Ground Motion Dataset Verification and Validation, Insights Into Unexpected Sources of Uncertainties Associated to Ground Motion Modeling
Description:
During the last years, the collection of weak motions and the implementation of local ground motion datasets has grown in popularity in low-to-moderate seismicity regions. Such ground motion datasets are then used, for example, to adjust existing Ground Motion Models (GMMs) developed in active regions of the world to the local context, to develop specific non-ergodic GMM or to perform Generalized Inversions (GIT) and assess seismological models.
Regardless of the final application, careful Verification and Validation (V&V) of the collected and processed data are essential to ensure the quality of the dataset and therefore the robustness of the following analyses.
In low-to-moderate seismicity regions as continental France, available ground motion recordings are mostly related to small magnitude earthquakes, for which the information associated to the earthquake source parameters (3D location and magnitude) is often affected by large uncertainties. Moreover, to obtain the largest usable frequency range over the largest number of records, ground motions recorded at both, accelerometers and broad-band seismometers are usually collected. However, information about seismic station installation conditions (housing, possible evolutions in station installation over time…) is most often lacking.
In this work we present the V&V strategy set up while implementing the EPOS-France ground motion dataset (Traversa et al., 2020 and Buscetti et al., 2024) for continental France. Using residual analyses with respect to an ad-hoc simple Ground Motion Model (GMM), inconsistencies and inaccuracies affecting recorded data and associated metadata could be identified and either, were corrected or led to data rejection. This analysis also allowed to assess the impact of the quality of the metadata (e.g. of the earthquake catalog) on the variability associated to a GMM developed based on the considered dataset.
This work highlights that significant epistemic uncertainties can still be hidden into what is interpreted as GMMs aleatory variability when datasets do not undergo V&V careful strategies.
Session: Modern Waveform Processing and Engineering Datasets - Accessibility, Quality Control, and Metadata [Poster]
Type: Poster
Date: 4/17/2025
Presentation Time: 08:00 AM (local time)
Presenting Author: Paola
Student Presenter: No
Invited Presentation:
Poster Number: 53
Authors
Paola Traversa Presenting Author Corresponding Author paola.traversa@edf.fr Electricity of France |
Margaux Buscetti margaux.buscetti@cea.fr French Alternative Energies and Atomic Energy Commission, University of Grenoble Alpes, Institute of Earth Sciences |
Vincent Perron vincent.perron@cea.fr French Alternative Energies and Atomic Energy Commission, University of Grenoble Alpes, Institute of Earth Sciences |
Pauline Rischette pauline.rischette@cea.fr French Alternative Energies and Atomic Energy Commission, University of Grenoble Alpes, Institute of Earth Sciences |
Fabrice Hollender fabrice.hollender@cea.fr French Alternative Energies and Atomic Energy Commission, University of Grenoble Alpes, Institute of Earth Sciences |
Pierre Arroucau pierre.arroucau@edf.fr Electricity of France |
|
|
|
Ground Motion Dataset Verification and Validation, Insights Into Unexpected Sources of Uncertainties Associated to Ground Motion Modeling
Category
Modern Waveform Processing and Engineering Datasets - Accessibility, Quality Control, and Metadata