Time Monitoring of Radio Jets and Magnetospheres in the Nearby Young Stellar Cluster R Coronae Australis
Hauyu Baobab Liu, Roberto Galván-Madrid, Jan Forbrich, Luis F. Rodríguez, Michihiro Takami, Gráinne Costigan, Carlo Felice Manara, Chi-Hung Yan, Jennifer Karr, Mei-Yin Chou, Paul T.-P. Ho, Qizhou Zhang
aa r X i v : . [ a s t r o - ph . S R ] N ov submit v2 Preprint typeset using L A TEX style emulateapj v. 5/2/11
TIME MONITORING OF RADIO JETS AND MAGNETOSPHERES IN THE NEARBY YOUNG STELLARCLUSTER R CORONAE AUSTRALIS
Hauyu Baobab Liu , Roberto Galv´an-Madrid , Jan Forbrich , Luis F. Rodr´ıguez , Michihiro Takami , Gr´ainneCostigan , Carlo Felice Manara , Chi-Hung Yan , Jennifer Karr , Mei-Yin Chou , Paul T.-P. Ho , andQizhou Zhang Academia Sinica Institute of Astronomy and Astrophysics, P.O. Box 23-141, Taipei, 106 Taiwan European Southern Observatory, Karl-Schwarzschild-Str. 2, 85748, Garching, Germany University of Vienna, Department of Astrophysics, T¨urkenschanzstraße 17, 1180, Vienna, Austria Centro de Radioastronom´ıa y Astrof´ısica, UNAM, A.P. 3-72, Xangari, Morelia, 58089, Mexico School of Cosmic Physics, Dublin Institute for Advanced Studies, 31 Fitzwilliam Place, Dublin 2, Ireland Armagh Observatory, College Hill, Armagh BT61 9DG Department of Earth Sciences, National Taiwan Normal University,Taipei, 117 Taiwan and Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138 submit v2
ABSTRACTWe report JVLA 8-10 GHz ( λ =3.0-3.7 cm) monitoring observations toward the YSO cluster RCoronae Australis (R CrA), taken in 2012, from March 15 to September 12. These observations wereplanned to measure the radio flux variabilities in timescales from 0.5 hours to several days, to tensof days, and up to ∼
200 days. We found that among the YSOs detectable in individual epochs, ingeneral, the most reddened objects in the
Spitzer observations show the highest mean 3.5 cm Stokes I emission, and the lowest fractional variabilities on < ∼ < Subject headings:
Stars: activity — Stars: circumstellar matter — Stars: evolution — Stars: formation— Stars: magnetic field INTRODUCTION
Young (proto)stars are known to show radio flux vari-ability on various timescales. Magnetic reconnections onthe (proto)stellar surface can cause non-thermal radioflares in timescales shorter than several minutes (Dulk1985; Bower et al. 2003; Forbrich et al. 2008; Chen et al.2013; Su et al. 2013), while the interaction of the decou-pled magnetic fields between the protostars and the diskscan result in non-thermal radio flares (Shu et al. 1997)on timescales from a few days to as long as the 10-15days expected from protostellar rotation (e.g. Forbrichet al. 2006, see also Carpenter et al. 2001). In addition,accreting young stellar objects (YSOs) can emit thermalradio emission from the regions where the magnetohydro-dynamic (MHD) wind (Konigl 1982; Pudritz & Norman1983, 1986; Shu et al. 1994, 1995) shocks the ambi-ent gas (e.g. Rodr´ıguez 1997; Rodr´ıguez 1999; Anglada1995; Anglada et al. 1998). If the accretion rate to theprotostar and the mass-loss rate from the protostar areintimately linked as theories suggest (Calvet et al. 1993;Shang et al. 2004; see also Chou et al. 2013), then thethermal radio flux is expected to vary also on the dynam-ical timescale of the accretion disk, the timescales of diskinstabilities (several years; e.g. Zhu et al. 2009), and onthe dynamic- and hydrogen recombination timescales ofthe thermal radio jet core (as short as 1-3 months; e.g. [email protected]
Galv´an-Madrid et al. 2004).Radio monitoring observations towards YSOs, plannedto resolve the flux variability, spectral indices, and po-larization percentages, can shed light on discriminat-ing the aforementioned magnetospheric emission mech-anisms (Forbrich et al. 2011). In addition, comparisonof the radio fluxes between a sample of YSOs occupying abroad range of evolutionary stages may provide hints onthe evolution of the (proto)stellar magnetosphere on theone-million-year YSO evolutionary timescale (e.g. Dzibet al. 2013; AMI Consortium et al. 2012). We there-fore resumed the multi-epoch 3.5 cm radio observationstowards the R Coronae Australis (R CrA) cluster since2012 March using the National Radio Astronomy Ob-servatory (NRAO) Karl G. Jansky Very Large Array(JVLA). This target was selected because it has a con-centration of early YSOs in a field of a few arcminutes,and also due to its proximity ( d ∼
130 pc; for a discussionof the distance, see Neuh¨auser & Forbrich 2008).The R CrA cluster is one of the nearest young, dense(i.e. >
25 Class 0-II YSOs pc − , see Myers 2009) clusterswhich remains embedded in the natal molecular cloud.The previous optical and near-infrared (OIR) observa-tions (Wilking et al. 1985, 1992; L´opez Mart´ı et al. 2005;Haas et al. 2008; Peterson 2011) found that the majority The National Radio Astronomy Observatory is a facility ofthe National Science Foundation operated under cooperative agree-ment by Associated Universities, Inc.
TABLE 1The 2012 JVLA observations of 3.5 cm emission.
