Seismology

The case study targets natural hazard detection and forecasting in complex geophysical settings, showing how GenAI can integrate conventional seismology with emerging FO-DAS data. Datasets merge continuous seismic signals and FO-DAS streams with contextual inputs (weather, oceanography, human activity), enabling high-resolution monitoring but raising challenges of storage and scalability. The case study develops a seismology FM for source identification, event characterisation, seismic parameter estimation etc., generalising across volcanoes, glaciers, fault zones, landslides, and urban areas, supporting hybrid seismic event catalogues with explainable annotations and metadata. FO-DAS bottlenecks are addressed via AI-driven compression (e.g. variational autoencoders), selective archiving, and data augmentation to ensure scalable monitoring.

Validation with EPOS, Data Terra, and NFDI ensures EOSC interoperability. The pilot progresses from prototype FMs (M18) to a demonstrator (M36), showcasing real-time seismology/FO-DAS hazard monitoring within EOSC. Involvement of civil protection agencies, research infrastructures (EPOS, EMSO), and NGOs is foreseen through table-top exercises and advisory exchanges, ensuring that the pilot reflects operational priorities.