Mining the VLA Archives for Radio Transients

AMERICAN JOURNAL OF UNDERGRADUATE RESEARCH VOL. 5, NO. 2 (2006)

Looking for the Elusive: Mining the VLA Archives for Radio Transients

Destry R. Saul and Geoffrey C. Bower Astronomy Department and Radio Astronomy Laboratory University of California - Berkeley 601 Campbell Hall Berkeley, California 94720 USA

Received: June 7, 2006 Accepted: July 24, 2006

ABSTRACT

Detecting extragalactic radio transients will provide valuable information on several astronomical phenomena. Orphan gamma-ray burst afterglows (OGRBAs) and radio supernovae detection will be important for the theories of stellar collapse while the detection of tidal flares and variable active galactic nuclei (AGN) will provide details on massive black holes and galactic centers. In this paper we present the results of a survey we conducted using archived observations from the Very Large Array (VLA). We did not detect any transients in this set of data and from this we computed an upper limit on the rate of radio transients above 400 μJansky at 0.23 transients per square degree per year. This rules out one of several theories on the rates of OGRBAs.

I. INTRODUCTION transients of 0.23 per square degree per year above 400 μJy.* This value agrees Transient radio sources are objects that did with some models and disagrees with not previously radiate detectably in the radio others. We will discuss our results in Section spectrum, but then begin to emit for a period III. of time, and then fade away. Of the several classes of predicted extragalactic radio a. A Brief Introduction to Extragalactic transients, only radio supernova explosions Radio Transient Theories and gamma-ray burst afterglows have been confirmed. We will discuss the different Current theories indicate that theories behind extragalactic radio extragalactic transient radio sources should transients in Section Ia. Detection of be detectable with the VLA. Figure 1 plots extragalactic radio transients will provide the theorized rates of different types of valuable information on the evolution of transients. Here we provide a brief stars, black holes, and galactic centers. With introduction to the theories of OGRBAs, tidal this motivation, we have begun to search flares, and radio supernovae. Gamma-ray through the VLA archive for radio transients. bursts are short flashes of gamma-rays that The survey that we present here used data are detected roughly once a day from taken with the VLA from 1982 to 1992. Our random directions. The origins of gamma- survey was superior to previous surveys due ray bursts are unknown, though there are to the combined sensitivity, area, and several theories for their production. duration of the survey. There were several Gamma-ray bursts are separated into two difficulties with the specific field of view of categories of ’fast-hard’ and ’slow-soft’ by our survey, and we will discuss these problems in Section II. We did not detect any viable transient candidates in our * In astrophysics, a Jansky is a unit of measurement of -2 -1 survey, and using these results we flux density, in units of watt · meter · hertz ; 1 Jansky is 10-26 W · m-2 · Hz-1 and is abbreviated Jy. A micro- calculated an upper limit to detectable Jansky is 10-6 Jy.

13 AMERICAN JOURNAL OF UNDERGRADUATE RESEARCH VOL. 5, NO. 2 (2006) the duration of the burst and the quality of

Figure 1. A plot of the different theories for radio transients. There are tidal flares (TF), radio supernova (RSNe), active galactic nuclei with intra-day variability (AGN IDV), and three models for OGRBAs: PL98 (Perna & Loeb [9] 1998), L02 (Levinson et al. [8] 2002), and TP02 (Tolani & Panaitescu [10] 2002). The different shaded regions represent the parameter spaces that observations with different arrays over different time durations would probe. The unshaded region is similar to the parameter space probed by our survey. Notice that the curve for OGRBA PL98 is well inside our survey. (Prepared by G. Bower)

the spectra. One theory for their cause is Following the detection of a gamma- the collapse of a massive star into a black ray bursts, there have been detections of hole [1]. The rate of gamma-ray bursts afterglows in the x-ray through radio provides information on the number and spectrum in the same location [3-5]. It is lifetime of massive stars. It would therefore thought that these afterglows are caused by be useful to have an accurate number for the deceleration of the ultra-relativistic jets the rate of gamma-ray bursts. Calculating emanating from the poles of the collapsing this rate has been complicated by theories star and the interaction of these jets with the that predict that gamma-ray bursts are surrounding gas. This afterglow emission produced in a tight beam emanating from should be isotropic and thus detectable the collapsing star’s poles [1, 2]. Thus we when the jets are not pointed toward us. can only detect gamma-ray bursts produced This detection of an afterglow without the by stars whose axis is along our line of sight. detection of the gamma-ray burst is an Identifying gamma-ray bursts that are not orphan gamma-ray burst afterglow. So far pointed directly at us involves the detection there have been no detections of OGRBAs of an orphan gamma-ray burst afterglow. in any wavelength even though there have