Epoch Time a Day b Array uv range c Medium API d Cloud e Synthesized beam rms f Flux/Pol.config. elevation rms θ maj × θ min; P.A. noise cal. g (UTC) (day) (m) (deg) (deg) (arcsec × arcsec, deg) ( µ Jy beam − )1 Mar.15 14:21 0 C 26-3387 19.1 1.2 sky clear 8 ′′ .8 × ′′ .4; 177 ◦
19 3C286/J2355+49502 Mar.16 14:27 1 C 26-3383 19.1 2.0 sky clear 8 ′′ .0 × ′′ .5; 178 ◦
18 3C286/J2355+49503 Mar.17 14:13 2 C 26-3387 19.1 1.2 10% covered 8 ′′ .1 × ′′ .5; 177 ◦
16 3C286/J2355+49504 Mar.17 14:43 2 C 28-3356 19.0 1.9 10% covered 8 ′′ .1 × ′′ .4; 2.6 ◦
23 3C286/J2355+49505 Mar.17 15:13 2 C 30-3314 18.5 1.4 10% covered 8 ′′ .4 × ′′ .6; 7.2 ◦
18 3C48/J2355+49506 Mar.17 15:43 2 C 26-3323 17.2 4.8 10% covered 9 ′′ .1 × ′′ .6; 14 ◦
21 3C48/J2355+49507 Mar.17 16:13 2 C 39-3099 15.3 2.5 20% covered 9 ′′ .6 × ′′ .7; 20 ◦
24 3C48/J2355+49508 Mar.22 13:53 7 C 26-3387 19.1 2.2 10% covered 9 ′′ .0 × ′′ .4; 178 ◦
20 3C286/J2355+49509 Mar.22 14:25 7 C 47-3352 19.0 2.7 sky clear 9 ′′ .0 × ′′ .4; 3.6 ◦
24 3C286/J2355+495010 Mar.25 14:14 10 C 28-3352 19.0 1.3 30% covered 8 ′′ .3 × ′′ .5; 3.4 ◦
22 3C286/J2355+495011 Mar.31 13:49 16 C 28-3362 19.0 2.4 50% covered 8 ′′ .5 × ′′ .3; 2.7 ◦
26 3C286/J2355+495012 Apr.02 13:08 18 C 26-3388 19.0 6.9 10% covered 7 ′′ .9 × ′′ .2; 174 ◦
32 3C48/J2355+495013 Jul.28 06:00 135 B 103-11069 19.1 7.1 10% covered 2 ′′ .6 × ′′ .1; 2.3 ◦
21 3C48/J2355+495014 Sep.12 01:45 181 BnA 166-10567 17.9 6.5 80% covered 2 ′′ .2 × ′′ .96; 166 ◦
34 3C286/J2355+4950
Note.— All epochs were observed using the correlator setting described in Table 2. The pointing center for all epochs of observations isR.A.=19 h m s (J2000), Decl.=-36 ◦ ′ ′′ .0 (J2000). a All epochs are observed in 2012. The observations (including calibrations) started 15 minutes before the time noted here, and ended 15minutes after it. The first ∼
10 minutes in each epoch were used for taking dummy observing scans as required by the system.b The relative day to the first epoch.c From the minimum to the maximum of the baseline projected lengths. We present it in units of meters rather than kilo-wavelengths becauseof the large range of observing frequencies.d The values of the atmospheric phase interferometer quoted from the observing log.e The sky condition commented by the JVLA operator.f Measured at the center of the IF1 Stokes I image generated utilizing the 1 GHz total bandwidth (centered at the sky frequency ν =8.5 GHz).g The observed quasar for absolute flux and polarization calibrations. of the objects younger than Class II are located in thecentral r ∼ ∼ ′ ) gas concentration (Loren 1979;Loren et al. 1983; Harju et al. 1993; Henning et al. 1994;Andreazza & Vilas-Boas 1996; Anderson et al. 1997a,1997b; Chini et al. 2003; Groppi et al. 2004). High an-gular resolution mapping observations and molecular linesurveys further confirmed abundant protostellar cores inthis region (Nutter et al. 2005; Sch¨oier et al. 2006; Lind-berg & Jørgensen 2012; Watanabe et al. 2012; Sicilia-Aguilar et al. 2011, 2013) and even found prestellar corecandidates (Groppi et al. 2007; Chen & Arce 2010). Theinstantaneous accretion rates of several YSOs in this fieldhave been constrained by a near infrared line survey in2002 July 12 and 13 (Nisini et al. 2005).In the radio part of the spectrum, VLA 6-cm observa-tions in 1985 resolved 11 radio emission sources at > σ significance (Brown 1987). Similar results were also givenby the Australia Telescope Compact Array (ATCA) 3-cm, 6-cm, and 20-cm observations in 1998 and 2000 (Mi-ettinen et al. 2008). Early VLA and Australia Tele-scope (AT) observations between 1985 and 1993 have re-vealed radio flux variability at 6-cm wavelength (Suters1996). More extensive, deep 3.5-cm observations wereperformed with the VLA in 1996-1998 (Feigelson et al.1998; Forbrich et al. 2006), and in 2005 (Forbrich et al.2007; Choi et al. 2008, 2009). Due to limited sensitiv-ity, those previous radio observations generally have anon-source integration time of several hours to achieve anadequate significance for detections.Thanks to the improved sensitivity of the JVLA, inves-tigation of the radio flux variability in the R CrA clusteron ≪ ∼ I radio flux variability. A preliminary interpreta-tion of our observational results is given in Section 4.3.Section 5 summarizes the main results of this paper. OBSERVATIONS AND DATA REDUCTION
We performed 16 epochs of filler-mode observationstowards the R Coronae Australis region using theJVLA C, B, and BnA array configurations in 2012from March to September. The pointing center forall epochs is R.A.=19 h m s (J2000), Decl.=-36 ◦ ′ ′′ .0 (J2000). Each observation epoch has anoverall duration of 30 minutes, and contains two ∼ on-source scans (separated by ∼
50 seconds).This relatively short calibration cycle as compared withthe typically longer than 20 minutes calibration dutycycles, helps compensate out the relatively large atmo-spheric effects for the low elevation target source (alsomentioned in Forbrich et al. 2006). We lost one epochof observations on 2012 March 25 because of missing thecalibration data; and we lost another epoch of observa-tions in 2012 September 13 because the weather condi-tions were too poor to allow robust antenna-based gaincalibrations. The details of the remaining observations The exact on-source time slightly varies among epochs of ob-servations because of the differences in antenna slewing time.
Fig. 1.—
The 3.5-cm radio image of the R CrA YSO cluster (gray scale). The right panel zooms into the sub-field around the groups ofcompact radio sources IRS7E and IRS7W (Table 3). This image is generated using Briggs weighting with Robust=1, incorporating all IF1data described in Tables 1 and 2. The θ maj × θ min =4 ′′ .3 × ′′ .0, P.A.=-179 ◦ synthesized beam is shown in bottom left of the right panel.The rms noise level is 8.5 µ Jy beam − . Contours in the left and right panels are [5 σ ] and [2.5 σ , 5 σ ], respectively. To avoid noisier edges,the images presented are not yet corrected for primary beam attenuation. The annotated sources and the primary beam attenuation attheir location can be found in Tables 3 and 4. The scale bars in both panels are drawn assuming a distance of 130 pc (see Deller et al.2013 and references therein). The red diamonds, orange crosses, and green crosses mark the locations of the Class I, Class II and flat SED,and Class III YSOs (Peterson et al. 2011) which were not detected in our JVLA observations, respectively. TABLE 2The correlator setup of the 2012 JVLA observations.
IF Spw ID a Central frequency b Bandwidth
Note.— There is no doppler tracking in our observations. a The ID of the observed spectral windows.b The sky frequency at the center of the spectral window. Thespectral windows spw 10 and 11 often had strong radio fre-quency interference (RFI) and thus were flagged out for allepochs of observations. are listed in Table 1. The correlator setup of our ob-servations can be found in Table 2. The total band-width after combining the 16 spectral windows in the 2independently tunable intermediate frequencies (IFs) is2 GHz (Table 2). We centered IF1 at a sky frequencyof 8.5 GHz to enable comparison with the extensive ear- lier VLA observations at the same frequency (Feigelsonet al. 1998; Forbrich et al. 2006; Forbrich et al. 2007;Choi et al. 2008, 2009). The IF2 was tuned to completea continuous 2 GHz total frequency coverage, and alsoto minimize differences in the primary beam coveragebetween the lowest and the highest frequency spectralwindows. The expected root-mean-square (rms) noiselevel for individual images of a 128 MHz spectral win-dow is ∼ µ Jy beam − in each epoch. However, thenoise may be degraded depending on the data flaggingand the unresolved radio frequency interference (RFI).The data were calibrated using the Common Astron-omy Software Applications (CASA) package release3.4.0. Each epoch of observations listed in Table 1was further independently phase self-calibrated to mini-mize the decoherence of fluxes. By the time these datawere calibrated, the JVLA data format did not includeweights, and the CASA task statwgt for reweighting vis-ibilities was still experimental. Because of being unableto calibrate the weightings, we were not able to correctlycombine the data in spectral windows with very differ-ent noise levels. Therefore, in this manuscript, we onlypresent the data in spectral windows 0-8, which have thelowest noise levels; and we only jointly imaged the datain IF1. Including the other spectral windows in the jointimaging either does not change the results, or may in-crease the noise level. However, the weighting issue doesnot affect the studies presented in this paper. We expectthis weighting issue to be resolved in the future, whichwill allow us to reprocess these data. http://casa.nrao.edu TABLE 3The 2-dimensional Gaussian components for initializing the source fits.
