AI4EPS/QuakeFlow

QuakeFlow: A Scalable Deep-learning-based Earthquake Monitoring Workflow with Cloud Computing

QuakeFlow: A Scalable Machine-learning-based Earthquake Monitoring Workflow with Cloud Computing

Overview

QuakeFlow is a scalable deep-learning-based earthquake monitoring system with cloud computing. It applies the state-of-art deep learning/machine learning models for earthquake detection. With auto-scaling enabled on Kubernetes, our system can balance computational loads with computational resources.

Current Modules

Models

• DeepDenoiser: (paper) (example)
• PhaseNet: (paper) (example)
• GaMMA: (paper) (example)
• HypoDD (paper) (example)
• More models to be added. Contributions are highly welcomed!

Data stream

• Plotly: ui.quakeflow.com
• Kafka
• Spark Streaming

Data process

• Colab example
• Kubeflow: (example)

Deployment

QuakeFlow can be deployed on any cloud platforms with Kubernetes service.

• For google cloud platform (GCP), check out the GCP README.
• For on-premise servers, check out the Kubernetes README.