van den Ende M. P. A. auteur aut IRD Ampuero Jean-Paul auteur aut IRD text journalArticle eng text/pdf born digital access Most seismological analysis methods require knowledge of the geographic location of the stations comprising a seismic network. However, common machine learning tools used in seismology do not account for this spatial information, and so there is an underutilized potential for improving the performance of machine learning models. In this work, we propose a graph neural network (GNN) approach that explicitly incorporates and leverages spatial information for the task of seismic source characterization (specifically, location and magnitude estimation), based on multistation waveform recordings. Even using a modestly-sized GNN, we achieve model prediction accuracy that outperforms methods that are agnostic to station locations. Moreover, the proposed method is flexible to the number of seismic stations included in the analysis and is invariant to the order in which the stations are arranged, which opens up new applications in the automation of seismological tasks and in earthquake early warning systems. specialized earthquake source seismology machine learning graphs 066 122 47 17 e2020GL088690 [11 ] 2020 0094-8276 https://www.documentation.ird.fr/hor/fdi:010079752 10.1029/2020gl088690 0094-8276 [F B010079752] https://www.documentation.ird.fr/hor/fdi:010079752 https://www.documentation.ird.fr/intranet/publi/2020/10/010079752.pdf Accès réservé (Intranet de l'IRD) IRD - Base Horizon / Pleins textes 2020-11-12 2025-02-24 fdi:010079752 fre