Source name R.A. Decl. Major axis FWHM a Minor axis FWHM a P.A. Flux P11 Classification b (J2000) (J2000) (arcsec) (arcsec) (deg) (mJy)IRS7B IRS7E 19:01:56.422 -36:57:27.6 2.88 1.24 5.0 1.46 Class IFPM15 19:01:56.476 -36:57:25.6 5.60 2.00 18.9 0.84IRS7B-S 19:01:56.326 -36:57:30.8 3.49 1.22 178.1 0.25IRS7A IRS7W 19:01:55.325 -36:57:22.1 2.74 1.31 175.9 6.63 Class IB9 19:01:55.291 -36:57:16.6 2.69 1.31 176.0 1.19FPM13 19:01:55.375 -36:57:13.0 5.56 1.71 178.3 0.66FPM10 19:01:54.974 -36:57:16.0 4.80 2.36 178.5 0.26JVLA3 (CXO 34) 19:01:55.793 -36:57:27.1 2.10 0.96 172.0 0.060 Class IJVLA2 (WMB55) 19:01:58.561 -36:57:08.6 2.70 1.20 178.0 0.10 Class IIRS5N 19:01:48.484 -36:57:14.8 2.10 0.69 166.0 0.049 Class IIRS5 19:01:48.061 -36:57:22.0 3.05 1.42 0.9 1.09 Class IIRS1 19:01:50.685 -36:58:09.7 2.98 1.21 179.3 0.64 Class IJVLA4 (Haas 4) 19:01:40.667 -36:56:05.2 2.10 0.96 169.0 0.084 Flat SEDIRS2 19:01:41.579 -36:58:31.3 2.88 1.20 178.4 0.37 Class IIRS6 19:01:50.484 -36:56:38.3 3.83 1.26 167.0 0.12 Class IIT CrA 19:01:58.784 -36:57:49.7 3.18 1.70 176.0 0.18 Class IIJVLA1 (CrA PMS1) 19:01:34.858 -37:00:55.7 2.69 1.11 178.2 0.13 Class IIIR CrA 19:01:53.686 -36:57:08.0 3.49 1.22 178.1 0.28 Class IIIB5 19:01:43.283 -36:59:12.0 2.71 1.14 174.9 0.68 Galaxy Note.— This target list is generated by fitting the compact sources in the deep Briggs Robust=0 weighted image, incorporating all IF1data described in Tables 1 and 2. IRS7E resolved at higher angular resolotion into IRS7B, FPM15, and IRS7B-S. IRS7W resolves intoIRS7A, B9, FPM10, and FPM13. The 1 σ rms noise levels at the individual locations of these sources are ∼ µ Jy beam − divided by theprimary beam attenuation factors listed in Table 4. a The listed values of FWHM are not yet deconvolved from the θ maj × θ min=2 ′′ .7 × ′′ .11 (P.A.=178 ◦ ) synthesized beam. Several of the listedsources are consistent with point sources at our angular resolution, thus cannot be deconvolved.b YSO classification quoted from Peterson et al. (2011), except for the extragalactic source B5. TABLE 4The averaged primary beam attenuation factor forindividual sources.
Source name Spw 1 Spw 8 IF1 a IRS7E 0.76 0.71 0.74IRS7W 0.79 0.75 0.78JVLA3 (CXO 34) 0.79 0.74 0.77JVLA2 (WMB55) 0.62 0.56 0.60IRS5N 0.94 0.94 0.95IRS5 0.97 0.96 0.96IRS1 0.97 0.97 0.97JVLA4 (Haas 4) 0.59 0.52 0.56IRS2 0.84 0.81 0.83IRS6 0.83 0.80 0.82T CrA 0.65 0.59 0.63JVLA1 (CrA PMS1) 0.20 0.13 0.18R CrA 0.84 0.80 0.83B5 0.81 0.77 0.80
Note.— The columns Spw 1 and Spw 8 are the primary beam at-tenuation factors for images generated using the data in spectralwindow 1 and in spectral window 8, respectively. a The averaged primary beam attenuation factors while incorpo-rating all spectral windows in IF1.
We performed the naturally weighted imaging usingthe CASA task clean . The image size is 3600 pixelsin each dimension, and the pixel size is 0.2 ′′ . A fewof the sources in the R CrA field are known to be as-sociated with extended radio emission (e.g., Miettinenet al. 2008). We therefore implemented a lower cut ofthe visibility uv distances ( √ u + ν ) of 4.4 k λ , which iscomparable to the shortest baseline of the BnA array ob-servations (Table 1), to all epochs of data before imaging. This ensures that the extended emission does not bias themeasurements of the flux variations. We found that mostof the emission comes from the compact components af-ter implementing the 4.4 k λ cut. The flux measurementsare not sensitive to the √ u + ν cut when it is longerthan 4.4 k λ . Cutting even at much larger uv distances(e.g. 45 k λ ) does not fundamentally change our measure-ments, however, it significantly degrades the sensitivityand the synthesized beam shapes of the C array observa-tions. The rms noise levels achieved after combining thespectral windows 0-7 are given in Table 1. We note thatthese observations are sensitive to events at the 0.6-0.9mJy level, such as the radio-jet knot eruption reportedby Pech et al. (2010) in IRAS 16293-2422. RESULTS
The Compact Sources in the 3.5 cm Stokes I Image
To yield a deep radio image, we combined and jointlyimaged the phase self-calibrated IF1 data (Table 2) fromall 14 epochs of observations listed in Table 1. The BriggsRobust=1 weighted combined image without the imple-mentation of the > λ uv distance limit (Section 2)is shown in Figure 1. The compact radio sources wereregistered by performing 2-dimensional Gaussian fittingson the Briggs Robust=0 weighted combined image, usingthe CASA task imfit . Since the Gaussian fitting is fun-damentally ambiguous (e.g., not necessarily converges toa unique solution), we implemented the minimal possi-ble number of Gaussian components which can recoverthe emission of the compact sources well. The Gaussiancomponents are listed in Table 3. The primary beamattenuation of the fluxes was corrected only after the 2-dimensional Gaussian fittings to avoid confusion by thenoise. The primary beam attenuation factors for the indi-vidual of sources are given in Table 4. Hereafter, we referto the group of sources IRS7B, FPM15, and IRS7B-S asIRS7E, and to the group of sources IRS7A, B9, FPM13,and FPM10 as IRS7W, because they are not resolvedin every epoch of our JVLA C array observations. Thesource IRS5 is known to be binary (its components areknown as IRS5a, IRS5b, see Chen & Graham 1993; Choiet al. 2008; Deller et al. 2013), however, they cannot beseparated given our angular resolution. The radio fluxvariability of the individual components in these groupswill not be independently discussed in this manuscript(Section 3.2). From high angular resolution 7 mm contin-uum images (Choi & Tatematsu 2004), the group IRS7Wis likely to be a cluster of young (proto)stars with asso-ciated thermal jet knots. The components FPM15 andIRS7B-S may trace discrete knots in the extended bipolarradio jet emanating from the young stellar object IRS7B(Forbrich et al. 2006; Choi et al. 2008). Alternatively,they may be tracing the base of this outflow. A futuresearch for proper motions may clarify the nature of thesecomponents.We found that the sources IRS7E, IRS7W, IRS5, andIRS6 are associated with extended radio emission (Figure1). The properties of the extended emission will be ad-dressed in a forthcoming paper incorporating the followup observations in the more compact JVLA array config-uration, and the observations at higher frequencies. Theradio source JVLA1 is detected for the first time, and islikely the known Class III object CrA PMS 1. By crosscomparing with the YSO source catalog from the previ-ous infrared surveys (Peterson et al. 2011), we also claimthe new detections of the additional two faint sourcesJVLA3 and JVLA4, which are likely the Class I objectCXO 34, and the flat spectrum object Haas 4, respec-tively. JVLA3 can be isolated from the extended emis-sion after the > λ uv distance limit is implemented.In the image incorporating data from all epochs of ob-servations, the source JVLA2 can be isolated from thenorth-east extended radio lobe emanated from IRS7E.It may be the weakly detected radio source WMB55 re-ported in Choi et al. (2008) (see also Wilking et al.1997), which is associated with the submillimeter coreSMM2 (Groppi et al. 2007).The radio source B5 which was previously proposed tobe a brown dwarf candidate (Feigelson et al. 1998), isnow confirmed to be extragalactic (Jan Forbrich, privatecommunication), thus will be omitted in the followingdiscussion.To provide a sense of the evolutionary stages of the de-tected YSOs, we quote the Peterson et al. (2011) classi-fication of YSOs in Table 3, and show the Spitzer color-color diagram in Figure 2. The
