Observations of V404 Cygni during the 2015 outburst by the Nasu telescope array at 1.4 GHz
Kuniyuki Asuma, Kotaro Niinuma, Kazuhiro Takefuji, Takahiro Aoki, Sumiko Kida, Hirochika Nakajima, Kimio Tsubono, Tsuneaki Daishido
aa r X i v : . [ a s t r o - ph . H E ] J un Publ. Astron. Soc. Japan (2018) 00(0), 1–9doi: 10.1093/pasj/xxx000 Observations of V404 Cygni during the 2015outburst by the Nasu telescope array at 1.4 GHz
Kuniyuki A
SUMA , Kotaro N
IINUMA ∗ , Kazuhiro T AKEFUJI , TakahiroA
OKI , Sumiko K IDA , Hirochika N AKAJIMA , Kimio T SUBONO andTsuneaki D AISHIDO Faculty of Science and Engineering, Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo169-8555 Asaka High School, Saiwaicho 3-13-65, Asaka-shi, Saitama 351-0015 Graduate School of Sciences and Technology for Innovation, Yamaguchi University, Yoshida1677-1, Yamaguchi 753-8512 Kashima Space Research Center, National Institute of Information and CommunicationsTechnology, 893-1 Hirai, Kashima-shi, Ibaraki 314-8501 Japan Aerospace Exploration Agency, 3-1-1 Yoshinodai, Chuo-ku, Sagamihara, Kanagawa229-8510, Japan The Research Institute for Time Studies, Yamaguchi University, Yoshida 1677-1, Yamaguchi753-8511 School of Education, Waseda University, 1-104 Totsukamachi, Shinjuku-ku, Tokyo 169-8050 ∗ E-mail: [email protected]
Received ; Accepted
Abstract
Waseda University Nasu telescope array is a spatial fast Fourier transform (FFT) interferometerconsisting of eight linearly aligned antennas with 20 m spherical dishes. This type of interfer-ometer was developed to survey transient radio sources with an angular resolution as high asthat of a 160 m dish with a field of view as wide as that of a 20 m dish. We have been per-forming drift-scan-mode observations, in which the telescope scans the sky around a selecteddeclination as the earth rotates. The black hole X-ray binary V404 Cygni underwent a newoutburst in 2015 June after a quiescent period of 26 years. Because of the interest in blackhole binaries, a considerable amount of data on this outburst at all wavelengths was accumu-lated. Using the above telescope, we had been monitoring V404 Cygni daily from one monthbefore the X-ray outburst, and two radio flares at 1.4 GHz were detected on June 21.73 andJune 26.71. The flux density and time-scale of each flare were ± mJy and . ± . days, ± mJy and . ± . days, respectively. We have also confirmed the extremevariation of radio spectra within a short period by collecting other radio data observed withseveral radio telescopes. Such spectral behaviors are considered to reflect the change in theopacity of the ejected blobs associated with these extreme activities in radio and X-ray. Our1.4 GHz radio data are expected to be helpful for studying the physics of the accretion andejection phenomena around black holes. Key words:
Radio continuum: stars — X-rays: binaries — Stars: black holes — Stars: individuals (V404 c (cid:13) Publications of the Astronomical Society of Japan , (2018), Vol. 00, No. 0
Cygni) — Instrumentation: interferometers
The Nasu telescope array operates as a spatial FFT in-terferometer. In this scheme, an antenna array and anFFT processor produce a direct image of an arriving ra-dio source. A number of topics dealt with here have beenreviewed by Thompson et al. (2017). The idea of a spa-tial FFT interferometer was proposed in 1984 and demon-strated using an eight-element test telescope at the cam-pus of Waseda University (Daishido et al. 1984; Daishidoet al. 1987). Afterwards, two 8 × ×
20 m) was designed for surveying transient radiosources, and the smaller one (1 . × . × × ∼ M ⊙ black hole and a ∼ M ⊙ companion with an orbital periodof 6.5 days. Later radio parallax measurements determinedits distance to be ∼ . Suppose that incident radio waves arrive at antennasplaced at equally spaced grid points in a plane. By spa-tially Fourier transforming the electric field sampled byeach antenna, we can obtain a map of the incident field in k -space. Thus, the combination of an antenna array and areal-time spatial FFT processor acts as a digital lens, whichproduces a direct image of the incoming waves. We call thistype of detector a spatial FFT interferometer. This typeof interferometer is suitable for surveying transient radiosources. Assume that an infinite number of antennas are distributedcontinuously on an x − y plane, and then a plane radiowave arrives at the antenna plane. The electric field onthe antenna located at r is written in the form E ( r , t ) = E e i ( k r − ω t ) , (1)where E , k and ω are the amplitude of the electricfield, the wave-number vector and the angular frequency ofthe incident wave, respectively. Note that E , E , k and r are all two-dimensional vectors on the x − y plane. If weexecute the Fourier transformation of E ( r ,t ) with respectto the spatial coordinate r , we obtain g E ( r , t ) = (cid:16) π (cid:17) Z ∞−∞ Z ∞−∞ E e i ( k r − ω t ) e − i kr d r = E e − iω t δ ( k − k ) . (2)Therefore, by finding the peak of e E in k -space, we candetermine the direction k of the incident wave. Figure 1illustrates the peaks appearing in k -space originating fromthe incident waves from (a) a point source and (b) two ublications of the Astronomical Society of Japan , (2018), Vol. 00, No. 0 Fig. 1. (a) Image on the k -plane formed by a point source in the k -direction. (b) Images formed by two independent point sources. independent sources. The incident angle θ with respect tothe vertical axis is calculated from the following relationsin θ = λ π k , (3)where λ is the wavelength of the incident wave.If the distribution of antennas is finite, the δ -functionin equation (2) is replaced by a broadened function. Forsimplicity, let the antennas be distributed inside the area | x | ≤ b/ | y | ≤ b/
2; then equation (2) with a definite in-tegral interval yields a spatial frequency response functionwith bandwidth∆ k ∼ πb . (4)From equation (3), we can derive the angular resolution as∆ θ ∼ λ b . (5)More practically, if the antennas are distributed discretely,not continuously, on the grid points with equal intervals d , the sampling theorem implies that the output of equa-tion (2) has a periodic pattern with interval∆ k = 2 πd . (6)Thus, if the incident angle θ satisfies | θ | ≪
1, the field ofview is given by∆ θ ∼ λ d . (7)As described above, the angular resolution is given bythe overall size of the antenna array, while the field ofview is determined by the spacing between the antenna el-ements. We can thus design the configuration of the spatialFFT interferometer appropriately according to the purposeof the observation.Most of the traditional Fourier synthesis telescopes inuse today were designed to obtain fine images of radiosources using a relatively small number of antennas bychoosing minimum- redundancy baselines or an arbitraryconfiguration of antennas. They are indirect imaging sys-tems that use correlators and integrators. On the otherhand, the spatial FFT interferometer, in which each an-tenna element is fixed in the maximum redundancy po-sition, generates real-time images of radio sources at the Nyquist sampling rate. Moreover, an N -element spatialFFT interferometer requires N log N multipliers, whilean N -element Fourier synthesis interferometer requires N ( N −
1) correlators. This means that the spatial FFTtype is more economical than the correlator type when thenumber of elements N is large (Daishido et al. 1984). Fig. 2.
Nasu telescope array with 20 m spherical antennas at the Jiyu-Gakuen Nasu Farm.
The Nasu telescope array is located at the Jiyu-GakuenNasu Farm in Tochigi Prefecture, 160 km north of Tokyo,at latitude 36 ◦ ′ . ′′ north and longitude 139 ◦ ′ . ′′ east. The telescope is a one-dimensional array of eight an-tennas with a spacing of 21 m, aligned from east to west(E-W). The main part of each antenna is a spherical dishwith a diameter of 20 m. A photograph of the 20 m an-tennas is shown in figure 2. The radius of curvature andthe aperture diameter are both 20 m; that is, each an-tenna is a 60 ◦ spherical cap. For an incoming radio wave,the spherical surface of the main reflector and a Gregoriansub-reflector form a focal point at the input of the feedhorn (see figure 3). The asymmetrical 3D surface of thesub-reflector was specially designed to compensate for theaberrations caused by the spherical reflector (Daishido etal. 2000).The main dish is fixed on the ground, whereas the sub-reflector and feed horn can be moved mechanically. Thesub-reflector is located 5 ◦ from the vertex axis of the spher-ical reflector, and thus, the elevation angle is fixed to 85 ◦ .By rotating the sub-reflector synchronously with the feedhorn in azimuth, the antenna covers the sky area in a dec-lination zone of 32 ◦ ≤ δ ≤ ◦ , which is 7.0% of the entiresky.A block diagram of the Nasu telescope array, includingthe data processing system, is shown in figure 4. The de-tector system operates at a center frequency of 1.415 GHz Publications of the Astronomical Society of Japan , (2018), Vol. 00, No. 0
Fig. 3.
