# GWOSC O3b 16 kHz Gravitational-Wave Strain

## Facts

- Provider: LIGO Scientific Collaboration, Virgo Collaboration, KAGRA Collaboration (via GWOSC); Parquet by astro-legacy-archive
- License: CC-BY-4.0 (https://gwosc.org/acknowledgement/)
- Size: 1,238 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/astro-legacy-archive/gwosc-o3b-strain
- Access: public, readable without an account
- Category: Astronomy & Space
- Source: https://gwosc.org/data/
- Page: https://scigantic.com/archive/868cc1c0-2a50-5fc4-a4ca-0dd6f5023f32
- Keywords: Astronomy & Space, gravitational waves, time series, Parquet, Hugging Face

## About

Calibrated 16,384 Hz strain time series from the LIGO and Virgo gravitational-wave detectors for observing run O3b, as released by the Gravitational Wave Open Science Center (GWOSC), repackaged as Parquet by astro-legacy-archive. The card lists GPS extent 1256652800-1269366784, detectors H1 (Hanford), L1 (Livingston) and V1 (Virgo), 7,933 source GWOSC HDF5 files of 4,096 s each, contiguous spans 155 H1, 114 L1 and 143 V1, and a complete holding of 1,236 Parquet files (412 each of strain, DQmask and Injmask); strain is float64 and the masks are 1 Hz uint32. The repository uses about 3.3 TB (measured 2026-11-14). Earlier runs O1, O2 and O3a are separate rows. Each Parquet file is one contiguous observing span for one detector, named DETECTOR-GPS_START-GPS_STOP-COLUMN.parquet, and the three files per span hold strain, the data-quality mask and the hardware-injection mask; configurations such as H1-Strain or L1-DQmask load with Hugging Face datasets (streaming advised) or pyarrow, and per-file metadata records the timestream schema. Use it for gravitational-wave searches, glitch classification, denoising and noise characterisation. GWOSC data are licensed CC-BY-4.0; acknowledge the LIGO, Virgo and KAGRA collaborations as GWOSC asks.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("astro-legacy-archive/gwosc-o3b-strain")
print(ds)
```

## Data structure

File types in sampled files: Parquet

Example file: `H1-1256660992-1256706048-DQmask.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/868cc1c0-2a50-5fc4-a4ca-0dd6f5023f32
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/868cc1c0-2a50-5fc4-a4ca-0dd6f5023f32
- Over MCP: `get_schema_card` with id `868cc1c0-2a50-5fc4-a4ca-0dd6f5023f32` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Time series datasets](https://scigantic.com/topics/time-series)
- [Astronomy & Space datasets](https://scigantic.com/topics/astronomy-and-space)