Spitzer fluxes in thecolor-color diagram are also from Peterson et al. (2011).The
Spitzer fluxes of the well known Class I YSO can-didate IRS9 (Forbrich & Preibisch 2007; Peterson et al.2011) cannot be measured because it is located too closeto the bright source R CrA. We do not detect 3.5-cm ra-dio emission from IRS9 either and will omit this sourcefrom the following discussion.The YSOs at earlier evolutionary stages should in gen-eral be redder and will appear at the top right of the
Spitzer color-color diagram. Contamination and short- period infrared variations of the YSOs may cause un-certainties in the
Spitzer colors. We note that althoughthe loci in the
Spitzer color-color diagram help to dividethe sample into the conventional Class 0-III evolution-ary stages, the actual evolutionary tracks of the YSOsmay be more continuous. For sources located very neara boundary, for example, IRS2 and IRS6, their classifi-cation as Class I or II is not intrinsically important. Wealso note that the source R CrA is in fact a Herbig Aestar (Peterson et al. 2011, and references therein). Bycomparing Figure 1 with Figure 2, we conclude that thenon-detections of some Class II and Class III YSOs arenot due to primary-beam attenuation. It is more likelythat the majority of these sources were fainter than oursensitivity limit at all epochs.Figure 2 shows that while practically all Class 0/Isources are detected in the radio, sources in the ClassesII/III are only rarely detected. A possible explanationis that in the Class 0/I sources, we are detecting free-free emission from an ionized outflow that is systemati-cally present in this type of objects. In contrast, in theClass II/III sources, we are probably detecting gyrosyn-chrotron emission from active magnetospheres (see alsoGibb 1999). This is a time-variable process that is notnecessarily present in all Class II/III stars (more discus-sion in Section 4.3). In addition, the left panel of Figure 1shows a clear spatial differentiation of the Clsss 0/I andClass II/III sources. While Class 0/I sources are con-fined to the inner part of the cluster, in a region of about2 ′ in extent, the Class II/III sources extend over 4 ′ -5 ′ .This may suggest that star formation did not take placesimultaneously in all the cluster, but that it propagatedinwards with time. Alternatively, it may be explained bythe fact that the Class II and III objects are old enoughand have had time to diffuse (a star moving at ∼ − travels 1 pc in 1 Myr). The 3.5 cm Stokes I Flux Variabilities.
To analyze the 3.5-cm Stokes I flux variabilities, im-ages with low noise and minimal phase decoherence arerequired. For each epoch of observations, we there-fore jointly imaged the √ u + ν > λ phase self-calibrated data in IF1 (Table 2). We perform 2-dimensional Gaussian fits to these broad band imagesto obtain the fluxes. The derived Gaussian componentsin Table 3 were used to initialize the Gaussian fits for allepochs. The residual noise level, as well as the shapes ofthe Gaussian models and the residuals, were inspectedto verify the convergence of the fits. The errors from theGaussian fits were obtained by taking the maximum ofthe two estimates described in the AIPS++ Note 244 ,and in Condon et al. (1998), as well as Richards et al.(1999). These two methods are based on the signal tonoise ratio of the fitted Gaussian component, and thegoodness of fit, respectively. For most of the epochs, thepositions of each Gaussian component only need to beshifted by < ′′ ) relative to the initial model,which is not significant as compared with the synthe-sized beam sizes (Table 1). The images of Epoch 6, 7,and 12 required to shift the Gaussian component rela-tive to the initial model, by up to 20% of the synthesizedbeam full width half maximum (FWHM). This probably Fig. 2.—
The
Spitzer color-color diagram of the detected YSOradio sources (black). The YSOs located within the JVLA primary-beam attenuation contour at 0.1 of the peak, but that were not de-tected in radio emission, are plotted in gray symbols. The plotteddata are quoted from Peterson et al. (2011). The dashed loci forYSO classification are drawn based on Allen et al. (2004) and Leeet al. (2006). The source R CrA is in fact a Herbig Ae star (Krauset al. 2009, and references herein). results from a combination of noise and spatial drifts ofthe images due to the phase self-calibration.Alternative methods to obtain the fluxes are summingthe fluxes within box regions enclosing the sources (e.g.Feigelson et al. 1998), or summing the fluxes in regions(partially) defined by contours at certain significance lev-els (e.g. 2 σ , Choi et al. 2008). We did not use the formermethod because it is hard to uniformly define the box re-gions given the variations of the synthesized beam sizesand position angles in our 2012 observations (Table 1).The latter method is potentially biased in observationswith high noise levels. In our observations, the differ-ences of the measured fluxes with all mentioned methodsare typically less than 10% for IRS7W and IRS7E, andare much smaller for point sources. This systematic ef-fect is smaller than the intrinsic flux variabilities for mostof the sources (Section 4.1). However, these methods aresubject to different errors (i.e., the error bars can be dif-ferent).The measured broad band 3.5-cm Stokes I fluxes areshown in Figure 3. We also quote the 3.5-cm fluxes inearlier VLA A-array observations on 1996 December 29(Choi et al. 2008), BnA-array observations on 1997 Jan-uary 19 and 20 (Feigelson et al. 1998), BnA array obser-vations on 1998 June 27 (Forbrich et al. 2006), 8 epochsof B-array observations on 1998, from July 19 to October13 (Forbrich et al. 2006), and BnA-array observationson 2005 February 03 (Choi et al. 2008). We note thatthe radio flux of IRS5 flared to up to ∼ Fig. 3.—
The 3.5 cm Stokes I fluxes of the detected sources.The source JVLA2 is too faint to be robustly detected in individualepochs, thus cannot be plotted. The red symbols connected withgray dashed lines show the 2012 observations listed in Table 1. Forparticular 2012 epochs in which the sources cannot be detected, the2 σ upper limits are given in green downward arrows. Otherwise, weshow the ± σ uncertainties with black error bars. In addition, wequote the 3.5-cm Stokes I fluxes observed in 1996 December (Choiet al. 2008), 1997 January (Feigelson et al. 1998), 1998 from Juneto October (Forbrich 2006), and 2005 February (Choi et al. 2008),in black dotted lines, orange dotted lines, cyan dotted lines, and reddotted lines, respectively. The 1998 flux of IRS6 (and of FPM10,FPM13, FPM15, which areincorporated in IRS7W and IRS7E), isaveraged from all 9 epochs of observations in 1998 because of thefaintness of the source. The source T CrA was not detected in 1998observations. The sources JVLA1 was not detected in any of theprevious observations. Fig. 4.—