14 AMERICAN JOURNAL OF UNDERGRADUATE RESEARCH VOL. 5, NO. 2 (2006) been several attempts [6-8]. Searching for gas is ejected from the star before the OGRBAs in the radio spectrum should give explosion in a process that is still hard to the best estimates on the rate of OGRBAs model. A better understanding of radio because the radio emission is not affected supernovae would be useful in by extinction that would obscure other understanding the last stages of a star’s life. frequencies. The current models of There have been roughly two dozen OGRBAs differ significantly from each other detections of radio supernovae and more and make it difficult to calculate the would be helpful in refining the theories expected rate of detections. Based on one involved. New detections of radio model [9], our survey should have seen supernovae could also lead to better several OGRBAs, but another model identification of the exploding star’s pre- predicted that we would need ten times our supernova properties and it is also possible observed area to make a detection [10]. We that radio supernovae could be used to find will present this data in Section III. the distances to the galaxies where they Radio transients could also be occur [16]. produced by the central regions of other Besides the types of radio transients galaxies. It has been established that there discussed above, there is always the are massive black holes at the centers of possibility that we will discover a previously most galaxies, and that some of these are undetected type of radio source. By using ‘active’. This means that there are huge more sensitive observations of greater amounts of energy being emitted by the duration than in previous surveys we are central region of the galaxy. These are investigating regions of parameter space called active galactic nuclei (AGN). Some that have not been observed in the past. AGN have been shown to vary in intensity The detection of a new type of radio source on timescales less than a day [11]. Previous would be a very exciting discovery. surveys for AGN variability have been sensitive down to around 100 mJy. Our II. OBSERVATIONS survey is sensitive to 400 μJy so it is To observe transient radio sources, possible that we could observe variability on observations need to be made either over a a much more sensitive scale. large area of the sky, or over a long period While some galactic nuclei are of time. Since extragalactic radio transients active, many are not. This does not mean should occur isotropically, the location of the that galaxies lacking AGN do not have observation should not affect the rate of massive black holes at their centers. An transient detection. Because we have not inactive galactic nucleus indicates that the had dedicated telescope time for our search, massive black hole is not currently accreting we have used data taken by other material. It is possible that material could astronomers for other purposes. The data fall into these ’quiet’ black holes and we used for this survey was taken with the produce a bright flare of energy. This could VLA from 1982 to 1992. The VLA is a 27 happen when a star passes too close to the dish interferometer located in Socorro, New massive black hole and is torn apart by tidal Mexico. Each dish in the array has a forces [12]. After the star has been diameter of 25 meters. The VLA is currently disrupted, it falls into the black hole and the best instrument for conducting blind emits brightly. This emission is called a tidal surveys because of the high sensitivity and flare. Tidal flares have never been detected the large field of view. The dishes in the in the radio spectrum, and there are only a array can be moved to four different few x-ray candidates [13, 14]. configurations. The largest configuration Of the different theories predicting has a maximum baseline of 36 kilometers extragalactic radio transients, radio and the smallest configuration fits all the supernovae are the only phenomena that dishes within one kilometer. The larger two have been observationally confirmed. The configurations provide higher resolution, but cause of radio emission from some also limit the field of view. We did not use supernovae is thought to be produced by the the observations taken in these outer interaction of the shock produced by the configurations because the images were supernova explosion with the gas significantly smaller than the images surrounding the exploding star [15]. This

15 AMERICAN JOURNAL OF UNDERGRADUATE RESEARCH VOL. 5, NO. 2 (2006)

Figure 2. An image of the full field of view of our observations—notice the two bright sources in the lower left quadrant.

produced in the tighter two configurations. important to establish these sources since This left us with 16 observations from some low level sources appeared in only December 1982 to September 1992. Since one or two images. This does not make our data was taken in the L Band (1.4GHz), these single detections transients because each image covered 0.43 square degrees. they are present in the other images, just We reduced the data using a software hidden in the noise. To identify sources we package developed by the National Radio used an AIPS ‘task’ called Search And Astronomy Observatory (NRAO) called the Destroy (SAD) that searched for regions of Astronomical Image Processing System the image above given flux thresholds, fit (AIPS). AIPS is the standard program for Gaussians to these regions, and subtracted reducing and analyzing VLA data. To those Gaussians. SAD does this process reduce the data into a usable image we iteratively, detecting weaker sources. We performed self-calibration on the data using then compared the detections from each the AIPS ‘task’ CALIB. Some of the observation to the combined image. Our observations had data that was corrupted for data was not ideal for a blind survey some baselines, so we removed, or ‘flagged’ because it is a complicated field. Figure 2 these baselines. Once the data was illustrates that there are many sources in the appropriately flagged and calibrated, we field and several very bright sources. The imaged the data using the AIPS task two bright sources in the lower left corner IMAGR. After reduction, we used a were troublesome because they introduced combined image of all of our data to identify a high noise level in that quadrant of the the constant sources in our field. It was image. Because of this we ignored