Principal parts of each antenna: main dish, sub-reflector, and feedhorn.
Fig. 4.
Block diagram of the Nasu telescope array including the data pro-cessing system. with a bandwidth of 20 MHz. The front-end signal fromeach feed horn is fed into an HEMT low-noise amplifierwith a noise temperature of ∼
40 K. After being filteredwithin a 1.405–1.425 GHz bandwidth and amplified by 40dB, the signal is downconverted to 20 MHz by a com-plex mixer with two 1.415 GHz references in quadrature.The complex signal is sampled by an 8-bit analog-to-digitalconverter every 50 ns. The digitized signal is subsequentlyprocessed by an FPGA-based processor. The phase fluctu-ation arising from the temperature-dependent delay in thetransmission line is compensated for by the phase equal-izer. The 8-ch complex signal x ( n ) ( n = 0 , , ...,
7) is thenspatially Fourier transformed to obtain X ( k ) as follows X ( k ) = X n =0 x ( n ) exp( − i π kn . (8)Finally, the time-averaged vector amplitude ch( k ) = | X ( k ) | (9)is calculated and sent to the storage device.The main parameters of the Nasu telescope array aresummarized in table 1. Table 1.
Main parameters of the Nasu telescope array.
Location Latitude 36 ◦ ′ . ′′ northLongitude 139 ◦ ′ . ′′ eastNumber of antennas Eight with 21 m spacing (E-W)Main reflector Fixed 20 m spherical dishSub-reflector Asymmetrical Gregorian typeAngular resolution 0 . ◦ (E-W)Field of view (HPBW) 0 . ◦ Declination coverage 32 ◦ ≤ δ ≤ ◦ Frequency range 1.415 ± The directivity of the spatial FFT interferometer is de-termined by the configuration of the antenna array. Onthe basis of the characteristics of the directivity pattern,two analysis methods have been developed: one is directimaging and the other is correlation analysis.
In a spatial FFT interferometer, each ch( k ) in equation (9)has its own directivity pattern, which we will derive here.First, neglecting the curvature of the antenna dish, we canobtain the directivity F C of a single round dish for anincoming electric wave e i k r as follows F C = Z r ≤ a e i k r dS = πa J ( k a ) k a , (10)where a is the radius of the dish. Next, the array factor F A for N -element linearly aligned antennas with an equalspacing of d is calculated as F A ( k ) = N − X n =0 e − i πN nk e in k d = e i ( N − k d − πN k ) sin( κ − k ) π sin( κ − k ) πN , (11)where we defined the modified incident direction κ as κ = N k d π . Then we can obtain the following normalizeddirectivity pattern G ( k ) for each ch( k ) ( k = 0 , , ,...,N − G ( k ) = | F C F A ( k ) | ublications of the Astronomical Society of Japan , (2018), Vol. 00, No. 0 −16 −12 −8 −4 0 4 8 12 16 incident wave (κ ) d i r e c t i v i t y G ( ) ch(0) −16 −12 −8 −4 0 4 8 12 16 incident wave (κ ) d i r e c t i v i t y G ( ) ch(1) −16 −12 −8 −4 0 4 8 12 16 incident wave (κ ) d i r e c t i v i t y G ( ) ch(2) −16 −12 −8 −4 0 4 8 12 16 incident wave (κ ) d i r e c t i v i t y G ( ) ch(3) −16 −12 −8 −4 0 4 8 12 16 incident wave (κ ) d i r e c t i v i t y G ( ) ch(4) −16 −12 −8 −4 0 4 8 12 16 incident wave (κ ) d i r e c t i v i t y G ( ) ch(5) −16 −12 −8 −4 0 4 8 12 16 incident wave (κ ) d i r e c t i v i t y G ( ) ch(6) −16 −12 −8 −4 0 4 8 12 16 incident wave (κ ) d i r e c t i v i t y G ( ) ch(7) Fig. 5.