Similar to Figure 3. The left panel zooms into the time periods of the C-array observations listed in Table 1. The right panelszooms in further to show only the 5 epochs taken on 2012 March 17. bright sources will be addressed in Section 4. Figure 4zooms in to better present the observations from 2012March 15 to April 02, and the consecutive 5 epochs ofobservations on 2012 March 17.In our JVLA field of view (Figure 1), three Class 0/Isources (IRS7E, IRS7W, IRS5), and one source in be-tween the Class I and Class II stages (IRS6) are asso-ciated with diffuse Stokes I emission. However, two ofthese sources (i.e. IRS7W and IRS5) show higher 3.6-cm fluxes in the more extended B-array and BnA-array measurements, which can only be explained by flux vari-ations on the unresolved spatial scales. The source IRS7E shows lower flux in the BnA array epoch (Epoch 14),but this is still consistent with short-term intrinsic fluxvariations. Because the uv distance ranges of the B-arrayand the BnA-array observations are not very different ascompared with the C-array observations (Table 1), we donot think the drop of the IRS7E flux in Epoch 14 is due tothe uv sampling. After implementing the √ u + ν > λ cut (Section 2), the source IRS6 is consistent with apoint source. As can be seen in Figure 3, the radio vari-ability of IRS6 is dominated by occasional short-durationflares, so contamination from the diffuse emission shouldbe negligible. The rest of the sources in our field are pointsources, so flux variability can be measured without anybias from the JVLA array configuration.We found that the fluxes of the four bright sourcesIRS7W, IRS7E, IRS5, and IRS2 dropped by ∼
10% si-multaneously in Epoch 6 (Table 1; Figure 4). Becauseof the larger API rms (Table 1) and larger errors of theGaussian fits, we think that this may be due to the loss ofcoherence caused by the phase noise. To some extent thelarger error bars can take care of this systematic effect.We do not manually correct the fluxes because we can-not rule out that this is a real, simultaneous flux drop.Nevertheless, we also found that manually correcting thefluxes by 10% does not qualitatively affect our statisti-cal analysis (Section 4). We did not identify the sameissue in other epochs of observations. The 1996 fluxesof all sources appear to be systematically lower. We hy-pothesize that it is due to both the poor signal to noiseratio and the loss of phase coherence, but it is not com-pletely clear. Although we will exclude these data pointsfrom the following discussion and statistical analysis, wefound that including them does not change our resultsqualitatively.Figure 3 shows that the 3.5-cm fluxes of the twoyoungest sources IRS7E and IRS7W (Figure 2) haveincreased since as early as 1997 January, and are nowfluctuating around their 2005 February values (Choi etal. 2008). The mean radio fluxes of these two sourcesare ∼ < . uv coverage ineach snapshot JVLA epoch. The 2012 fluxes of IRS5fluctuated about its 2005 value (Choi et al. 2008), whichis ∼ − to 10 days. The source IRS6 is only detected inoccasional flares, similar to the Class II and the Class IIIsources T CrA, R CrA, and JVLA1. The source R CrA(the brightest in the near infrared) occasionally falls be-low our detection limits ( < σ ) in our 2012 observations,while it was detected in all epochs of earlier observations.However, from Figure 4, the low state of R CrA may onlylast for <
30 minutes. Because all previous observationsrequired at least several hours of on source integrationto achieve the adequate signal-to-noise (S/N) ratio, theymight not be sensitive to the low states of R CrA.The 2012 fluxes of the extragalactic source B5 fluc-tuated within the same range as in the earlier observa-
Fig. 5.—
The 3.5-cm spectral indices of the five most significantlydetected YSO sources. The horizontal axis shows the fluxes of the3.5-cm emission measured from the IF1 data (Table 2). The errorbars represent the 1 σ uncertainties. The spectral indices derivedfrom the 5 epochs of observations on 2012 March 17 (Table 1) arepresented in cyan color. The spectral indices derived from ourJVLA B array and BnA array observations are presented in redcolor. The spectral indices derived from other of the 2012 JVLA Carray observations are presented in black color. The dotted line andthe shaded area show the results of linear regression for spectralindices derived from the JVLA C array observations and the 1 σ uncertainties returned by the IDL fitting program POLY FIT. tions for timescales > ± > s variability of theinnermost accretion flow around a supermassive blackhole (e.g. Miniutti et al. 2006), which in fact makesthe gain calibration of JVLA data possible (Section 2).The small flux variations observed in B5 on < The Stokes I 8.2-9.1 GHz Spectral Index
We provide a preliminary analysis of the spectral in-dex and the fluctuations of the spectral index of brightsources by comparing the radio fluxes measured fromspectral window 1 and spectral window 8 (Table 2).These two spectral windows have low noise levels, and are
TABLE 5The earlier measurements of spectral indices between 6cm and 3 cm.
Source name 1996 Dec 29 a b IRS7A 0.04 ± ± ± · · · FPM13 -0.2 ± · · · IRS7B -1.41 ± ± ± ± ± · · · IRS2 · · · · · ·
Note.— a Derived from the VLA observations reported in Choi et al.(2008).b Derived from the ATCA observations reported in Miettinen etal. (2008). We note that these ATCA observations cannotresolve FPM15 from IRS7B; and cannot resolve FPM10 andFPM13 from IRS7A and B9. adequately separated in frequency. Ideally, the spectralindex analysis should incorporate the fluxes from higherfrequency spectral windows. This is presently hinderedby the higher noise in those spectral windows, and ourinability to correctly combine the data (Section 2).We smoothed the image of spectral window 8 to thesame angular resolution of spectral window 1 before mea-suring the fluxes. The images of spectral windows 1 and 8are subject to a higher noise than the broad band images(Section 3.2, Table 1) because of the smaller bandwidth.Therefore, we trimmed both images to the 4 σ level toavoid confusion by noise. In some epochs, the sourcescannot be detected in the images of spectral window 8 af-ter trimming, thus the spectral indices were not derived.The obtained spectral indices, if available, are presentedwith the 3.5-cm fluxes in Figure 5. Because the spectralwindow 1 is more sensitive and is subject to a smallerprimary beam attenuation, the measurements presentedin Figure 5 preferentially picked positive spectral indicesfor weak ( < σ trimming can bedifferent for images with different angular resolution.For comparison, in Table 5 we quote the spectral in-dices previously measured between 6 cm and 3 cm. Thequoted spectral indices were derived from observationstaken on the same date, but not exactly simultaneous.Although IRS2 is too faint to obtain a meaningful con-straint on its spectral index, the rest of the measurementspresented in Figure 5 are consistent with earlier obser-vations. In particular, our measurements of the spectralindex of IRS7E ( α obs IRS7E ) vary within a range consistentwith the previously reported spectral index of IRS7B. α obs IRS7E shows a general trend of being more negative whenthe 3.5 cm flux is higher, but it has an exception with α obs IRS7E ∼ +2. α obs IRS7E converged towards ∼ α obs IRS7W stayed within the range ± − α obs IRS5 also converged towards ∼ α obs IRS5 , however, cannot be represented by a linear rela-
Fig. 6.—
The detected 3.5 cm Stokes V emission (see also Table6) in the 2012 observations (Table 1). For particular 2012 epochswhen the sources cannot be detected, the 2 σ upper limit is given ingreen downward arrows. Otherwise we show the ± σ uncertaintiesin black error bars. tion, but is rather bimodal. For IRS1, α obs IRS1 fluctuatesaround 0, with a marginal trend to be more negativewhen the flux is higher. More discussion about the spec-tral indices is deferred to Section 4.3.
