Continuum source catalog for the first APERTIF data release

TL;DR

This paper presents a new machine learning method to model the primary beam in radio astronomy, applies it to the first Apertif survey data, and releases a comprehensive source catalog with cross-matched data and transient source analysis.

Contribution

It introduces a Gaussian process regression approach for primary beam modeling in radio surveys, enhancing data accuracy and source detection capabilities.

Findings

Produced a catalog of 249,672 radio sources, many detected for the first time.

Cross-matched sources with NVSS and LOFAR catalogs, enabling spectral and redshift analysis.

Identified long-term transient sources with significant flux changes over 25 years.

Abstract

The first data release of Apertif survey contains 3074 radio continuum images covering a thousand square degrees of the sky. The observations were performed during August 2019 to July 2020. The continuum images were produced at a central frequency 1355 MHz with the bandwidth of $\sim$150 MHz and angular resolution reaching 10". In this work we introduce and apply a new method to obtain a primary beam model using a machine learning approach, Gaussian process regression. The primary beam models obtained with this method are published along with the data products for the first Apertif data release. We apply the method to the continuum images, mosaic them and extract the source catalog. The catalog contains 249672 radio sources many of which are detected for the first time at these frequencies. We cross-match the coordinates with the NVSS, LOFAR/DR1/value-added and LOFAR/DR2 catalogs resulting in 44523, 22825 and 152824 common sources respectively. The first sample provides a unique opportunity to detect long term transient sources which have significantly changed their flux density for the last 25 years. The second and the third ones combined together provide information about spectral properties of the sources as well as the redshift estimates.