Normalized directivity pattern G ( k ) for each ch( k ) ( k = 0 , , ,..., as a function of κ . Here we assume that k is parallel to d and d = 2 a .The solid line shows the antenna pattern for each ch( k ) and the broken lineshows that of a single dish. = (cid:12)(cid:12)(cid:12)(cid:12) J ( k a ) k a sin( κ − k ) π sin( κ − k ) πN (cid:12)(cid:12)(cid:12)(cid:12) . (12)Assuming that N = 8, k is parallel to d and d = 2 a , theantenna pattern G ( k ) as a function of κ is shown in figure5. A direct image can be obtained by making a contour mapon the array block of ch( k ) in equation (9). Figure 6 showsan example of a direct image obtained from the brightradio source 4C 33.57 (1.1 Jy at 1.4 GHz; Condon et al.1998). As shown in figure 6 (a), the same two sets of ch( k )reduce the complexity of the folded pattern arising fromthe Fourier transformation. Furthermore, we can obtainthe clearer image, as shown in figure 6 (b) by masking theunnecessary part of the array before making the contourmap.In principle, the direct imaging of a radio source showsintensity variations with a short duration. For this pur-pose, however, an accurate calibration of the image is cru-cial, although the optimum method that should be usedis still under investigation. Nevertheless, we can applydirect imaging to distinguish artificial noise, such as radio-frequency interference (RFI), from genuine astronomicalradio signals. Direct imaging cannot be applied to veryfaint radio sources. Fig. 6.
Example of direct image produced by the bright radio source 4C33.57 (1.1 Jy at 1.4 GHz). The integration time of ch( k ) is 0.6 s. (a) Rawimage appearing in the same two sets of ch( k ). (b) Masked view of thesame image. As shown in figure 5, each ch( k ) has its own antenna pat-tern. We can therefore find signals buried in an outputstream by using correlation analysis or a pattern-matchingmethod. This technique is a very powerful tool in the dataanalysis of gravitational waves, especially for the chirpwaves generated by a binary coalescence (Abbott et al.2016). In correlation analysis, the observed signal-to-noiseratio (SNR) for each ch( k ) is formulated as follows (see,for example, Creighton et al. 2011): Fig. 7.
Example of three-hour data analyzed by the pattern-matchingmethod. The upper figure shows the raw data of ch(0). The middle fig-ure shows the combined SNR of the 8-ch output data. SNR peaks largerthan 7 are marked with a small circle; the figures by the peaks are observedSNRs. Peaks with an open circle show the side-lobe from strong sources.The lower figure shows the expected flux density of the radio sources listedin the NRAO VLA Sky Survey (NVSS) catalog (Condon et al. 1998) takingaccount of the difference in the declinations between the source and theantenna position.
SNR = 4Re Z + ∞ ˜ x ( f )ˆ s ∗ ( f ) e πift P one ( f ) df. (13)In our case, ˜ x ( f ) is the Fourier transform of ch( k ) withrespect to time t , ˆ s ( f ) is the normalized Fourier transformof the antenna pattern with respect to time t , and P one ( f )is the one-sided power spectral density of ch( k ). This ex-pression is essentially a noise-weighted correlation of the Publications of the Astronomical Society of Japan , (2018), Vol. 00, No. 0 anticipated wave form with the actual data.An example of three-hour data analyzed by the pattern-matching method is shown in figure 7. The upper figureshows the raw data of ch(0). The middle figure shows thereduced SNR of the 8-ch output data. SNR peaks largerthan 7 are marked with a small circle, and the values shownat the peaks are the observed SNRs. The lower figureshows the expected flux density of the radio sources listedin the NVSS catalog, taking account of the difference in thedeclinations between the source and the antenna positions.We obtained good agreement between the expected andobserved source distributions and their intensities.The observed one-sigma noise level was ∼
20 mJy whenthe averaging time of ch( k ) is 0.6 s. Thus, this method iseffective even for rather weak radio sources. The Nasu telescope array started an observation run on2015 May 18, monitoring the sky region inside the declina-tion zone of 33 . ◦ ± . ◦ in the drift-scan mode, in whichthe telescope scans the sky around a selected declination asthe earth rotates. V404 Cygni is located at 20 h m . s
83 inright ascension (RA) and 33 . ◦ in declination (Dec); thus,this source was within our survey area. Fig. 8.
