Broadband Ground-Motion Synthesis by Conditioning Denoising Diffusion Models on Low-Frequency Waveforms
Abstract
Seismic hazard and structural analyses require broadband ground-motion time histories, yet the high-frequency band (above 1~Hz) is the hardest to obtain: physics-based simulations are too costly at these frequencies, and stochastic methods lose the deterministic phase, directivity, and long-period coherence that control structural demand. We present HighFEM-2, a latent denoising-diffusion waveform model trained with rectified flow matching on a multi-modal diffusion transformer. It synthesizes three-component acceleration up to 50~Hz, conditioned on a low-frequency waveform below the source corner frequency and on magnitude, hypocenter and station location, and VS30. The low-frequency waveform supplies the deterministic source, path, and directivity, so the network learns only the regional high-frequency scattering. Conditioning each station on its own waveform lets HighFEM-2 generate records for all stations of an earthquake that share its source and low-frequency path terms. Trained on 195{,}138 Italian INSTANCE records (Mw 1.4--6.5, within 200~km), HighFEM-2 produces realistic waveforms: it reproduces observed peak amplitudes with small bias (0.042 ± 0.088 log units for PGA, 0.057 ± 0.090 for PGV) and matches observed envelopes, Fourier spectra, and integrated velocity and displacement. The conditioning waveform can come from a simulation or a filtered record, and ensembles of realizations sample the high-frequency variability given that input. HighFEM-2 thus provides fast, regionally calibrated broadband time histories for hazard and risk assessment, including near-fault scenarios where recordings are scarce.
Related
- Fault-conditioned Seismic Image Generation using Denoising Diffusion Probabilistic Modeling and Neural Style Transfer
- Earthquake Seismogram Denoising Across Time, Time‐Frequency, and Hybrid Domain Approaches
- Latent Diffusion Modeling of Two- and Three-dimensional Fracture Networks with Sparse Observation Conditioning
- Xaurora: Generative Weather Forecasting with Denoising Stochastic Interpolants from a Foundation Model Prior
- Deep Learning Denoising of Real SWOT Sea Surface Height Observations
BibTeX
@misc{palgunadi2026broadband,
title = {{Broadband Ground-Motion Synthesis by Conditioning Denoising Diffusion Models on Low-Frequency Waveforms}},
author = {Kadek Hendrawan Palgunadi and Simon Dirmeier and Maria Koroni and Laura Ermert and Men-Andrin Meier},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.04334},
eprint = {2610.04334},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.04334},
}
RIS
TY - UNPB
TI - Broadband Ground-Motion Synthesis by Conditioning Denoising Diffusion Models on Low-Frequency Waveforms
AU - Kadek Hendrawan Palgunadi
AU - Simon Dirmeier
AU - Maria Koroni
AU - Laura Ermert
AU - Men-Andrin Meier
PY - 2026
DA - 2026/10/03
PB - arXiv
DO - 10.48550/arxiv.2610.04334
UR - https://arxiv.org/abs/2610.04334
AB - Seismic hazard and structural analyses require broadband ground-motion time histories, yet the high-frequency band (above 1~Hz) is the hardest to obtain: physics-based simulations are too costly at these frequencies, and stochastic methods lose the deterministic phase, directivity, and long-period coherence that control structural demand. We present HighFEM-2, a latent denoising-diffusion waveform model trained with rectified flow matching on a multi-modal diffusion transformer. It synthesizes three-component acceleration up to 50~Hz, conditioned on a low-frequency waveform below the source corner frequency and on magnitude, hypocenter and station location, and V_S30. The low-frequency waveform supplies the deterministic source, path, and directivity, so the network learns only the regional high-frequency scattering. Conditioning each station on its own waveform lets HighFEM-2 generate records for all stations of an earthquake that share its source and low-frequency path terms. Trained on 195{,}138 Italian INSTANCE records (M_w 1.4--6.5, within 200~km), HighFEM-2 produces realistic waveforms: it reproduces observed peak amplitudes with small bias (0.042 ± 0.088 log units for PGA, 0.057 ± 0.090 for PGV) and matches observed envelopes, Fourier spectra, and integrated velocity and displacement. The conditioning waveform can come from a simulation or a filtered record, and ensembles of realizations sample the high-frequency variability given that input. HighFEM-2 thus provides fast, regionally calibrated broadband time histories for hazard and risk assessment, including near-fault scenarios where recordings are scarce.
ER -