The Stokes V Flares
We used the method introduced in Section 3.2 to mea-sure the fluxes of the 3.5-cm Stokes V emission. Theresults are shown in Figure 6 and Table 6. We de-tected Stokes V flares towards the three Class 0/I sourcesIRS7E, IRS7W, and IRS5. From the position of the de-tection, the Stokes V flares observed in IRS7W are likelyto be dominated by the component IRS7A, rather thanby B9 or any component. However, given the angularresolution of our observations (Table 1), we cannot ruleout that B9 contributed partially.Stokes V emission from IRS7A and IRS5 was also re-ported in previous VLA observations (Feigelson et al.1998; Forbrich et al. 2006; Choi et al. 2008, 2009). Whilethe Stokes V flares of IRS5 were observed to have dura-tions >
30 days (Forbrich et al. 2006; Choi et al. 2008,2009), the Stokes V emission from IRS7A was detectedin only one previous observation (Choi et al. 2008). Ourresults are qualitatively similar to previous reports, andwill be briefly discussed in Section 4.3. DISCUSSION
We examine the statistics of the measured Stokes I fluxes in the 2012 JVLA observations in Section 4.1. Wecompare our observations with the earlier VLA observa-tions in Section 4.2. Our tentative interpretation of theobservational results is provided in Section 4.3. Statistics of Stokes I Emission in 2012
We used the biweight method in robust statistics(Hoaglin et al. 1983) to estimate the steady flux levelsand the dispersions of the fluxes. This method is advan-tageous because it can objectively lower the weights orreject the measurements which are largely deviated from0
Fig. 7.—
The biweight mean of the 3.5-cm Stokes I flux F , the (biweighted) standard deviation ( σ bw ) of the flux, the fractional standarddeviation σ bw / F , the absolute maximal flux deviation relative to the biweighted mean ∆ F max , and the fractional maximal flux deviation∆ F max /F. The statistics in these diagrams incorporate all the 2012 IF1 data summarized in Tables 1 and 2. The horizontal axis can beused as an indicator of the YSO evolutionary stage (see also Figure 2). We note that the three sources TCrA, IRS6, and JVLA1 were onlyoccasionally detected (presented in gray dots). The values of F and σ bw for these sources are therefore upper limits ( σ bw / F is not verymeaningful), and the values of ∆ F max /F are lower limits. σ bw can represent the uncertainty in the biweighted mean as well as its trend. TABLE 6The 2012 detections of Stokes V emission and thepolarization percentage. IRS7A IRS7B IRS5V V/I a V V/I a V V/IEpoch ( µ Jy) (%) ( µ Jy) (%) ( µ Jy) (%)7 183 ±
32 2.2 ± ±
33 4.5 ± · · · · · · · · · · · · · · · · · · ±
26 10.7 ± · · · · · · · · · · · · ±
28 10.4 ± · · · · · · · · · · · · ±
29 19.7 ± · · · · · · · · · · · · ±
35 19.5 ± ±
44 3.4 ± · · · · · · · · · · · · Note.— These measurements incorporate all data in IF1 (Ta-ble 2). These data are plotted in Figure 6. a Lower limits to total polarization percentage. the mean value. Therefore, the steady flux levels and dis-persions derived using this method are less biased by fastflaring or fading of the YSOs as well as the occasional im-pact of potential calibration issues (e.g. see Section 3.2).We used the
BIWEIGHT MEAN routine in the IDL Astron-omy User’s Library (Landsman 1993) to iteratively es-timate the biweight mean ( F ), the biweighted standard deviation ( σ bw ), and the normalized biweighted standarddeviation σ bw / F for the 3.5-cm Stokes I fluxes (Section3.2). For the non-detections in individual epochs, we usethe 1 σ rms noise in one synthesized beam (i.e., units inmJy beam − ). The values derived from the JVLA ob-servations taken in 2012 from March 15 to September12 (Table 1), and derived from the 5 epochs of JVLAobservations taken on 2012 March 17, are given in Fig-ure 7 and 8, respectively. In the same figures, we alsoprovide the (fractional) maximal deviation of the fluxes(i.e., ∆ F max and ∆ F max /F) in these two periods.The results in Figure 7 and 8 show similar trends, sug-gesting that some of the detected flux variations within ∼
10 to 10 days can be attributed to phenomena withshorter durations. Observations separated by days mayalso be capable of characterizing (at least partially) themean flux level and the variability in shorter periods. Inaddition, adding or removing a few records does not seemto impact the statistics qualitatively, thanks to the mod-erate resistance of the biweight method. We note thatthe steady flux level of JVLA1 is potentially comparableto that of R CrA. However, JVLA1 was only detectedwhen it flared to & Fig. 8.—
Similar to Figure 7, however, we only incorporate the 5 epochs taken on 2012 March 17 in the analysis. The two sources TCrAand IRS6 were only detected once in these epochs. The values of F and σ bw for these sources are therefore upper limits, and the valuesof ∆ F max /F are lower limits. The biweight mean and standard deviation of JVLA1 in this figure are significant, because it is detected inmore than half of the epochs in these observations. beam attenuation (Table 4). The Class II source T CrA,and the Class I/II source IRS6 are the least active sourcesat all timescales (Figure 3, 4). For these two sources, onlythe upper limit of the biweight mean 3.5-cm fluxes canbe given. The faintest sources JVLA2, JVLA3, JVLA4,and IRS5N, which cannot be detected in any of the in-dividual epochs, are omitted from Figures 7 and 8, butwill be discussed in Section 4.3.For the more reddened YSOs, we found that although σ bw is the largest, F is large enough such that σ bw / F is < .
