Doctoral thesis
OA Policy
English

TABASCAL: A Bayesian Approach to RFI Subtraction for Radio Interferometry

Directors

Kunz, Martin

Imprimatur date

2025-04-15

Abstract

Radio frequency interference (RFI) limits radio astronomy observations, while flagging discards contaminated data and reduces sensitivity. This thesis develops TABASCAL (Trajectory-Based RFI Subtraction and Calibration), a Bayesian method for removing post-correlation interference from radio-interferometric measurements. TABASCAL jointly models astronomical visibilities, antenna gains, and signals from multiple moving RFI sources. It combines trajectory and geometric-delay models with Gaussian-process descriptions of unknown, time-varying RFI amplitudes. This structure exploits the spatial signatures of astronomical and interfering sources while accommodating closure errors and using strong interference to constrain phase calibration. Tests on simulations show that TABASCAL can recover astronomical signals and calibration solutions from contaminated observations, producing results comparable to uncontaminated data and outperforming data flagging. The method is computationally demanding and the thesis focuses on simulated observations, but early real-data tests are promising. The work establishes a foundation for scalable RFI subtraction in future radio telescope facilities worldwide, including the Square Kilometre Array.

Keywords

  • Radio interferometry
  • Radio-frequency interference
  • Bayesian inference
  • Calibration
  • Gaussian processes
  • Square Kilometre Array

Citation (ISO format)

FINLAY, Christopher. TABASCAL: A Bayesian Approach to RFI Subtraction for Radio Interferometry. Thèse, 2025. doi: 10.13097/archive-ouverte/unige:195230

Main files (1)

Thesis

Secondary files (1)

Imprimatur

Identifiers

17views
176downloads

Technical informations

Creation

12/08/2026 11:08:02

First validation

13/08/2026 06:19:14

Update

13/08/2026 06:19:14

Status update

13/08/2026 06:19:14

Last indexation

13/08/2026 06:19:15
All rights reserved by Archive ouverte UNIGE and the University of Geneva
OpenAIRE Indexed & Validated