English
 
Privacy Policy Disclaimer
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT

Released

Journal Article

Ground-Motion Modeling as an Image Processing Task: Introducing a Neural Network Based, Fully Data-Driven, and Nonergodic Approach

Authors
/persons/resource/lilienka

Lilienkamp,  Henning
2.6 Seismic Hazard and Risk Dynamics, 2.0 Geophysics, Departments, GFZ Publication Database, Deutsches GeoForschungsZentrum;

/persons/resource/svs

von Specht,  S.
0 Pre-GFZ, Departments, GFZ Publication Database, Deutsches GeoForschungsZentrum;

/persons/resource/gweather

Weatherill,  Graeme
2.6 Seismic Hazard and Risk Dynamics, 2.0 Geophysics, Departments, GFZ Publication Database, Deutsches GeoForschungsZentrum;

Caire,  Giuseppe
External Organizations;

/persons/resource/fcotton

Cotton,  Fabrice
2.6 Seismic Hazard and Risk Dynamics, 2.0 Geophysics, Departments, GFZ Publication Database, Deutsches GeoForschungsZentrum;

External Ressource
No external resources are shared
Fulltext (public)

5010751.pdf
(Postprint), 3MB

5010751_Erratum.pdf
(Postprint), 211KB

Supplementary Material (public)
There is no public supplementary material available
Citation

Lilienkamp, H., von Specht, S., Weatherill, G., Caire, G., Cotton, F. (2022): Ground-Motion Modeling as an Image Processing Task: Introducing a Neural Network Based, Fully Data-Driven, and Nonergodic Approach. - Bulletin of the Seismological Society of America, 112, 3, 1565-1582.
https://doi.org/10.1785/0120220008


Cite as: https://gfzpublic.gfz-potsdam.de/pubman/item/item_5010751
Abstract
We construct and examine the prototype of a deep learning-based ground-motion model (GMM) that is both fully data driven and nonergodic. We formulate ground-motion modeling as an image processing task, in which a specific type of neural network, the U-Net, relates continuous, horizontal maps of earthquake predictive parameters to sparse observations of a ground-motion intensity measure (IM). The processing of map-shaped data allows the natural incorporation of absolute earthquake source and observation site coordinates, and is, therefore, well suited to include site-, source-, and path-specific amplification effects in a nonergodic GMM. Data-driven interpolation of the IM between observation points is an inherent feature of the U-Net and requires no a priori assumptions. We evaluate our model using both a synthetic dataset and a subset of observations from the KiK-net strong motion network in the Kanto basin in Japan. We find that the U-Net model is capable of learning the magnitude–distance scaling, as well as site-, source-, and path-specific amplification effects from a strong motion dataset. The interpolation scheme is evaluated using a fivefold cross validation and is found to provide on average unbiased predictions. The magnitude–distance scaling as well as the site amplification of response spectral acceleration at a period of 1 s obtained for the Kanto basin are comparable to previous regional studies.