[DTC-E2] WF7201: AI-based Seismic Catalogue Generation (QuakeFlow)

Version 1

Overview

This repository contains the Common Workflow Language (CWL) and RO-Crate metadata definition for WF7201, the AI-based Seismic Catalogue Generation workflow of DTC-E2, developed under the DT-GEO project.

WF7201 integrates advanced machine learning models for seismic signal detection and catalog generation within a unified, scalable computational framework. It builds upon QuakeFlow — a deep-learning-based earthquake monitoring pipeline — to automate waveform ingestion, event detection, association, location, and catalog generation.

The workflow enables near-real-time seismic catalogue creation from waveform data, supporting downstream digital twin components for earthquake analysis and forecasting.

Workflow Structure

The workflow integrates several sequential stages for data ingestion, AI-driven detection, and catalog production:

1. Workflow Inputs

   • WAVEFORMS – Continuous seismic waveform data from multiple stations.
   • DT7205 (AlpArray Catalogue) – Example dataset from the AlpArray Network (EIDA download via ObsPy).

2. Processing Steps

   • ST720101: Waveform ingestion and pre-processing (e.g., filtering, normalization).
   • ST720102: AI-based phase picking using neural network models from QuakeFlow.
   • ST720103: Event association to link picks to potential earthquake events.
   • ST720104: Event location and magnitude estimation.
   • ST720105: High-resolution catalogue generation and export.

3. Workflow Output

   • HIGH_RESOLUTION_CATALOG – AI-generated earthquake catalogue containing high-precision event locations and magnitudes.