VLASS Epoch 1 Quick Look Users Guide
VLASS is a 2-4 GHz radio interferometric survey covering Declinations greater than -40deg and conducted by the VLA in an on-the-fly survey mode. Each observation of a point in the sky is observed for a few seconds and 40 square degrees is covered in a single observing block. This observing mode allows the VLA to efficiently survey a large area, but it does complicate the algorithms used for imaging and analysis. The Quick Look (QL) images use a relatively simple imaging algorithm that results in limits on the accuracy of the images.
The following applies to QL images from Epoch 1 of VLASS, and is subject to modification as the images are examined further. Users are referred to VLASS Memo #13 (obtainable from the VLASS Memo collection of the NRAO Library) for more technical details. We expect these limitations will be significantly mitigated when more accurate Single Epoch images are produced.
Positional Accuracy: For Dec > −20, the positional uncertainty is ≈ 0.5”, increasing to ≈ 1” at Dec −40◦ . This is primarily due to uncorrected wide-field imaging effects. There may also be residual effects due to phase errors introduced by the troposphere and ionosphere.
Flux Density Accuracy: The flux density accuracy for the first campaign of the first epoch (VLASS1.1) and the second campaign (VLASS1.2) differ, with VLASS1.1 typically worse (most likely due to a problem with the antenna positions that was fixed for 1.2). For objects in VLASS1.1 with flux densities below ≈ 1Jy, the peak flux densities are systematically low by ≈ 15%, and the total flux densities by ≈ 10%, with a systematic scatter of ≈ ±8%. Above 1 Jy the flux densities can be very unreliable and should not be used. In VLASS1.2, the corresponding offsets in flux density are about 8% low for the peak flux densities and 3% low for the total flux densities. Occasionally, for any sources that happen to be located near a region that was affected by uncorrected Ghost phenomena, (see below) flux densities could be low by as much as 25%.
Sizes of Compact Sources: The issues that affect the positions, as noted above, will also cause a small increase in the apparent size of compact sources. We estimate this based on the limited tests described above to be a the level of 5–15% of the PSF beam area. Further investigation is underway.
Sensitivity to Extended Emission: The shortest baselines covered by S-band B-array are approximately 58" (see the Largest Angular Scale (LAS) in Table 3.1.1 in the Observational Status Summary), but it is unknown how much flux density is recovered on those scales, For complex sources, the recovery of extended emission will vary with morphology. Tiles in the Southern sky with Dec < −14◦ were observed in the BnA hybrid configuration with the North arm of the VLA extended out a factor of 3 longer, and thus for these (in particular those at higher Declinations) the LAS would be more similar to that in A-configuration (18"). In addition, the QL images were made by combining data over the full 2–4 GHz band during imaging, implying a factor of two in scale over the band with an uncertain impact on the LAS in these wideband images. We have no quantitative assessment at this time on the sensitivity and accuracy of the VLASS QL images to extended structure, and we caution users to keep this in mind when using these images.
Ghost Artifacts: There were errors during observing that caused faint “ghost” copies of sources to appear at multiples of 1 or more field offsets (178.74" in most Tiers) in RA. Although we have identified and removed cases of this where our source finding could identify these in the mosaic images (requiring 5-sigma detection of the fainter ghost image), there are undoubtedly further cases that fell below our detection threshold. Even if the ghosts are to faint to see in an image, they will rob the true source of flux density, biasing the flux measured flux density low. These ghost remnants would be from observations made in the Pilot, VLASS1.1 before 2017-Nov-15, and in VLASS1.2 between 2019-Mar-13 and May-13. VLASS QL users are recommended to check components of interest during these times for the effects of faint ghosts.
Image artifacts
The algorithms used to generate the Quick Look images were designed run quickly, in particular to allow searches for transient radio sources. This means that, in addition to the minor issues described above, some images are affected by systematic failures in the clean algorithm and/or poor phase calibration.
Bright sources: there is a systematic variation in the PSF across a VLASS mosaic that is not allowed for in the clean. Thus, the presence of bright (> 100 mJy) sources with high S:N sidelobes can sometimes cause the clean algorithm to diverge. For the most part these fields are found during QA and typically they are placed in the QA rejected category if the ratio of the residual from the clean to the RMS over the remainder of the image exceeds 5:1, but in general users should exercise caution when extracting detections of faint objects in the vicinity of bright ones, as false detections and/or distorted morphologies are likely to be found.
Phase errors: no self-calibration is applied to the Quick Look images. Thus, their dynamic range is very dependent on observing conditions. These can lead to a false impression that an object has a particular morphology or size. As the phase errors will persist over a large fraction of a tile, the keys to identifying these is that all point-like sources share a morphology, and the distortions around the point sources are asymmetric in nature. Quantitatively, the near-field dynamic range 8-arcsec (~3 beams) from a point source (defined as the ratio of the peak brightness in the center to the magnitude of the largest artifact on an 8" radius circle) ranges from about 10:1 to 100:1 in the Quick Look images. At the lower end of this range, particularly when using a logarithmic or histogram stretch in a viewer it is easy to mistake such artifacts for real morphological features. An example of such an unusually poor image is shown in Figure 1.
Figure 1: a point source in a tile with bad image quality. Left: a linear stretch from the image minimum to maximum, center: a logarithmic stretch from minimum to maximum, and right a +/-5-sigma stretch, where the rms, sigma, is measured in the image far from bright sources. Note the characteristic three-fold pattern, asymmetric about the source center. The magenta circle shows the 8" radius used for calculating the near-field dynamic range. In this example, the value of the near-field dynamic range is 13.
Users are encouraged to submit questions and comments to the VLASS Department of the NRAO Helpdesk to improve the usability of this document.
