Astron. Astrophys. 337, 69-79 (1998)

3. Observations and imaging

3.1. Observations

While at most frequencies 0710+439 has only been observed once, at 5 GHz it has been observed with a global VLBI array at 5 epochs spread fairly evenly over a period of 13 years (see Table 1). The first three epochs were analysed by Conway et al. (1992). Here we reanalyse these first three epochs and add new data from two additional epochs; a global 10 station long track observation made on 25th September 1989 and a multi-snapshot 13 station global observation made on 11th June 1993.

Table 1. Summary of 5 GHz global VLBI observations of 0710+439

3.2. Data reduction

Observations in all epochs were made in left circular polarisation (IEEE convention) and a bandwidth of 2 MHz was recorded using the MkII recording system (Clark 1973). The data were cross-correlated with the JPL-Caltech VLBI Processor. The data were fringe-fitted in AIPS (Schwab & Cotton 1983) and averaged to 1 minute; error bars for the averaged data were estimated from the internal scatter of the data over the averaging interval. Amplitude calibration for each antenna was derived from measurements of the antenna gain and system temperature during the observations.

After amplitude calibration the data were edited and mapped using IMAGR (AIPS package 1995) and Difmap (Shepherd et al. 1995). Many iterations of phase self-calibration were performed before applying amplitude self-calibration at the end. Windows for clean components were added to provide support and reject sidelobes. Initially each epoch was mapped separately starting with a point source model. The fitted restoring beams at each epoch were typically 1 mas in the East-West direction and 1.5 mas in the North-South direction. The highest dynamic range image was obtained from the 5th epoch data set (see Fig. 1 a). Given this best map we therefore remade maps at all epochs using it as the starting model for self-calibration. As described in a Sect. 4 we then used these images to detect or set limits on internal motions within 0710+439.

Fig. 1. a Fifth epoch map of 0710+439. b Diagram with positions and sizes of the Gaussian model fit components

Modelfitting with gaussian components was also carried out to the visibility data at each epoch using the program MODELFIT in the Caltech VLBI package (Pearson 1991), which fits to the amplitudes and closure phases directly, and also with the gaussian modelfitting option within Difmap. The latter fits amplitudes and phases but allows phase self-calibration against the model, so that the model can converge to fit the closure phases. In all cases it was possible to obtain good fits to the data using only a few Gaussian components. The modelfitting process was started using gaussian components fitted to the 5th epoch CLEAN map using the AIPS task JMFIT. After varying all the gaussian parameters the best fitting model to the fifth epoch data contained 9 components (see Fig. 1b, Table 2). This model provided a good fit with reduced Chi-squared agreement factor for amplitude Q_AMP=1.258 and for closure-phases Q_CLP=1.180 (for definition of agreement factor see Henstock et al. 1995). To characterise temporal changes in the source we obtained fits to the data at each epoch, using this 5th epoch model as a starting point (see Sect. 4.1).

Table 2. Model for the 5th epoch of 0710+439

3.3. Source structure

The CLEAN map of the 5th epoch data (rms noise = 0.9 mJy beam^-1) shows clearly the overall triple structure of the source (see Fig. 1a). Maps at each epoch show three main components which we name (from North to South) A, B and C. Each of these main components shows substructure which is represented in the modelfits as separate gaussians (e.g. A1, A2 and A3; see Fig. 1b).

The CLEAN maps show that the northern (A) and the southern (C) components show some faint extended emission around them. In addition to this both modelfitting and imaging indicate a compact feature (A2) within component A which we interpret (see Sect. 5.1) as a hotspot. The CLEAN images suggest a weak bridge of emission between A and the middle component B, however imaging simulations show that this feature may not be reliable (see Appendix). We were not able to detect a similar jet-like feature connecting the middle and southern components, which was found on a 1.6 GHz map (Xu 1994), possibly due to lack of surface brightness sensitivity. We were also not able to detect any emission located to the East of the C component, which is seen on the maps made by Wilkinson et al. (1994).

Fitting the B component required 4 gaussian subcomponents. Fig. 2a,b show these gaussians as fitted to the 1st and 5th epochs respectively (convolved with a circular restoring beam of FWHM 0.7 mas). These images show that the B component is narrow at the South and becomes wider to the North. There also appears to be a slight kink in the jet with the major axis of the B3 component inclined at with respect to the B2-B4 direction. A similar kink is seen in the 15 GHz maps and models (Taylor et al. 1996). Despite being separated by almost 13 years the model components for all 5 epochs are very similar. One possible change is in the size of B1, however we conclude based on our imaging simulations that this may be an imaging artifact. In Sect. 4 we describe our detailed analysis of the multi-epoch data which shows components B2 and B4 to be stationary relative to each other, with a possible small northward motion of B3.

Fig. 2a and b. Modelfit gaussians within middle component of 0710+439 at 5 GHz convolved with a 0.7 mas restoring beam: a First epoch, b Fifth epoch

Overall our maps and modelfits to B agree closely with those estimated at 15 GHz by Taylor et al. (1996). However in our modelfits we did not require a compact component at the position of the core seen at 15 GHz. This is not unexpected given that at high frequency this component has a self-absorbed spectral index of = (Taylor et al. 1996), where flux density S . At 4.9 GHz with this spectral index we would expect the core to have a flux density of between 4.8 mJy and 11.8 mJy. At the position of the 15 GHz core on a 5th epoch super-resolved map (0.5 mas FWHM, not shown) we did see a component with flux density mJy, if this feature is real it implies a spectral index of between 4.9 GHz and 15 GHz consistent with a synchrotron self-absorbed spectrum core component.

© European Southern Observatory (ESO) 1998

Online publication: August 6, 1998
helpdesk.link@springer.de