The LOFAR Transients Pipeline (“TraP”) provides a means of searching a stream of N-dimensional (two spatial, frequency, polarization) image “cubes” for transient astronomical sources. The pipeline was developed specifically to address data produced by the LOFAR Transients Key Science Project, but is also applicable to a range of other instruments.

The TraP codebase provides the pipeline definition itself, as well as a number of supporting routines for source finding, measurement, characterization, and so on. Some of these routines are also available as stand-alone tools.

Key features

  • Customisable ‘quality-control’ steps to weed out bad images before processing.
  • Built-in sourcefinder optimized for radio-synthesis images.
  • Optional source-fitting constraints (only fit point-sources, avoid fitting sources near to edge of image, etc).
  • Source-association incorporates knowledge of positional errors (using the DeRuiter radius algorithm) - this means much less trial-and-error tweaking of source-association parameters when working with a new dataset.
  • ‘Skyregion’ tracking - this keeps a record of which parts of sky have been previously surveyed, and to what faint limit, allowing for better separation of real transients and marginal steady-source detections.
  • Variability metrics and cataloguing for every source - no need to choose transient-detection thresholds ahead of time, simply sort through the data after processing and judge for yourself.
  • Position monitoring and null-detection tracking. ‘Forced’ source-fits are attempted at positions where a source has been previously detected, or where a monitoring location has been manually specified, allowing for better detection of sources near to the faint limit.
  • All source measurements are stored in a standard SQL database; users can write their own custom data-extraction and analysis tools if desired.
  • Ready-made web-based data-exploration interface. TraP is accompanied by Banana, a web-based tool which allows astronomers to sort and search source-catalogues without requiring any local installation or programming. Provides interactive plots, links to external catalogue searches, and more.
  • Support for multiple data formats and telescopes. TraP can process both FITS and CASA MeasurementSet formats, and it is usually quite straightforward to add support for a new telescope.

High-level overview

The TraP consists of a tightly-coupled combination of a “pipeline definition” – effectively a Python script that marshals the flow of data through the system – with a library of analysis routines written in Python and a database, which not only contains results but also performs a key role in data processing.

Broadly speaking, as images are ingested by the TraP, a Python-based source-finding routine scans them, identifying and measuring all point-like sources. Those sources are ingested by the database, which associates them with previous measurements (both from earlier images processed by the TraP and from other catalogues) to form a lightcurve. Measurements are then performed at the locations of sources which were expected to be seen in this image but which were not detected. A series of statistical analyses are performed on the lightcurves constructed in this way, enabling the quick and easy identification of potential transients. This process results in two key data products: an archival database containing the lightcurves of all point-sources included in the dataset being processed, and community alerts of all transients which have been identified.

Exploiting the results of the TraP involves understanding and analysing the resulting lightcurve database. The TraP itself provides no tools directly aimed at this. Instead, the Transients Key Science Project has developed the Banana web interface to the database, which is maintained separately from the TraP. The database may also be interrogated by end-user developed tools using SQL.

Documentation layout

The documentation is split into four broad sections:

Getting Started
Provides a guide to installing the TraP and its supporting libraries on common platforms and some basic information to help get up and running quickly.
User’s Reference
Here we provide a complete description of all the functionality available in the TraP and describe the various configuration and setup options available to the end user.
Developer’s Reference
A guide to the structure of the codebase, the development methodologies, and the functionality available in the supporting libraries. This is of interest both to developers within the project and to those who want to build upon TraP functionality for their own purposes.
Stand-alone Tools
Some functionality developed for the TraP is also available in these simple, end-user focused tools.

This documentation focuses on the technical aspects of using the TraP: all the pipeline components are described, together with their user-configurable parameters and the systems which have been developed for connecting them together to form a pipeline. However, it does not provide detailed rationale for all of the scientific choices made in the pipeline design. It is the position of the author that achieving high quality results requires understanding both the technical and the scientific choices made. For help with the latter, the reader is referred to Swinbank et al.