SeisBench Models

The SeisBench model module provides a collection of deep learning models for seismic data analysis. These models cover a wide range of tasks including phase picking, earthquake detection, denoising, and depth phase identification. SeisBench offers both classical waveform models for standard seismic data and models for Distributed Acoustic Sensing (DAS) data. SeisBench models support accelerated processing on GPUs and Apple’s MPS.

All models in SeisBench expose two unified interfaces, making it easy to load pretrained weights, apply models to new data, and train models on custom datasets. With the first interface, implementing the annotate and classify functions, models can be applied directly to data in convenient formats such as ObsPy streams for waveform data or xdas arrays for DAS data. The second interface provides direct, low-level access to the Pytoch implementation of the models underneath and can be used for training the models.

Hint

If you are looking to quickly apply pretrained models to your seismic data, check out the Examples section for practical tutorials on using these models for phase picking, detection, and other tasks.

• SeisBench Waveform Models
• SeisBench DAS Models
• Loading pretrained models
• Speeding up model application