Daily records of the one-hour SNR data from June 19 to June 22.An enlarged view of a part of the figure is also shown. Peaks with an opencircle show the side-lobe from strong sources. On June 21 17:24 UT, wefound a significant peak at a position near V404 Cygni, although on June 2017:28 UT we could not observe any noticeable signal in the vicinity. A nearbyradio source, QSO J2025+337, was chosen as a calibration star, whose fluxdensity is 1268 mJy at 1.4 GHz (Condon et al. 1998).
Figure 8 shows the daily records of the one-hour SNRdata calculated by the correlation analysis method. In theplot, the horizontal axis shows the RA of the target skyregion. Four days of data from June 19 to June 22 are plot-ted here to demonstrate how to find substantial changes in the brightness among the daily records. On 2015 June 21,17:24 UT, we found a significant peak at a position nearV404 Cygni, although on June 20 17:28 UT, we could notobserve any noticeable signal in the vicinity. A nearby ra-dio source, QSO J2025+337, whose flux density is 1268mJy at 1.4 GHz, was chosen as a calibrator (Condon etal. 1998). The calibrated flux density of the detected sig-nal on June 21 was 313 ±
30 mJy (Tsubono et al. 2015a).Since the center position of V404 Cygni and the calibratoris almost on the observation line, the difference in direc-tivity was not considered here. The daily variation of thedetected flux density during the two months from May 18(MJD 57160) to July 18 UT (MJD 57221), which includesthe period of the V404 Cygni outburst, is plotted in figure9. Except for the 10 days from June 21 (MJD 57194) toJune 30 UT (MJD 57203), the detected signal was belowthe detection limit.
Fig. 9.
Daily variation of the detected flux density around V404 Cygni duringthe two months from May 18 (MJD 57160) to July 18 UT (MJD 57221) includ-ing the period of the V404 Cygni outburst. Except for the 10 days from June21 (MJD 57194) to June 30 UT (MJD 57203), the detected signal was belowthe detection limit (filled red triangle). These detection limits can sometimesbe high, mainly due to bad weather at the observation site. The date of theX-ray outburst is also shown here.
On June 21 17:24 UT, we found a significant peak of 313 ±
30 mJy at the position of V404 Cygni. Then, the ob-served flux density decreased to 105 ±
30 mJy on June 2317:17 UT. However, the flux density recovered from thatpoint and increased to 364 ±
30 mJy on June 26 17:05UT. On June 27 17:01 UT, we observed fast decay of theintensity (Tsubono et al. 2015b). After this rapid decay,the flux density seemed to slowly decay. These character-istics appeared to be similar to those of the radio decaycurve reported for the 1989 outburst of V404 Cygni (Han& Hjellming 1992).
As mentioned in Section 1, Barthelmy et al. (2015) andNegoro et al. (2015) reported that Swift-BAT and MAXI- ublications of the Astronomical Society of Japan , (2018), Vol. 00, No. 0 GSC captured the huge X-ray outburst, which precededthe radio outburst, in the black-hole X-ray binary ob-ject V404 Cygni. After this X-ray observation and in re-sponse to the alert about its detection, follow-up observa-tions were performed by ground-based telescopes, includ-ing not only radio telescopes but also other telescopes suchas optical telescopes. On the other hand, fortunately, wehad been monitoring the region at a declination line of33 . ◦ ± . ◦ , at which V404 Cygni is located, every daysince 2015 May 18, which was almost one month beforethe X-ray outburst and during which time V404 Cygniwas in the quiescent state in the 1.4 GHz radio band.Although an accurate understanding of the flux increasein the flaring phase of non-thermal synchrotron emission isessential for investigating the coupling between the accre-tion state and the jet ejection mechanism, which is one ofthe most important questions, it is quite difficult to obtainearly-stage information of the flare especially for radio ob-servation, not only at higher frequencies but also at lowerfrequencies. Because of their smaller field of view (FoV),most radio telescopes have not performed unbiased sur-veys, but have carried out follow-up observations of tran-sient phenomena triggered by X-ray/ γ -ray observations byall-sky monitoring satellites.As shown in figure 9, it was revealed that the radiobehavior of this source was clearly very quiescent aroundJune 15 when the first X-ray outburst was recognized inthis object. It is easy to specify the date on which theradio outburst occurred at 1.4 GHz with a time resolutionof one day. This daily light curve clearly shows two radiooutbursts at 2015 June 21.73 (the first flare: F1) and 26.71(the second flare: F2). In order to estimate the time scaleof flares, we performed the structure function analysis (e.g.,Simonetti et al. 1985). The first-order structure function SF ( τ ) is described as follows: SF ( τ ) = D [ S ( t ) − S ( t + τ )] E (14)where S is the observed flux density, and τ is the time lag. SF ( τ ) derived from daily light curve at 1.4GHz gives usa characteristic time scale of ∼ F ( t ) = F c (cid:16) e t − ttr + e t − t td (cid:17) + F b (15)where, F b is a baseline level of light curve in quiescentstate, F c is an amplitude coefficient, t is estimated timeof the flare peak, t r and t d are e-folding times in rise anddecay phase, respectively. Here, we define the duration of individual flare as T flare ( ≡ t r + t d ), and T flare of F1 and F2derived from the well-fitted results with Eq.(15) are 1 . ± .
