East Asian VLBI Network observations of active galactic nuclei jets: imaging with KaVA+Tianma+Nanshan

The TMRT is operated by Shanghai Astronomical Observatory (SHAO) as a joint project between the CAS and the city of Shanghai. The project was approved at the end of October 2008. After laying the foundation in December 2009, the construction of the telescope started in March 2010 and completed with the first-light detection in October 2012. The location of the telescope is in Sheshan, Songjiang District of Shanghai (Jiang et al. 2018).

Through cooperation with another 25-m-diameter radio telescope in Sheshan, the first interferometric fringes of TMRT were detected with a 6.1-km baseline in November 2012. After that, the TMRT started commissioning and made great contributions to a broad range of radio astronomy researches such as blazars, microquasars, molecular spectral lines, pulsars, X-ray binaries and geodynamics based on both single-dish and VLBI modes (e.g., Li et al. 2016; Yang et al. 2019; Hou & Gao 2020). As a representative radio telescope in China, TMRT joined EAVN from the beginning of this project since 2013. TMRT also regularly joined the European VLBI Network (EVN) since 2014, in particular at low frequencies (e.g., 5 GHz).

At present, TMRT is the largest fully steerable telescope in East Asia. It has a wide range of observing wavelengths covering 1–50 GHz with eight bands, namely L (1.25–1.75 GHz), S (2.2–2.4 GHz), C (4–8 GHz), X (8.2–9.0 GHz), Ku (12–18 GHz), K (18.0–26.5 GHz), Ka (30–34 GHz) and Q (35–50 GHz) bands (Yan et al. 2015). The S/X- and X/Ka-band receivers are the dual-frequency co-axis feed. The common observing frequencies of TMRT with KaVA are 22 GHz and 43 GHz (and partially also 6.7 GHz). The maximum aperture efficiency of TMRT that can be reached at 53° elevation is around 50% at 22 GHz and 45% at 43 GHz with the active surface control system (see Table 1). The active surface controller is set 'ON' by default. The nominal root mean square (rms) of surface accuracy is 0.6 mm without an active surface and 0.3 mm with active surface at the elevation around 53°. The sub-reflector surface accuracy is 0.1 mm rms. The typical pointing accuracy is 3 arcsec when the wind speed is 4 m s⁻¹ and 30 arcsec if the wind speed reaches 20 m s⁻¹. The slew rate is 0.5° s⁻¹ and 0.3° s⁻¹ in azimuth and elevation directions respectively. Dual-pixel receivers are installed in TMRT at both 22 and 43 GHz which enable simultaneous observations of multiple lines (Zhong et al. 2018). These two beams have fixed separation angles of 140 arcsec at 22 GHz and 100 arcsec at 43 GHz. One of the beams is placed at the antenna focus for VLBI observations. The half power beam width (HPBW) is the measured beam sizes listed in Table 1.

The NSRT in Urumqi, operated by Xinjiang Astronomical Observatory, is another key station in CVN, constructed in 1993. The telescope is located at the northern foot of Tianshan with high elevation above 2029 m and over 75 km away from Urumqi city which is surrounded by a superior observing environment with dry climate and low-level radio frequency interference. Originally the telescope was designed with a diameter of 25 m, but a refurbishment of the telescope was made from early 2014 and completed in late 2015 (Xu et al. 2018). This resulted in an enlargement of the main reflector to 26m and improvement of the antenna surface accuracy. After this reconstruction was finished, NSRT participated in EAVN commissioning. NSRT also joined EVN since early 1994 and International VLBI Service for Geodesy and Astrometry (IVS) after 1996. Over the past 20 years, NSRT has made rich contributions to pulsar timing and physics, molecular line survey and star formation, AGN jets and flux monitoring based on both single-dish and VLBI modes (e.g., Wang et al. 2005; Yuan et al. 2010; Wu et al. 2018; Cui et al. 2010; Liu et al. 2012).

