A Systematic Visualisation Framework for Radio-Imaging Pipelines

Abstract

Pipelines for calibration and imaging of radio interferometric data produce many intermediate images and other data products (gain tables, etc.) These often contain valuable information about the quality of the data and the calibration, and can provide the user with valuable insights, if only visualised in the right way. However, the deluge of data that we're experiencing with modern instruments means that most of these products are never looked at, and only the final images and data products are examined. Furthermore, the variety of imaging algorithms currently available, and the range of their options, means that very different results can be produced from the same set of original data. Proper understanding of this requires a systematic comparison that can be carried out both by individual users locally, and by the community globally. We address both problems by developing a systematic visualisation framework based around Jupyter notebooks, enriched with interactive plots based on the Bokeh and Datashader visualisation libraries.