49 days and 1 . ± .
16 days, respectively. Surprisingly,the flux density of F1 drastically increased by more than10-times within a day compared to that in the quiescentstate (e.g., in 2015 June 20.73). On the other hand, it isdifficult to understand the situation in which t d of F2 isclearly shorter than t r in the framework of the synchrotronregime. Therefore, the flux peak at 1.4 GHz in F2 possiblyappeared between June 25.71 and 26.71. T flare suggests that the size of the emitting regionwhere both radio outbursts occurred is estimated to be R < ∼ . × cm, which is derived from the light-crossingtime ( R < c ∆ t ) under the assumption of non-relativisticphenomena. Other radio follow-up observations with hightime resolution but short duration clearly revealed in-traday variability from this object (e.g., Gandhi et al.2017; Tetarenko et al. 2017; Tetarenko et al. 2019) espe-cially at higher radio frequencies ( > B discussed in Chandra & Kanekar (2017), the time scaleof synchrotron cooling at 1.4 GHz should be t syn ≈ . (cid:16) B .
25 G (cid:17) − (cid:16) γ (cid:17) − (cid:16) δ (cid:17) − yr , (16)where γ is Lorentz factor of electron, δ is Doppler factorof moving blob. This time scale is a thousand times longerthan T flare for both F1 and F2. Therefore, adiabatic ex-pansion (van der Laan 1966) is considered to be one of themost dominant cooling mechanisms that can explain thecooling time scale of these outbursts.Figure 10 compares radio spectra in the quiescent phase(phase 1) and outburst phases (phases 2 & 3) by usingarchived data obtained within a couple of hours, includingour 1.4 GHz observation, as shown in Table 2. The timeevolution of the radio spectra between the quiescent phaseand the outbursts can be seen in this figure. Althoughthe peak frequency ( ν p ) in phase 1 is located higher than ∼
10 GHz, ν p in phases 2 & 3 is located at around a fewGHz. Such a change in spectral features possibly causedby non-thermal jet ejection is similar to that of the out-burst detected in blazars at mm to sub-mm wavelengths(e.g., Le´on-Tavares et al. 2011), but the time scale wasmuch shorter (i.e., within several days) compared to suchextragalactic phenomena that occurred in the vicinity ofsupermassive black holes. Additionally, in phase 3, dou-ble peaks at ∼ ∼ Publications of the Astronomical Society of Japan , (2018), Vol. 00, No. 0
Fig. 10.
Radio spectra of V404 Cygni in the quiescent (phase 1; solid line)and outburst phases (phase 2&3: dotted and dashed lines, respectively) at1.4 GHz. Each color indicates radio spectra obtained in different phases.Filled green triangle shows an upper limit for 1.4 GHz radio flux density inphase 1. Additionally, filled red diamond shows 0.34 GHz radio flux density,but it was obtained more than 10 hours later compared to other radio fluxdensities in phase 2.
