erdem Akagündüz, Salih Tileylioglu

TL;DR

This work tackles site-specific strong-motion generation by introducing TimesNet-Gen, a time-domain conditional generator that leverages station IDs and a latent bottleneck to produce diverse yet site-faithful records. The model is trained in two phases and evaluated using HVSR-based site frequencies and distribution-based metrics on the AFAD dataset, showing strong station-wise alignment and realism. A comparison with a spectrogram-based VAE baseline demonstrates TimesNet-Gen's superior ability to reproduce station-specific spectral content and HVSR characteristics. The approach enables data-driven site-specific hazard assessment and holds promise for downstream seismic tasks such as phase picking and ground-motion parameter estimation.

Abstract

Effective earthquake risk reduction relies on accurate site-specific evaluations. This requires models that can represent the influence of local site conditions on ground motion characteristics. In this context, data driven approaches that learn site controlled signatures from recorded ground motions offer a promising direction. We address strong ground motion generation from time-domain accelerometer records and introduce the TimesNet-Gen, a time-domain conditional generator. The approach uses a station specific latent bottleneck. We evaluate generation by comparing HVSR curves and fundamental site-frequency $f_0$ distributions between real and generated records per station, and summarize station specificity with a score based on the $f_0$ distribution confusion matrices. TimesNet-Gen achieves strong station-wise alignment and compares favorably with a spectrogram-based conditional VAE baseline for site-specific strong motion synthesis. Our codes are available via https://github.com/brsylmz23/TimesNet-Gen.

TimesNet-Gen: Deep Learning-based Site Specific Strong Motion Generation

TL;DR

This work tackles site-specific strong-motion generation by introducing TimesNet-Gen, a time-domain conditional generator that leverages station IDs and a latent bottleneck to produce diverse yet site-faithful records. The model is trained in two phases and evaluated using HVSR-based site frequencies and distribution-based metrics on the AFAD dataset, showing strong station-wise alignment and realism. A comparison with a spectrogram-based VAE baseline demonstrates TimesNet-Gen's superior ability to reproduce station-specific spectral content and HVSR characteristics. The approach enables data-driven site-specific hazard assessment and holds promise for downstream seismic tasks such as phase picking and ground-motion parameter estimation.

Abstract

Effective earthquake risk reduction relies on accurate site-specific evaluations. This requires models that can represent the influence of local site conditions on ground motion characteristics. In this context, data driven approaches that learn site controlled signatures from recorded ground motions offer a promising direction. We address strong ground motion generation from time-domain accelerometer records and introduce the TimesNet-Gen, a time-domain conditional generator. The approach uses a station specific latent bottleneck. We evaluate generation by comparing HVSR curves and fundamental site-frequency distributions between real and generated records per station, and summarize station specificity with a score based on the distribution confusion matrices. TimesNet-Gen achieves strong station-wise alignment and compares favorably with a spectrogram-based conditional VAE baseline for site-specific strong motion synthesis. Our codes are available via https://github.com/brsylmz23/TimesNet-Gen.

Paper Structure

This paper contains 16 sections, 6 equations, 6 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. TimesNet-Gen
  5. Conditioning
  6. Sampling
  7. Variational Autoencoders
  8. Dataset
  9. Two-Phase Training
  10. Evaluation
  11. Experiments and Results
  12. Sample Results
  13. Spectral Analysis
  14. $f_0$ Distribution Confusion Matrices
  15. Conclusions
  16. ...and 1 more sections

Figures (6)

  • Figure 1: TimesNet-Gen architecture.
  • Figure 2: TimesNet-Gen sampling via $k$-sample encoder-feature averaging within a station pool.
  • Figure 3: Left: Real samples from the dataset. Right: similar TimesNet-Gen generated. Bottom: Corresponding Fourier amplitude spectra. Results are shown for three different stations.
  • Figure 4: $f_0$ distributions for selected stations.
  • Figure 5: Average HVSR curves for selected stations.
  • ...and 1 more figures