16 AMERICAN JOURNAL OF UNDERGRADUATE RESEARCH VOL. 5, NO. 2 (2006)

Figure 3. Three sources from our survey with corresponding light curves. In the light curves, the triangles represent observations in the D, or most compact configuration, and the diamonds represent observations in the C configuration. The variation in the flux is not true variability, but a product of the different configurations of the VLA.

detections that were above the overall noise of seven possible transients. All seven of level, but near those bright sources. After these candidates happened to be in regions analyzing the 16 observations we had a list of high noise, or were just at the flux

17 AMERICAN JOURNAL OF UNDERGRADUATE RESEARCH VOL. 5, NO. 2 (2006) threshold we set for believability. It was not timescale transients. Figure 1 was prepared conclusive that any of these candidates for a theoretical survey consisting of two were transient radio sources. observations separated by one year, so the computed rates are per year. Figure 1 also III. DISCUSSION illustrates what theories we tested with our survey of about 100 hours. As you can see, Our 16 observations covered a total our survey was not sensitive enough, or had of ten years and 0.43 square degrees. To great enough area, to test all of the transient compute the upper limit on the rate of predictions, though we did probe some transients we look at the limit where the theories. Our observations do rule out the length of time that the transient would be theory of OGRBAs by Perna & Loeb [9] detected is larger than the spacing between since our upper limit is below the rate our observations and the limit where the predicted by their calculations. But lack of transient is detectable for a time less than detections in this survey is not that the duration of our observations. Our survey surprising. We should have detected was much more sensitive to the first case numerous OGRBAs if Perna & Loeb’s theory where the transient is detectable for at least was accurate, but newer predictions put several months. This is because our OGRBAs at an order of magnitude or two observations were on the order of twelve less frequent. Following these models, we hours in duration, but on average separated would need to at least double the size our by more than half a year. OGRBAs, tidal survey to detect an OGRBA. The upper limit flares, and radio supernovae should all have on the rate of tidal flares predicts that we detection durations on the order of months. could have observed a tidal flare, but the In this limit we can compute the rate using theory for tidal flares still produces a broad N range for the rate of tidal flare detections. Rate = det , −× TTA 120 ))(( IV. CONCLUSION

After analyzing 6.88 square degrees where Ndet is the number of detections, A0 is of data taken over ten years we were able to the area per observation, T1 is the date of set an upper limit on the rate of radio the first observation, and T2 is the date of transients with timescales on the order of the last observation. Setting Ndet equal to one we can compute an upper limit on these months at 0.23 per square degree per year types of transients of 0.23 radio transients above 400 μJy. Though it was unlikely for per square degree per year above the this survey to detect a transient radio threshold of our observations. The source, we were able to refine our process sensitivity of our observations varied from of analysis and are currently looking at more near 100 to 400 μJy. So we set our upper data from the VLA archives. This other set of limit on the rate using 400 μJy. For observations is larger and more sensitive transients that are observable on time than the survey we present here, providing scales similar to the length of our us with a much higher probability of finding a observations (twelve hours) we can use radio transient.

ACKNOWLEDGEMENTS N det Rate = , This research was completed under the ( 0 δ ×× NtA obs ) guidance of Geoffrey C. Bower. I would like where δt is the duration of a single to thank Dr. Alberto Bolatto for his observation in hours, and Nobs is the number contributions and Julie Comerford for her of observations. Using this equation we can assistance in the preparation of this paper. compute an upper limit of 0.012 radio Funding for this research was provided by transients with characteristic timescales less the Berkeley Radio Astronomy Laboratory. than 12 hours per square degree, per hour, REFERENCES above 400 μJy. This translates to an upper limit of 106 of these radio transients per 1. E. Waxman, S. R. Kulkarni, and D. A. square degree per year. This limit is so high Frail. Implications of the Radio Afterglow because this survey is not sensitive to short from the Gamma- Ray Burst of 1997

