.. This file is generated by scripts/render_dataset_docs.py. Do not edit by hand.

SeisBench
=========

Synthetic-backed waveform picking adapter aligned to the SeisBench ecosystem.

Overview
--------

SeisBench is the public waveform adapter used by PyHazards for the earthquake picking path.

The current implementation is synthetic-backed, but it preserves the picking task shape, labels, and metrics expected by the shared earthquake evaluator.

At a Glance
-----------

.. list-table::
   :widths: 28 72
   :stub-columns: 1

   * - Provider
     - SeisBench ecosystem surfaced through a PyHazards adapter
   * - Hazard Family
     - Earthquake
   * - Source Role
     - Waveform Benchmark
   * - Coverage
     - Benchmark-aligned earthquake phase-picking samples
   * - Geometry
     - Multichannel waveform windows
   * - Spatial Resolution
     - Benchmark-defined waveform channels and sample windows
   * - Temporal Resolution
     - Short waveform windows with phase-pick targets
   * - Update Cadence
     - Generated locally for smoke and benchmark-alignment runs
   * - Period of Record
     - Synthetic-backed benchmark adapter
   * - Formats
     - PyTorch tensors via the dataset registry
   * - Registry Entry
     - ``seisbench_waveforms``

Data Characteristics
--------------------

- Multichannel waveform windows paired with P- and S-arrival sample targets.
- Registry-backed benchmark adapter rather than a raw external waveform ingestion path.
- Intended for phase-picking validation and smoke tests.

Typical Use Cases
~~~~~~~~~~~~~~~~~

- PhaseNet and EQNet smoke tests.
- Shared earthquake benchmark runs on picking tasks.
- Regression checks for waveform-based seismic models.

Access
------

Use the links below to access the upstream source or its public documentation.

- `SeisBench paper <https://joss.theoj.org/papers/10.21105/joss.04418>`_
- `SeisBench repository <https://github.com/seisbench/seisbench>`_

PyHazards Usage
---------------

Use this adapter when you want the public SeisBench-aligned waveform surface exposed by the earthquake benchmark.

Registry Workflow
~~~~~~~~~~~~~~~~~

Primary dataset name: ``seisbench_waveforms``

.. code-block:: python

   from pyhazards.datasets import load_dataset

   data = load_dataset(
       "seisbench_waveforms",
       micro=True,
       channels=3,
       length=256,
   ).load()

   train = data.get_split("train")
   print(train.inputs.shape, train.targets.shape)

Related Coverage
~~~~~~~~~~~~~~~~

**Benchmarks:** :doc:`Earthquake Benchmark </benchmarks/earthquake_benchmark>`, :doc:`SeisBench </benchmarks/seisbench>`

**Representative Models:** :doc:`PhaseNet </modules/models_phasenet>`, :doc:`EQNet </modules/models_eqnet>`

Inspection Workflow
-------------------

This dataset is currently surfaced as a registry-backed benchmark adapter,
so there is no standalone inspection CLI documented for it.

Notes
-----

- This is a synthetic-backed benchmark adapter rather than a full SeisBench ingestion pipeline.

Reference
---------

- `SeisBench - A Toolbox for Machine Learning in Seismology <https://joss.theoj.org/papers/10.21105/joss.04418>`_ (`repo <https://github.com/seisbench/seisbench>`__).