A Catalog of 971 FR-I Radio Galaxies from the FIRST Survey via Hybrid Deep Learning and Ridgeline Flux Density Distribution Analysis

TL;DR

The paper delivers a large, robust catalog of FR-I radio galaxies (971 total; 845 new) selected from the VLA FIRST survey using a two-stage pipeline: a deep-learning-based detector (Mask Transfiner/RGCMT) to identify candidates, followed by a ridgeline flux density distribution analysis to confirm edge-darkened jets. Host galaxies are matched with DESI LS DR10 and redshifts are assembled from spectroscopic and photometric sources, resulting in 813 FR-I hosts with $M_r$ and $M_{ m BH}$ estimates, and a detailed breakdown of LERG/HERG classifications. The study provides comprehensive radio properties (spectral indices, LAS, LLS, $P_{1400}$) and host properties (colors, luminosities, black hole masses) and shows that a simple power-based FR class separation does not cleanly distinguish FR-Is from FR-IIs in the expanded sample. The catalog, with its substantial size and multi-wavelength characterization, advances statistical studies of FR-I populations and offers a valuable training set for future AI-based morphological classification in radio astronomy, with redshifts up to $z=2.307$ and LLS predominantly below 700 kpc.

Abstract

We present a catalog of 971 FR-I radio galaxies (FR-Is) identified from the Very Large Array Faint Images of the Radio Sky at Twenty-Centimeters (FIRST) survey. The identifications were made using a hybrid method that combines deep learning with ridgeline flux density distribution analysis. Among these sources, 845 are new discoveries. The catalog comprises sources characterized by edge-darkened double jets, an absence of significant bent morphology, and angular sizes ranging from 23 to 159 arcseconds. Optical and/or infrared counterparts have been identified for 813 FR-Is. Among these, the host galaxies are predominantly (88.1\%) red galaxies, with the remainder (11.9\%) being blue galaxies; notably, most blue galaxies exhibit high radio power. The FR-I sample spans a radio power range of $1.20 \times 10^{21} \leq P_{\rm 1400} \leq 3.55 \times 10^{27} \, {\rm W\,Hz}^{-1}$ at 1400 MHz and reaches redshifts up to $z = 2.307$. The host galaxies have $r$-band absolute magnitudes in the range $-24 \lesssim M_r \lesssim -20$ mag. For the 512 FR-Is with estimates, the black hole masses fall within $10^7 \lesssim M_{\rm BH} \lesssim 7.94 \times 10^9 \, M_{\odot}$. Based on optical emission-line ratios and mid-infrared colors, spectroscopic classification shows that 571 hosts are low-excitation radio galaxies (LERGs) and 59 are high-excitation radio galaxies (HERGs).

Figures (10)

• Figure 1: This figure shows how the ridgeline flux density distribution-based identification method applies to two typical morphological structures (upper and middle) in FR-I sources, as well as to one hybrid morphology radio source (bottom). Left panel: FR-I source from FIRST data with the source's boundary points (green) derived from the predicted mask, the peak flux density position (cyan $\times$), and the ridgeline (blue dotted line). Right panel: the flux density distribution at the ridgeline points. The x-axis represents the distance of each ridgeline point from the origin point, which is defined as the ridgeline point closest to the location of the brightest peak flux density. The green rectangular boxes are centered on the x-axis at positions corresponding to the peaks in flux density, with their widths equal to the local beam sizes of the FIRST image data. The red dashed line represents a flux density equal to 3 times the local RMS noise.
• Figure 2: An example of the final host galaxy candidate identified for a FR-I radio galaxy from our sample. All matched host candidates are marked with '+' or '$\times$', where those with signal-to-noise ratio (S/N) $\geq$ 5 in any of the $g$, $r$, $z$, W1, or W2 bands are marked with '$\times$' and others with '+'. The final host candidate is marked with a green '$\times$'. The blue bounding box represents the predicted region of the FR-I source, while the blue '$\times$' markers indicate candidates located outside the predicted bounding box.
• Figure 3: The $g-r$ vs. $z-$W1 color-color diagram for the 142 undetermined host candidates, classified by types. Labels PSF, REX, EXP, DEV, and SER correspond to host candidates with the following light-profile types: Point Spread Function, Round Exponential, deVaucouleurs, Exponential, and Sérsic, respectively. The label 'ALL' designates the combined set of host candidates, incorporating all five morphological types.
• Figure 4: Radio-optical overlay of 12 newly identified FR-Is from our catalog, showing VLA FIRST radio continuum contours overlaid on DESI LS $r$-band optical images (plasma colour scale). The contour levels start at 3$\sigma_{\rm rms}$ (where $\sigma_{\rm rms}$ is the local RMS noise) and increase by multiplicative factors of $\sqrt{2}$. The host galaxy positions are indicated by green $\times$ symbols.
• Figure 5: Distribution of spectral indexes ($\alpha_{\rm 150}^{\rm 1400}$) for the FR-I sample. The black dotted curve shows a Gaussian fit to the distribution (mean=-0.59$\pm$0.01, standard deviation=0.20), with the red dashed line marking the mean value.