3.1. Images and their fidelity
Images of 3C 130 at the four frequencies used are shown in Fig. 2.
Some apparent evidence of anomalous spectral behaviour can immediately be seen from these maps. For example, the inner jets of the source appear fainter at 5 GHz than they do at 8.4 GHz, implying an inverted spectrum; this would be very surprising in an extended source region. However, since the uv plane coverage is sparsest at long baselines in the 5-GHz data, these apparent spectral differences may simply be a result of image infidelity. The CLEAN algorithm can be viewed as an attempt to interpolate over the missing spacings in the uv plane, but image fidelity clearly depends on the amount of interpolation needed.
To examine the degree to which image infidelity was a problem in these datasets, I made a test model of the source (consisting of 100,000 CLEAN components from an 8-GHz map with all baselines k). Using the AIPS task UVSUB, I replaced the real data in all four datasets with the Fourier transform of the CLEAN component model; this simulates the effect of observing an identical source with different uv coverage. The four simulated datasets were then mapped with IMAGR . The ratios of fluxes in the resulting maps indicates the fidelity with which a particular component is likely to be reproduced. While the L- and X-band datasets produce very similar maps of the model data, the U-band sampling seems to underestimate the flux of the southern jet, and the C-band sampling gives a noticeably poorer reproduction of the original model, with regions of lower flux particularly in the jets and at the edges of the plumes.
I next made another model, consisting of 100,000 CLEAN components from a similar 1.4-GHz map. The 1.4-GHz data has denser sampling in the centre of the uv plane. Simulating observations of this model with the uv coverages of the four datasets shows that even the X- and U-band datasets do not perfectly reproduce structure on the largest scales; the effect is to produce spurious spectral `steepening' at the very edges of the plumes, increasing the spectral index by 0.2-0.3 at the edges of detectability. The steep-spectrum sheath, however, is too strong an effect to be entirely or mostly due to this undersampling. But these results illustrate the danger of the common assumption that simply matching resolutions or longest and shortest baselines will give maps that are safe to use for spectral index analysis. All spectra at the very edges of the plumes, the U-band spectra of the jets, and the C-band data throughout the source, must therefore be treated with caution.
As an experiment, I tried making a map of spectral index between the 5-GHz data and the 8.4-GHz data remapped with the 5-GHz sampling. Assuming that the original 8.4-GHz images have ideal fidelity, we might expect this resampling to compensate fully for the poor sampling of the 5-GHz data. Although sampling the 8.4-GHz data sparsely does reduce the flux in the jets, the resulting 5-8.4-GHz spectral index is still very flat (0.1) while the spectral indices in the plumes are much more reasonable. This suggests that even identical sampling does not give completely reliable results for snapshot observations.
3.2. Flux densities and spectra
Flux densities of the various components of the source are tabulated in Table 1. Except for the core and hotspot flux densities, which were derived from a fit of Gaussian and zero level using the AIPS task JMFIT, these were measured from polygonal regions defined on the 15-GHz map using MIRIAD . In Fig. 3 the resulting component spectra are plotted.
Table 1. Flux densities of components of 3C 130
The spectra of the core, jets and northern plume are not unexpected. There is some slight evidence for a steeper spectrum in the N jet between 8 and 15 GHz, perhaps indicating a cutoff in the spectrum of the jet at high frequencies, but it is possible that the N jet is affected by undersampling. The S jet is certainly somewhat affected by undersampling at 15 GHz. The N plume shows a slightly concave spectrum, suggesting the presence of multiple spectral components. The hotspot in the N plume shows a flat spectrum with , consistent with a model in which it is produced by particle acceleration at the shock at the end of the N jet; its spectrum has only steepened slightly by 15 GHz.
The immediately striking result of these measurements is the steep spectral index between 8 and 15 GHz in the southern plume. We can be sure that this is not an artefact of poor sampling; the S plume lacks compact structure, and simulations show that we would not expect any flux on the scales of the observed emission to be missing from the 15-GHz maps. It appears that there is a genuine break in the spectrum between 8 and 15 GHz in the southern plume which is not present in the northern plume. As shown in Fig. 4, this effect is not limited to a single region in the plume, but is visible throughout.
The steep-spectrum `sheaths' around the plumes, particularly the southern plume, which were visible in the 8-GHz data presented in Paper I, are missing in the 15-GHz images, although again simulated images show that the 15-GHz observations have sampling which should be adequate to reproduce them. This implies that the sheath regions have very steep spectra between 8.4 and 15 GHz. Using regions defined with MIRIAD on the 1.4-8.4-GHz spectral index map, in which the sheath region is well defined, I find that for the sheaths around both north and south plumes, whereas .
© European Southern Observatory (ESO) 1999
Online publication: September 2, 1999