2. The biweight mean of IRS7W appears to be farlarger than for the rest of the sources, most likely due tothe fact that we cannot resolve the multiple embeddedYSOs and jet knots (Choi et al. 2004), and also becauseIRS7W is currently in a high state (Figure 3). The 3.5 cmfluxes of the less reddened YSOs have larger variationscompared to their steady flux level, as seen from theirlarger σ bw / F and ∆ F max /F.The accretion rates of the four sources IRS5, IRS1,IRS2, and IRS6, in 2002, were constrained by a near in-frared spectroscopic survey (Nisini et al. 2005) . We Nisini et al. (2005) estimated the differences between the ob-served bolometric luminosity and the stellar luminosity. We referto this original paper for uncertainties in their estimates. We areplanning to obtain new values of accretion rates in future programs. compare the biweight mean of their 3.5-cm Stokes I fluxes from March 15 to September 12 with the reportedaccretion rates (Figure 9). A weak point of this compari-son is that the radio fluxes from all available observationsare separated from the observations of the accretion ratesby ∼ M in ∼ × − M ⊙ yr − ). The accre-tion rate of IRS 2 was ˙ M in ∼ × − M ⊙ yr − . Thesource IRS6 showed no obvious accretion signature inNisini et al. (2005) and only an upper limit on the accre-tion rate of ˙ M in . × − M ⊙ yr − was given. Amongthese three sources, the accretion rates and the steady3.5-cm fluxes seem to be correlated. The younger binarysource IRS5 is deviated from this correlation (Figure 9),which suggests that comparison between different typesof sources may not be straightforward. We note that thelarge polarization percentage of IRS5 as compared withIRS7W and IRS7E (Table 6), and the large variation of2 Fig. 9.—
The biweight mean of the 3.5-cm Stokes I fluxes (mJy)and the VLT-ISAAC measured YSO accretion rates ( M ⊙ yr − )presented in Nisini et al. (2005). The 3.5 cm-flux and the accretionrate of IRS6 are both upper limits. Note that the accretion rateand the radio fluxes are not simultaneously observed. the spectral index (Figure 5), indicate that a good frac-tion of the Stokes I flux from IRS5 is non-thermal emis-sion. Because of the > V flare observed in IRS5 (Figure 6), we think that for thisparticular source, the non-thermal emission cannot befiltered out by the biweight statistics, and thus will con-tribute significantly to the steady flux level. By observ-ing where the spectral index of IRS5 converges (Figure5), we hypothesize that the flux of the more stationarythermal emission may be at most 0.4 to 0.6 mJy. The Time-Domain Structure Function
We compare our observations with the previous obser-vations reported by Feigelson et al. (1998), Forbrich etal. (2006), and Choi et al. (2008, 2009). We note thatthere were 4 additional epochs of radio observations on2005 August 09, 10, 12, and 13, reported by Forbrich& Preibisch (2007). These observations were executedin the most compact VLA D array configuration andcan contain extended emission that hinders a comparisonwith observations in more extended array configurations.They may possibly be included in our statistical analysisin the future, after the effect of the extended emissionis better modeled. We do not include these observationsin our current analysis. Nevertheless, the daily flux vari-abilities provided by these 2005 observations were alsowell sampled in our 2012 March observations (Table 1).We modified the structure function analysis introducedin Bondi et al. (1994), which was used to derive thetimescale of variability. For each YSO source (Table 3),for each pair of data points i , j , we calculate the time-lag t ij = t i - t j , and the normalized flux dispersion S ij =[ B i - B j ] /σ , where B i is the flux observed in epoch i , and σ bw is the biweighted standard deviation of the Stokes I fluxes calculated from 2012 March 15 to September 12.This analysis can only be performed for IRS7W, IRS7E,IRS5, IRS1, IRS2, and R CrA, because the other sources are too faint to be detected in individual epochs of theearlier observations. The derived t ij and S ij are plottedin Figure 10. Because our sampling of the time-domainbaselines is not very uniform, the biweight mean and thebiweighted standard deviation of S ij can only be calcu-lated in arbitrarily selected bins of time-lag. The S ij inthe time-lag range of [0.1, 1] days cannot be sampled byground based radio observations. For the time-lag binswith relatively poor statistics, possible flaring or fadingevents can dramatically bias the means and the standarddeviations of S ij . The most obvious example is observedin IRS2, in which the 0.67 mJy flux on 1997 January19/20 contributed to the large S ij values at > S ij in the time-lag range of [300, 1000] days are affectedby this poor-statistics issue for all sources. In the othertime-lag intervals, the behavior of IRS2 is similar to thatof IRS1.We observe that from the top to the bottom panels inFigure 10 (i.e., from early to late YSOs), the timescaleof the most significant flux variability shifts from over1000 days to about 1 day. In most time-lag bins, the S ij of R CrA is consistent with 1 within 2 σ . Its long-term flux variability appears to be less significant thanshort-period variations (Section 4.1). Also, we do notfind obvious decadal variability in IRS1 and IRS2. ForIRS5, the Stokes I flare with a duration of ∼ S ij in the corresponding time-lag bins. Because ofthe large circular polarization percentage during the 1998IRS5 flare event, it is likely to be (at least partially) non-thermal (Choi et al. 2009). Since no VLA observationwas taken between 1998 October 14 and 2005 February03, we cannot know for how long that IRS5 flare eventlasted. Figures 3 and 10 consistently suggest that thedecadal variability of IRS7E is marginally larger than itsshort-period variabilities, and the decadal variability ofIRS7W is significantly larger than its short-term variabil-ity. Choi et al. (2008) also suggested that source IRS7Wmay be undergoing a long duration outflow eruption. Interpretation
We think that the detected 3.5-cm Stokes I emissionfrom the young YSOs IRS7W, IRS7E, IRS5, IRS1, andIRS2 is produced by a mixture of thermal radio emissionfrom the jet cores and gyrosynchrotron emission fromthe magnetic reconnection events. The measured spec-tral indices (Figure 5) provide hints on this. Statisticalstudies on observations of solar flares with durations of1 to 1000 seconds (e.g., Nita et al. 2004) suggested thatthe centimeter-band spectral energy distribution can bedescribed by the following gyrosynchrotron-like spectralshape (Stahli et al. 1990) S ( ν ) = e A ν α lf [1 − exp( − e − B ν − β )] , (1)where S( ν ) is the radio flux at a frequency ν , and A and B are parameters affecting the normalization. The asymp-totic behavior of this spectral function above and belowthe peak frequency follows the positive spectral index α lf and the negative spectral index ( α lf − β ), respectively(e.g. Lim et al. 1994). The detected values of α lf and( α lf − β ) from the sun can be up to ∼ ±
6. At a certainobserving frequency, whether one sees positive or nega-3
Fig. 10.—
