ATCA observations of Fermi unassociated sources



Fermi two-year catalogue 2FGL lists 1872 γ-ray objects. Among those sources that have associations, for 751 objects radio emission at 8 GHz was detected from compact parsec-scale regions. The catalogue of VLBI associations in html or in text formats provides positions of sources with sub-milliarcsecond accuracies, estimates of their correlated flux densities and provides radio images of the bases of their relativistic jets.

However, 573 sources still do not have associations. We set a goal to observe at 5 and 9 GHz the areas of 95% probabilities of 2FGL localization and find there all radio sources with flux densities down to 2 mJy for a search of possible associations. The broad goal of our project is to increase the number of radio–γ associations and to find a population of radio silent γ-ray sources or to demonstrate that such a population does not exist.

The program has several steps:

Those sources within 95% probabilities of 2FGL localization that show detectable emission at 8 GHz at parsec scales (0.1–10 mas) are considered as likely associations.


We observed areas within positions of all Fermi sources with < +10° with ATCA at 5.5 and 9 GHz simultaneously. Each source was observed in three scans of 30 seconds each separated by four hours.

This will allow us

The existence of a population of radio weak(silent) γ-ray sources has a potential for interesting discoveries.

Independently, a list of 215 sources with δ > 0° was observed with VLA.

Source list

Map of 2FGL unassociated sources:

Current list of targets. Sources marked with # in the first column were not associated in the original 2FGL release, but currently are considered associated owing to new observations.


Early results from this program were submitted for publication to Monthly Notices of the Royal Astronomical Society: L. Petrov, E. K. Mahony, P. G. Edwards, E. M. Sadler, F. K. Schinzel, D. McConnel, ATCA observations of Fermi unassociated sources, Mon. Not. of Roy. Astr. Soc., 2013, in press. Preprint:

Summary of results:

We observed our program for 29 hours on 19–20 September 2012 with ATCA in the H214 configuration. The data were recorded at 4.5–6.5 and 8.0–10.0 GHz. The correlator output (58 Gb) is available. We performed post-processing with Miriad package. Analysis procedure included calibration, imaging, and source analysis.

We found 626 objects in 267 fields. Of them, 210 objects have position offsets with respect to γ-ray 2FGL objects exceeding 2.448σ, i.e. beyond the areas of 95 per cent probability of their localisation, and they were discarded.

The rms of images is typically in a range of 0.15–0.25 mJy. The detection limit is 1.8 mJy for sources in the center of the field of view and 9 mJy at the edge of the field of view, at 6.5'. There are 23 fields that have a number of resolved objects.

We found 416 counterparts within 95 per cent probability of localization of 267 2FGL sources. For 69 2FGL sources more than one counterpart was found. Among 416 counterparts, 95 have a spectrum flatter than -0.5, 75 have a spectrum steeper than -0.5, and for 246 objects no spectral index has been determined. The flat spectrum sources are considered as tentatively associated with Fermi objects. For 107 sources the probability of association is more than 10 times exceeds the probability to find a background source with that flux density by chance. These flags help to identify a subset of more probable associations. However, taken alone they do not establish a firm association.


Here a set of 822 radio images at 5.5 and 9.0 GHz subbands is available. NB: Flux densities written in margins of images are not for beam attenuation. Point sources that were automatically detected are circled. They were inspected my analysts who flagged points that were either spurious detections, or too weak, or far beyond the center of the view. They are marked with red. Remaining objects are marked green.


The list of 416 sources falls into three categories:
  1. Category I: 142 objects detected at both 5 GHz and 9.0 GHz subbands within 2.7 arcmin of the pointing direction. We provide in Table 1 γ-ray source name; IAU name of detected radio source; tentative association status, estimates of its J2000 coordinates followed by $1\sigma$ uncertainties in arcseconds (uncertainty of right ascension is not scaled by cosδ factor), Δα and Δδ flux densities at 5 GHz and 9 GHz in mJy, F5 and F9, corrected for beam attenuation followed by their standard deviations, Δ5 and Δ9, and spectral index and its error, Sp and ΔSp. We provide distance of a source from the pointing direction D, followed by Nσ, the ratio of this distance to its standard deviation derived from reported 2FGL position localisation errors. If the source was associated with an object either from NVSS, SUMSS or MGPS-2 catalogues, its flux density at 1.4 GHz (NVSS) or 0.843 GHz (SUMSS and MGPS-2) is shown in column F1 followed by the 1 GHz catalogue code (N for NVSS, S for SUMSS, M for MGPS-2), followed by the source identifier in that catalogue. If the source was associated with WISE object, its WISE source ID is shown in the last column.

    Column 3 shows two flags: "a" if a source has likelihood ratio greater than 10, and therefore, considered a likely association, "e" if a source is extended, and "f" if a source has the spectral index flatter than -0.$.

  2. Category II: 210 objects detected only at 5 GHz subband within 6.5 arcmin of the pointing direction. We provide in Table2 estimates of flux density at 5 GHz corrected for beam attenuation. The contents of Table2 is similar to Table1, except columns F9 and Δ9 and which are missing. The spectral index is computed over the sub-band [4.58, 6.41] GHz. The spectral index estimates with uncertainties greater than 0.3 are omitted.
  3. Category III: 64 objects either detected beyond 6.5' of the pointing direction or detected only at 9 GHz band. Since calibration for beam attenuation becomes uncertain at large distances, we can only provide a low estimate of their flux density: 10 mJy. Table3 lists these sources. Its contents similar to Table2 except columns F5 and Δ5 which are missing.
Dependence of flux density of detected sources of categories I and II on galactic latitude. Flat spectrum sources are shown with green hollow circles. Other sources are shown with filled blue discs.

Future work

We were allotted three 15 hours blocks in September 2013:
2013-09-25, 21:30, 2013-09-26, 12:30, C2624, H214
2013-09-26, 21:30, 2013-09-27, 12:30, C2624, H214
2013-09-27, 21:30, 2013-09-28, 12:30, C2624, H214
We are going to re-observe with ATCA those sources that were detected at 5 GHz, but are not detected at 9~GHz, and had position offsets exceeding FWHM of the 9~GHz beam. We will re-observe sources listed in Table3 with pointing to their positions found from this survey as well. We are also going to re-observe sources that were detected at the edge of the beam or by the sidelobe.

In the second phase of the project we are going to re-observe 144 fields where we did not detected a source in the mosaic mode. This will allow us to detect all the sources in the 95 per cent probability of 2FGL localisation that are brighter than 2 mJy.

In the third phase of the project we will observe detected sources with LBA and determine correlated flux densities from regions smaller than 50 mas. This will allow us to associate detected radio sources with blazars.

Current status


Team members (in alphabetic order)

Phillip Edwards ATNF, Australia
Yury Kovalev Astrospace Center, Russia
Elizabeth Mahony the Netherlands Institute for Radio Astronomy
David McConnell ATNF, Australia
Tara Murphy University of Sydney, Australia
Leonid Petrov Astrogeo Center, USA
Elaine Sadler University of Sydney, Australia
Frank Schinzel University of New Mexico, USA
Greg Taylor University of New Mexico, USA


  1. Abdo A. A., 2012, ApJS, 119, 31
  2. Ackermann M., 2012, ApJ, 753, 83
  3. Kovalev Y. Y. 2009, ApJ, 707, L56
  4. Murphy T. et al. 2010, MNRAS, 420, 2403
  5. Petrov L., Kovalev Y. Y., Fomalont E. B., Gordon D. 2008, AJ, 136, 580

Back to astronomy projects

This web page was prepared by Leonid Petrov ()
Last update: 2016.03.09_15:09:15