June 26, a flux density of 364 ±
30 mJy at 1.4 GHz wasmeasured by the Nasu telescope. As seen in the phase 3line of figure 10, this Nasu data point at 1.4 GHz togetherwith the nearby RATAN’s one at 2.3 GHz can be a strongsupport for the double-peak hypothesis. As also discussedin Chandra & Kanekar (2017), there is a possibility thatthis double-peak spectrum implies a mixture of multiplesynchrotron blobs ejected by outbursts. Actually, a verylong baseline interferometry (VLBI) follow-up observationcarried out immediately after the X-ray outburst directlyrevealed the ejection of multiple non-relativistic jets andchanging jet orientations (Miller-Jones et al. 2019). It isthought that the ν p of ∼ ∼ ν p of ∼ We used the Nasu telescope array, which is a spatial FFTinterferometer, to carry out daily monitoring of radio tran-sient phenomena by drift-scan observation. This telescopemakes it possible to perform observation of all right ascen-sion with a sensitivity of a few tens of mJy at a specificdeclination with a positional uncertainty of ± . Table 2.
Multi-frequency radio data in the quiescent and outburstphases.
Epoch ν S S err
Date Ref.(GHz) (mJy) (mJy)Phase1 1.4 <
30 - 2015-Jun-18.73 (1)4.6 76 4 2015-Jun-18.95 (2)8.2 108 5 2015-Jun-18.95 (2)11.2 129 6 2015-Jun-18.95 (2)21.7 210 25 2015-Jun-18.95 (2)Phase2 0.34 186 6 2015-Jun-22.54 (3)1.4 313 30 2015-Jun-21.73 (1)2.3 730 50 2015-Jun-21.94 (2)4.6 540 30 2015-Jun-21.94 (2)8.2 317 10 2015-Jun-21.94 (2)11.2 282 10 2015-Jun-21.94 (2)21.7 200 20 2015-Jun-21.94 (2)Phase3 0.24 188 27 2015-Jun-26.89 (4)0.61 470 49 2015-Jun-26.89 (4)1.28 739 77 2015-Jun-26.89 (4)1.4 364 30 2015-Jun-26.71 (1)2.3 320 30 2015-Jun-26.93 (2)4.6 534 50 2015-Jun-26.93 (2)8.2 370 40 2015-Jun-26.93 (2)11.2 292 30 2015-Jun-26.93 (2)21.7 150 20 2015-Jun-26.93 (2)97.5 65.2 0.2 2015-Jun-26.93 (5)140.5 46.9 0.3 2015-Jun-26.93 (5)
Column
1: The state of source activity.
Column
2: Observed fre-quencies in GHz,
Columns
Column
5: Date on which each datum was obtained. Here we as-sume that radio data at low frequencies obtained within severalhours show the same behavior.
Column
6: References. (1) Ourobservation, (2) Trushkin et al. (2015), (3) Kassim et al. (2015), (4)Chandra & Kanekar (2017), (5) Tetarenko et al. (2015) complementary to other radio telescopes, which can con-duct not only monitoring of specific objects with high timeresolution or high angular resolution, but also unbiasedsurveys with large FoV in other regions (e.g., CHIME/FRBCollaboration et al. 2018).As a successful demonstration, our daily monitoringwith the Nasu telescope from one month before the out-burst of V404 Cygni occurred in June 2015 allowed us tocapture the detailed behavior, which showed an abruptchange in flux density caused by an outburst in the 1.4GHz radio band. The radio light curve showed two hugeoutbursts that occurred during a period of 10 days, andat least the 1.4 GHz flux density during one of the twooutbursts increased by more than ten times within a day, ublications of the Astronomical Society of Japan , (2018), Vol. 00, No. 0 compared to the one in the quiescent state. We also haveconfirmed the extreme variation of radio spectra withinthe short period mentioned above by collecting other radiodata observed at several radio observatories. Such spectralbehaviors are considered to reflect the change in the opac-ity of the ejected blobs associated with radio and X-rayoutbursts.Although there is room to improve the calibration is-sues, which is crucial for accurate measurement of the am-plitude and position of radio transients, as mentioned inThompson et al. (2017), this new type of wide-field radiotelescope realized by adopting a spatial FFT technique isone of the most promising key techniques in the era of“time-domain and multi-messenger astronomy”. Acknowledgments
We are grateful to the anonymous referee, whose suggestionsimproved our paper substantially. This work was supportedin part by Cooperative and Supporting Program for Researchand Education in Universities of the National AstronomicalObservatory of Japan (NAOJ) and by Japan Society forthe Promotion of Science (JSPS) KAKENHI Grant NumberJP15K05029 (K.T.), JP18H03721 (K.N.), JP15H00784 (K.N.).We thank Dr. Trushkin for providing us with their RATAN-600data.