The structure function S ( τ ) of the frequently oralways detectable sources (Section 4.2). Green points show thevalues S ij calculated from the individual pairs of observations. Theblack symbols show the biweight mean and the biweighted ± σ standard deviation in the following 9 bins of time-lag (in units ofdays): [0.01, 0.1], [1.0, 3.0], [3.0, 10.0], [10.0, 30.0], [30.0, 100.0],[100.0, 300.0], [300.0, 1000.0], [1000.0, 3000.0], [5000.0, 10000.0].These irregularly separated time-lag bins are chosen to have enoughof data points each. tive spectral indices from the flares more often dependon the distribution function of the peak frequency. Dueto averaging during integration times much longer thanthe duration of the flares, the observed spectral indices α obs are likely to be closer to those of the most frequentevents. Regarding the thermal contribution to the flux, the spectral index of optically-thin thermal radio emis-sion is ∼− α obs IRS7E , α obs IRS7W , α obs IRS5 , and α obs IRS1 to-wards 0 or slightly positive values during their lower fluxstatus may be consistent with steady emission from thethermal radio jet cores (Section 3.3, Figure 5). The fre-quent α obs < > α obs IRS5 on2012 March 17 (Figure 5) varies from ∼ ∼ α obs IRS2 , besides that it seems to bepositive. We tentatively think that the emission mech-anism of IRS2 is similar to IRS1, because of the simi-larity in the Stokes I flux variability (Figures 3 and 4,and Section 4.2) as well as the similarity in their Spitzer colors (Figure 2). A non-thermal emission mechanismwas also suggested from comparisons between the X-rayand the radio emission associated with YSOs (Feigelson1998), though the underlying connection is not yet fullyunderstood (e.g., G¨udel 2002). However, X-ray emissionis a clear sign of magnetospheric activity, and in Table3, only the two radio faint Class 0/I YSOs JVLA2 andJVLA4 were not detected in the earlier deep X-ray sur-vey (Forbrich & Preibisch 2007). These non-detectionscould, however, be due to foreground extinction.The Stokes V flares detected from IRS7E and IRS7Wsupport the idea that the active magnetospheres weredeveloped while some magnetic loops occasionally breakthrough the optically thick radio emission wind (Figure6). The very low circular polarization percentages ofthese two sources as compared with that of IRS5 (Ta-ble 6) suggest that either the dominant emission mech-anism of these two sources is thermal, or that the non-thermal emission originated from the obscured inner re-gions. This is also supported by the decadal flux vari-ability of these two sources (Section 4.2). The magne-tosphere of IRS5 may be less obscured by a radio jetcore, thus its circular polarization percentage during theStokes V flares is closer to what usually is observed fromthe gyrosynchrotron sources (Andre 1996).The variability in timescales <
30 minutes of the 3.5-cm Stokes I emission from the rest of the sources, exceptJVLA2, JVLA3, JVLA4, and IRS5N (Figure 4), suggeststhat gyrosynchrotron emission is dominant. Their Stokes V emission, unfortunately, could only be detected withS/N > Spitzer colors of IRS6 are similar to IRS1 and IRS2 (Fig-ure 2), the radio flux and variability of this weakly ac-creting YSO (Section 4.1) behave more similar to thoseof the Class II source T CrA (Figure 3, 4). The ClassIII source JVLA1 (CrA PMS 1) is also only occasion-ally detected at 3.5 cm, but shows a peak flux ∼ α obs ∼ ∼
15 years). Ac-cretion disk instabilities could cause variability on sucha timescale. A deep infrared spectral-line survey is re-quired to check whether the radio faint Class 0/I YSOsare weakly accreting sources. From inspecting the fluxvariability of the radio-faint Class 0/I YSOs, it seemsthat the large-scale magnetosphere is more active in theradio when strong mass-loss occurs (see also Figures 7and 8). Another possible exception is the luminous ra-dio emission from some FU Orionis stars (Rodriguez etal. 1990; Anglada et al. 1994; Vel´azquez & Rodr´ıguez2001), which might be classical T Tauri stars (Class II)during their quiescent phase (Hartmann & Kenyon 1996).Because of the small number of observed YSOs in ourstudy, in particular in the Class II and Class III stages,the proposed scenario needs to be verified by more ex-tensive surveys. SUMMARY AND OUTLOOK
We performed 8-10 GHz monitoring observations to-wards the young stellar cluster R CrA in 2012, using the JVLA. Efforts have been taken to ensure that the changesof the JVLA array configurations do not interrupt orhinder the analysis of long-term radio flux variations.We found that for this particularly nearby field, afterimplementing a cut in uv distance > kλ , the effectsof changing the JVLA array configuration are negligiblecompared with the daily and hourly radio flux variabili-ties.From comparison with previous observations, radioflux variability was detected in timescales from <
30 min-utes, up to ∼
15 years. Our current consensus is thatthe 3.5-cm radio emission from YSOs is dominated bythe active magnetosphere and by the thermal emissionwind. The active magnetosphere, which produces hourlyand daily radio flux variability, is developed as earlyas when the transition from the Class 0 to the Class Iphase occurs. The thermal wind seems to be correlatedwith the accretion rate and varies in a longer, dynamicaltimescale. The optically thick wind can partially obscurethe active magnetosphere during the earliest stages, thusalleviating the non-thermal confusion in the diagnosis ofthe thermal radio jet variability. In stages later thanClass II, the mass loss becomes weak, and the radio fluxis dominated by gyrosynchrotron emission from the stel-lar magnetic field.Our scheme needs to be verified because of the smallnumber of observed YSOs. In particular, there is onlyone detectable Class II YSO and two detectable Class IIIYSOs in our sample. Besides, the radio flux variabilityof exceptional cases like the FU Orionis objects cannotyet be incorporated. Those exceptional cases might bevery important for understanding protostellar evolution.A more extensive JVLA survey may shed light on theseissues.We also note that the right radio emission mechanismsshould show not only the correct timescales, spectral in-dices, or polarization percentages, but also the charac-teristic flux scales. In the future, more sensitive obser-vations using the Square Kilometer Array (SKA) will beimportant for characterizing the physical mechanisms offainter, shorter duration emission.HBL thanks Dr. Joseph L. Hora for his help and theuseful comments while organizing this project. R.G.-M.acknowledges funding from the European Community’sSeventh Framework Programme (/FP7/2007-2013/) un-der grant agreement No. 229517R.
Facilities:
JVLA
REFERENCESAMI Consortium, Ainsworth, R. E., Scaife, A. M. M., et al. 2012,MNRAS, 423, 1089Anderson, I. M., Harju, J., Knee, L. B. G., & Haikala, L. K.1997a, A&A, 321, 575Anderson, I. M., Harju, J., & Haikala, L. K. 1997b, A&A, 326,366Andre, P. 1996, Radio Emission from the Stars and the Sun, 93,273Andreazza, C. M., & Vilas-Boas, J. W. S. 1996, A&AS, 116, 21Anglada, G. 1995, Revista Mexicana de Astronomia y AstrofisicaConference Series, 1, 67Anglada, G., Rodriguez, L. F., Girart, J. M., Estalella, R., &Torrelles, J. M. 1994, ApJ, 420, L91Anglada, G., Villuendas, E., Estalella, R., et al. 1998, AJ, 116,2953 Allen, L. E., Calvet, N., D’Alessio, P., et al. 2004, ApJS, 154, 363Chen, B., Bastian, T. S., White, S. M., et al. 2013, ApJ, 763, L21Benz, A. O., & Guedel, M. 1994, A&A, 285, 621Bower, G. C., Plambeck, R. L., Bolatto, A., et al. 2003, ApJ, 598,1140Bondi, M., Padrielli, L., Gregorini, L., et al. 1994, A&A, 287, 390Brown, A. 1987, ApJ, 322, L31Calvet, N., Hartmann, L., & Kenyon, S. J. 1993, ApJ, 402, 623Carpenter, J. M., Hillenbrand, L. A., & Skrutskie, M. F. 2001,AJ, 121, 3160Chen, W. P., & Graham, J. A. 1993, ApJ, 409, 319Chen, X., & Arce, H. G. 2010, ApJ, 720, L169Chini, R., K¨ampgen, K., Reipurth, B., et al. 2003, A&A, 409, 235Choi, M., & Tatematsu, K. 2004, ApJ, 600, L555