NSRT features a Cassegrain-type design with a 26-m diameter main reflector and a 3-m sub-reflector on an alt-azimuth mount. Receivers at six frequency bands, L, S/X, C, K (22–24.2 GHz) and Q (30–50 GHz), are equipped, while the new cryogenic 43 GHz receiver was installed in 2018 and is under evaluation. The surface accuracy is 0.4 mm (rms) for main-reflector and 0.1 mm (rms) for sub-reflector. The slewing rates of the main reflector are 1.0° s⁻¹ in azimuth and 0.5° s⁻¹ in elevation. The pointing precision is 10 arcsec (rms). The aperture efficiency of NSRT is 60% at 22 GHz (see Table 1).

Thanks to its unique location in the northwest of China, the east-west baseline coverage of EAVN is significantly enhanced with the participation of NSRT (from 2270 km to 5078 km). This offers a fringe spacing (namely angular resolution θ = λ/D, λ is the observing wavelength and D is the maximum baseline length) down to 0.55 milliarcseconds (mas) at 22 GHz (and 0.26 mas at 43 GHz), which is only 2.3 times smaller than KaVA (see Table 2).

Table 2. Array Specifications. ^a Array symbols for different combinations of stations. ^b Number of antennas. ^c Number of baselines. ^d Baseline length in km. ^e Angular resolution in milliarcsecond. ^f Image sensitivity σ_I is the typical value adopted in the EAVN status report under the assumption of an integration time t = 4 hr and a total bandwidth B = 256 MHz.

Arraya	Stations	${N}_{{\rm{Ant}}}^{b}$	${N}_{{\rm{bl}}}^{c}$	${L}_{{\rm{bl}}}^{d}$ (km)		θe (mas)		${\sigma }_{{\rm{I}}}^{f}$ (μJy beam−1)
				min	max	22 GHz	43 GHz	22 GHz	43 GHz
A1	KaVA	7	21	305	2270	1.24	0.63	155	268
A2	KaVA+TMRT	8	28	305	2270	1.24	0.63	95	165
A3	KaVA+TMRT+NSRT	9	36	305	5078	0.55	–	75	–

Here we selected two representative epochs from the EAVN 2017 campaign as summarized in Table 3. These observations were conducted on March 18th (observing code: a17077a; hereafter, Session-K) and March 27th (observing code: a17086a; hereafter, Session-Q) at 22 and 43 GHz, respectively. Yonsei (KYS) did not join in Session-K due to an issue at the site. TMRT participated in both epochs along with KaVA, while NSRT was only available at 22 GHz. The overall weather condition was good at each site. For TMRT, antenna pointing calibration was made at the beginning of each session. We observed M87 as a primary target of the EAVN campaign while the other sources (3C 273, 1219+044, M84) were observed with less on-source time. In detail, Session-K lasted for 7 hr where scans of 3C 273 (6 min), 1219+044 (4 min), M87 (47 min) and M84 (4 min) were repeated for 6 cycles. Session-Q was performed for 5 hr where scans of 3C 273 (6 min), 1219+044 (2 min), M87 (54 min), and M84 (2 min) were repeated for 4 cycles. RT Vir (H₂O/SiO masers) was inserted with four scans (4 min / each scan) for a system/frequency check of both experiments. However, the SiO masers of the source at Q-band were too weak to be detected during our observations (due to their variable nature, Brand et al. 2020). The recording rate was 1 Gbps (2-bit sampling) where a total bandwidth of 256 MHz was divided into eight 32-MHz intermediate frequency (IF) bands. Only left-hand circular polarization was received. All the data were correlated at the Daejeon hardware correlator installed in the Korea Astronomy and Space Science Institute (KASI).

In Figure 2(a) we display the uv-coverage of M87 for Session-K. The baselines of KAVA+TMRT and KAVA+NSRT are signified with blue and red colors, respectively. The addition of TMRT to KaVA improves the uv-coverage within 180 Mλ, while NSRT significantly extends the southeast-northwest baseline coverage by a factor of 2. Figure 2(b) and (c) features the corresponding beam patterns with a naturally-weighted scheme for KaVA and KaVA+TMRT for Session-Q. Similarly, Figure 2(d)-(f) depicts the beam patterns for Session-K. In Table 4, column MSLL, we list some representative levels of maximum side lobes obtained from the beam patterns for each source, which are a good proxy to foresee the improvement of imaging performance. One can clearly see that the side lobe levels are significantly reduced by adding TMRT(+NSRT).

