VIPCALs: A fully automated calibration pipeline for very long baseline interferometry data

TL;DR

VIPCALs tackles the problem of time-consuming, manual VLBI post-correlation calibration by delivering a fully automated, end-to-end pipeline that reproduces the standard AIPS workflow in an unsupervised manner. Implemented in Python via ParselTongue, VIPCALs handles data loading, TEC-based ionospheric corrections, geometric refinements, fringe fitting, and amplitude/bandpass calibration, plus automatic reference-antenna and calibrator selection, with comprehensive diagnostics. Validation on 1000 SMILE VLBA sources shows 955 fully calibrated datasets, with over 91% achieving fringe detections in at least half of the solution intervals and a median visibility retention of 0.87; the average per-dataset runtime is under 10 minutes on a single core, indicating strong scalability. This automation reduces the barrier to large-scale VLBI surveys, enabling reproducible, community-scale analyses of archival data and paving the way for automated imaging and large statistical studies of radio-loud sources.

Abstract

Very long baseline interferometry (VLBI) is a powerful technique that can achieve sub-milliarcsecond resolution. However, it requires complex and often manual post-correlation calibration to correct for instrumental, geometric, and propagation-related errors. Unlike connected-element interferometers, VLBI arrays typically provide raw visibilities rather than science-ready data, and existing pipelines are largely semi-automated and reliant on user supervision. We present VIPCALs, a fully automated, end-to-end calibration pipeline for continuum VLBI data that operates without human intervention or prior knowledge of the dataset. Designed for scalability to thousands of sources and heterogeneous archival observations, VIPCALs addresses the needs of initiatives such as the Search for Milli-Lenses (SMILE) project. Implemented in Python using ParselTongue, VIPCALs reproduces the standard AIPS calibration workflow in a fully unsupervised mode. Besides the usual calibration tasks, the pipeline also performs automatic reference antenna selection, calibrator identification, and generates diagnostic outputs for inspection. We validated it on a representative sample of Very Long Baseline Array (VLBA) data corresponding to 1000 sources from the SMILE project. VIPCALs successfully calibrated observations of 955 of the test sources across multiple frequency bands. Over 91% of the calibrated datasets achieved successful fringe fitting on target in at least half of the solutions attempted. The median ratio of calibrated visibilities to initial total visibilities was 0.87. The average processing time was below 10 minutes per dataset, demonstrating both efficiency and scalability. VIPCALs enables robust, reproducible, and fully automated calibration of VLBI continuum data, significantly lowering the entry barrier for VLBI science and making large-scale projects like SMILE feasible.

Figures (7)

• Figure 1: Overview of the VIPCALs calibration pipeline. The orange block represents the preloading stage, the yellow block corresponds to loading and precalibration tasks, and the blue block encompasses the phase and amplitude calibration steps. Purple boxes indicate the AIPS tables generated at each stage of the workflow.
• Figure 2: Example of a single baseline before and after calibration, as displayed in the VIPCALs GUI. The plots show amplitude and phase as a function of frequency. The top figure corresponds to uncalibrated data, while the bottom figure shows the same baseline after applying the full calibration pipeline. Calibration flattens the phase response across the band and properly scales the amplitudes, correcting also for band-dependent variations.
• Figure 3: Properties of the VLBA test sample used in this study. Each panel corresponds to a different subsample: Top left: Full sample of 1000 sources. Top right: Subsample uniformly distributed in (VLA) flux (450 sources). Bottom left: Subsample uniformly distributed in exposure time (450 sources). Bottom right: Subsample randomly selected from observations at 15 and 22 GHz (100 sources). Each panel contains four subplots: (a) Histogram of source fluxes at 8.4 GHz from VLA observations of the CLASS survey Myers2003. (b) Histogram of on-target exposure times. (c) Distribution of observation dates. (d) Sky distribution in equatorial coordinates.
• Figure 4: Time needed to calibrate each of the 1372 VLBA observations using VIPCALs. Mean and median run times were 9.5 and 4.2 minutes, respectively.
• Figure 5: Breakdown of the relative time spent across different pipeline steps for the 1372 VLBA observations calibrated with VIPCALs. More than 50% of the calibration time is spent in I/O intensive steps (data loading and plotting).