18 AMERICAN JOURNAL OF UNDERGRADUATE RESEARCH VOL. 5, NO. 2 (2006)

May 8. Astrophysical Journal, 497:288– August 1996. 293, April 1998. 7. A. Gal-Yam, E. O. Ofek, A. V. 2. D. A. Frail, S. R. Kulkarni, R. Sari, S. G. Filippenko, R. Chornock, and W. Li. Djorgovski, J. S. Bloom, T. J. Galama, SDSS J124602.54+011318.8: A Highly D. E. Reichart, E. Berger, F. A. Variable Active Galactic Nucleus, Not an Harrison, P. A. Price, S. A. Yost, A. Orphan Gamma-Ray Burst Afterglow. Diercks, R. W. Goodrich, and F. PASP, 114:587–592, June 2002. Chaffee. Beaming in Gamma-Ray 8. A. Levinson, E. O. Ofek, E. Waxman, Bursts: Evidence for a Standard Energy and A. Gal-Yam. Orphan Gamma-Ray Reservoir. Astrophysical Journal, Burst Radio Afterglows: Candidates and 562:L55–L58, November 2001. Constraints on Beaming. Astrophysical 3. E. Costa, F. Frontera, J. Heise, M. Journal, 576:923–931, September 2002. Feroci, J. in: ’t Zand, F. Fiore, M. N. 9. R. Perna and A. Loeb. Constraining the Cinti, D. dal Fiume, L. Nicastro, M. Beaming of Gamma-Ray Bursts with Orlandini, E. Palazzi, M. Rapisarda, G. Radio Surveys. Astrophysical Journal, Zavattini, R. Jager, A. Parmar, A. 509:L85–L88, December 1998. Owens, S. Molendi, G. Cusumano, M. 10. T. Totani and A. Panaitescu. Orphan C. Maccarone, S. Giarrusso, A. Coletta, Afterglows of Collimated Gamma-Ray L. A. Antonelli, P. Giommi, J. M. Muller, Bursts: Rate Predictions and Prospects L. Piro, and R. C. Butler. Discovery of for Detection. ApJ, 576:120–134, an X-ray afterglow associated with the September 2002. gamma-ray burst of 28 February 1997. 11. S. J. Wagner and A. Witzel. Intraday Nature, 387:783–785, June 1997. Variability In Quasars and BL Lac 4. J. van Paradijs, P. J. Groot, T. Galama, Objects. ARA&A, 33:163–198, 1995. C. Kouveliotou, R. G. Strom, J. Telting, 12. M. J. Rees. ’Dead quasars’ in nearby R. G. M. Rutten, G. J. Fishman, C. A. galaxies? Science, 247:817–823, Meegan, M. Pettini, N. Tanvir, J. Bloom, February 1990. H. Pedersen, H. U. Nordgaard-Nielsen, 13. S. Komossa. The extremes of (X-ray) M. Linden-Vornle, J. Melnick, G. van der variability among galaxies: Flares from Steene, M. Bremer, R. Naber, J. Heise, stars tidally disrupted by supermassive J. in ’t Zand, E. Costa, M. Feroci, L. black holes. In IAU Symposium, pages Piro, F. Frontera, G. Zavattini, L. 45–48, November 2004. Nicastro, E. Palazzi, K. Bennet, L. 14. L.-X. Li, R. Narayan, and K. Menou. The Hanlon, and A. Parmar. Transient Giant X-Ray Flare of NGC 5905: Tidal optical emission from the error box of Disruption of a Star, a Brown Dwarf, or a the gamma-ray burst of 28 February Planet? Astrophysical Journal, 576:753– 1997. Nature, 386:686–689, 1997. 761, September 2002. 5. D. A. Frail, S. R. Kulkarni, S. R. 15. R. A. Chevalier. The radio and X-ray Nicastro, M. Feroci, and G. B. Taylor. emission from type I supernovae. The radio afterglow from the gamma-ray Astrophysical Journal, 259:302–310, burst of 8 May 1997. Nature, 389:261– August 1982. 263, 1997. 16. K. W. Weiler, S. D. van Dyk, M. J. 6. E. V. Gotthelf, T. T. Hamilton, and D. J. Montes, N. Panagia, and R. A. Sramek. Helfand. The Einstein Observatory Radio Supernovae as Distance Detection of Faint X-Ray Flashes. Indicators. Astrophysical Journal, Astrophysical Journal, 466:779–793, 500:51–58, June 1998.

19 AMERICAN JOURNAL OF UNDERGRADUATE RESEARCH VOL. 5, NO. 2 (2006)