Zoom In Zoom Out Reset image size

Table 4. Image parameters of Figs. 6 and 7 with natural weighting. ^a Total integration time on each source. ^b Image total flux density. ^c Array symbols are the same as Table 2. ^d Image rms noise level. Core (or off-core) region indicates the region within a 5-mas (or over 40-mas) distance from the core. ^e Image peak intensity. ^f Dynamic range defined by DR = I_peak/I_rms. ^g The maximum side lobe levels are measured within the fringe space derived by the shortest baseline length, namely ±5 mas from the center for Session-K and ±2.5 mas from the center for Session-Q, along PA = +28/–152°.

Epoch	Source	ta	${I}_{{\rm{total}}}^{b}$	Arrayc	Beam size	${I}_{{\rm{rms}}}^{d}$ (mJy beam−1)			${I}_{{\rm{peak}}}^{e}$	DRf		MSLLg
		(min)	(mJy)		(mas×mas, deg)	Theo.	Core	Off-core	(mJy beam−1)	Core	Off-core
a17077a	1219+044	24	607	A1	1.38, 1.06, –10	0.76	0.63	0.65	598	944	917	0.43
(22 GHz)				A2	1.66, 0.96, –8	0.42	0.34	0.37	598	1759	1608	–0.05
				A3	1.51, 0.53, 14	0.37	0.34	0.38	594	1732	1561	–0.03
	3C 273	36	13067	A1	1.45, 1.03, –20	0.97	4.53	4.29	5730	1264	1335	0.41
				A2	1.61, 0.91, –11	0.67	3.11	2.98	5540	1783	1861	–0.03
				A3	1.43, 0.58, 10	0.63	3.37	2.77	5090	1511	1840	–0.01
	M84	24	172	A1	1.29, 0.98, –1	1.13	0.93	1.02	148	159	145	0.39
				A2	1.55, 0.90, –7	0.59	0.46	0.48	149	325	307	–0.03
				A3	1.37, 0.63, 12	0.57	0.45	0.46	146	326	316	–0.03
	M87	282	2009	A1	1.34, 1.08, –21	0.26	0.67	0.48	1326	1980	2754	0.36
				A2	1.56, 0.96, –12	0.16	0.40	0.29	1313	3284	4544	–0.03
				A3	1.35, 0.58, 12	0.15	0.35	0.25	1154	3321	4586	–0.01
a17086a	1219+044	8	692	A1	0.73, 0.56, –22	1.84	1.57	1.50	672	428	448	0.27
(43 GHz)				A2	0.82, 0.52, –17	1.35	1.01	0.95	672	662	707	–0.03
	3C 273	24	8617	A1	0.78, 0.52, –31	1.65	9.04	8.11	3840	425	474	0.38
				A2	0.81, 0.50, –25	1.56	6.73	6.46	3840	571	634	0.13
	M87	216	1577	A1	0.68, 0.57, –23	0.37	0.74	0.46	1180	1596	2565	0.28
				A2	0.74, 0.51, –17	0.31	0.51	0.31	1180	2314	3764	0.01

The EAVN data were calibrated in the standard manner of VLBI data reduction procedures and under the guideline of EAVN data reduction. We employed the National Radio Astronomy Observatory (NRAO) Astronomical Image Processing System (AIPS; Greisen 2003) software package for the initial calibration of visibility amplitude, bandpass and phase. As a large dish, the visibility amplitude of TMRT shows some offsets compared with that of KaVA. The system temperature information of NSRT was absent due to the malfunction of the data acquisition system. Subsequent imaging/CLEAN and self-calibration which were properly performed with the DIFMAP software (Shepherd et al. 1994) can successfully recover the amplitude information from TMRT and NSRT. More details are described in the EAVN memo ² .

In the phase calibration process, solution intervals of 1 min and 30 s (with SNR_ij threshold 5) were used for Session-K and Session-Q, respectively, which are typical coherence time at each frequency. For comparison, in Figure 3 we display baseline SNR_ij obtained by FRING (on three baselines) for the point source 1219+044, for which the correlated flux densities are similar over the whole baselines. The measured SNRs of KTN-TMRT at 22 GHz were ∼4.4/∼3.4 times higher than those of KTN-MIZ/KTN-KUS, which are consistent with our expectation under slightly rainy weather conditions at TMRT. At 43 GHz, on the other hand, the measured SNR on KTN-TMRT was ∼4.2 times higher than that at KTN-MIZ. At 22 GHz, the SNR of KTN-NSRT is comparable with that of intra-KaVA baselines. This is broadly consistent with our expectation given the typical sensitivity of NSRT. For M84, the fringes at 43 GHz were not detected even with TMRT. This would be reasonable given the low core flux (∼100 mJy) and relatively shorter integration time than Session-K. In Figures 4 and 5, we depict the full visibility amplitude of 1219+044, 3C 273, M84 and M87 at 22/43 GHz as a function of uv-distance.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Figures 6 and 7, we show the structure images of each source obtained from Session-K and Session-Q, respectively. For each source, images obtained from KaVA, KaVA+TMRT and KaVA+TMRT+NSRT are displayed separately (with identical contour levels). Overall, the images indicate a significant improvement of image quality when TMRT and NSRT were added to KaVA, and the side lobes seen in KaVA images were greatly reduced for all the sources at both frequencies. The detailed image parameters and resulting image dynamic ranges (DR = I_peak/I_rms, where I_peak is the peak flux density and I_rms is image rms noise level.) are summarized in Table 4. Here we calculate DR for two distinct regions. One is the core region which indicates the area within a 5-mas distance to the core where the deconvolution errors usually dominate. The other one is the off-core region which represents the place over 40-mas away from the core where thermal noises dominate. The corresponding off-core DR is also visualized in Figure 8. For all cases with different arrays and frequencies, the highest DR was obtained for M87. This is simply because the total integration time of M87 was the longest and the corresponding uv-coverage was much better than for the other sources.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

For a proper comparison of relative improvement of image DR among different arrays, a good measure would be the ratio of DR rather than the absolute DR for each image. In Table 5 we list the ratios of DR which are calculated as follows:

$$\begin{eqnarray}&&{R}_{21}=\displaystyle \frac{D{R}_{{\rm{A}}\mathrm{A}}}{D{R}_{{\rm{A}}1}},\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&{R}_{31}=\displaystyle \frac{D{R}_{{\rm{A}}3}}{D{R}_{{\rm{A}}1}},\end{eqnarray} \tag{ 2 }$$

where A1, A2 and A3 indicate the array with KaVA, KaVA+TMRT and KaVA+TMRT+NSRT, respectively. Overall, DR is increased significantly when TMRT is added to KaVA, with actual improvement factors varying from ∼34% to ∼112%. The highest relative improvement was achieved in M84, which is the weakest source in our sample. Then the point-like source 1219+044 comes second followed by M87. The improvement was the lowest in 3C 273. The east-west angular resolution of the images including NSRT is twice better than KaVA+TMRT images. However, the DR does not become worse when the resolution becomes better.

The EAVN images of 1219+044 featured in Figures 6(a)–(c) and 7(a)–(b) are characterized by a quasi-unresolved structure at both frequencies, which is consistent with the flat visibility amplitude distributions in the uv domain (Figs. 4 and 5). For all images from different arrays, we confirmed that the resulting image rms levels were close to thermal limits, as expected for such a simple source structure.

The improvement of image DR for this source is remarkable: by adding TMRT(+NSRT) to KaVA, a factor of ∼1.7–1.9 and ∼1.5–1.6 enhancement was seen at 22 GHz and 43 GHz, respectively. Thanks to the significant improvement of image quality, the KaVA+TMRT(+NSRT) image(s) at 22 GHz allows us to detect a hint of weak extended emission towards the south, which is consistent with the jet emission seen in the literature VLBI images (Lister et al. 2019).

As apparent in Figures 6(d)–(f) and 7(c)–(d), the addition of TMRT to KaVA increased the image dynamic range of 3C 273 by ∼40% at both frequencies. Since the source structure of 3C 273 is complex with a bright core, the resulting image rms noise levels were ∼3 times larger than the thermal noise limit, indicating that EAVN imaging of 3C 273 is highly dynamic range limited.

Another challenge of 3C 273 imaging is that, due to the complicated source structure, the calibrated visibility data could leave larger systematic errors than the case for a simple structure source. Hence, the image DR improvement of KaVA+TMRT+NSRT at 22 GHz was relatively modest. Nevertheless, a factor of 2 enhancement in the east-west resolution allowed us to resolve the fine-scale structure of the core and jet of 3C 273.

The jet in quasar 3C 273 holds a well-known appealing structure which appears to be bent at mas scales (Readhead et al. 1979) while it is highly collimated at arcsecond scales (Bahcall et al. 1995). Previous high-resolution VLBI images at ≤22 GHz consistently indicate a characteristic persistent "kink" morphology at ∼7–10 mas from the core, where the bright edge of the jet flips from one side to the other (Zensus et al. 1988; Bruni et al. 2017; Akiyama et al. 2018; Lister et al. 2019; Zensus et al. 2020). Thanks to the image DR improvement by a factor of ∼1.4, the addition of TMRT (and NSRT) to KaVA is better at tracing the bending structure of 3C 273 in 7–10 mas regions (see Fig. 6(d)–(f)).

The addition of TMRT/NSRT to KaVA allowed us to robustly detect weak signals of this source at 22 GHz. The EAVN images clearly (>10σ) detected a weak jet structure towards the north thanks to the factor of 2 improvement of image DR compared to KaVA. In the KaVA+TMRT+NSRT image, there is a possible hint of slight extension also towards the south, which could be associated with the counter-jet as seen in the literature VLBI images (Giovannini et al. 2001; Ly et al. 2004). A deeper imaging observation with a longer integration time may confirm this structure in the future.

For M87, the resulting dynamic range of EAVN images reached >4500 at 22 GHz (Fig. 6(l)) and >3700 at 43 GHz (Fig. 7(f)), thanks to the longest on-source time among the four sources. As a result, the extended jet emission was clearly detected up to 30 mas (Fig. 9) at 22 GHz, which is important to investigate the collimation profile over long distances. The DR improvement of EAVN images with respect to KaVA is more significant at 22 GHz (∼70%) than at 43 GHz (∼50%). This should be mainly due to the overall higher array sensitivity as well as the longer on-source time at 22 GHz.

Zoom In Zoom Out Reset image size

The significant enhancement of sensitivity and east-west angular resolution by adding TMRT+NSRT enabled us to robustly detect/resolve the "counter-jet" at 22 GHz, which extends up to ∼2 mas at the eastern side of the core (see Fig. 6(l)). The detection of this feature was cross-checked by different data analysis to make sure that its appearance is not an artifact. The counter-jet of M87 was firstly suggested in Ly et al. (2004) and later confirmed in a number of VLBA images at various frequencies (Ly et al. 2007; Kovalev et al. 2007; Hada et al. 2016; Kim et al. 2018; Walker et al. 2018). Imaging of the counter-jet near the core is vital to better understand the properties of jet-launching regions (Hada et al. 2016; Kim et al. 2018; Walker et al. 2018). In particular, accurate measurements of the jet-to-counter-jet brightness ratio (BR) allow us to constrain some key parameters such as the viewing angle and launching velocity, based on the standard theory of Doppler-boosting/deboosting of relativistically moving plasma.

At 43 GHz, thanks to the higher angular resolution transverse to the jet axis, the KaVA+TMRT image clearly resolved the limb-brightening structure that is well known in this jet. Interestingly, the southern limb of the jet appears to be brighter than the northern limb, which might be connected to the asymmetric brightness distribution of the black hole shadow seen in the EHT image (EHTC et al. 2019a).