Entrevista con Plinio Vento.

The results from source extraction were compared with those reported in Bower et al. (2007) and are summarised in Table 5.1 (the LTraP results have been corrected for primary beam attenuation, in concordance with the Bower results). When ex- amining the results there are two aspects to consider. Firstly, did the ParselTongue pipeline produce the correctly calibrated images? Secondly, did the LTraP correctly fit the transient sources? In the initial testing, the LTraP struggled to pick out the published transients in the field. After further refinement it was found that the RMS background map size had been set too small - approximately 30_{× 30 pixels. The} RMS background map size grids the input image into boxes, for which the RMS is measured. If the grid size is set too small, in this case 30_{× 30 pixels, and the source} size is around five pixels in diameter, then the noise in that region is large (when compared with that away from point sources) and the transient does not exceed a 5σ detection threshold. A number of the transient sources reported by Bower et al.

Figure 5.2: Top panel: Averaged image of 40 epochs (at 4.8 GHz) produced us- ing the automated VLA reduction pipeline. The two squares indicate the locations of the transients RT 1986-01-15 (Northern square) and RT 1984-05-02 (Southern square), which are discussed further in this chapter. Bottom panel: The concate- nated deep image of all 4.8 GHz data presented in Bower et al. (2007). On both plots the small circle represents the half-power radius, the larger circle represents the twice the half-power radius.

Table 5.1: Summary of the Guassian fits produced from the LTraP compared with those reported in Bower et al. (2007).

Transient epoch VLA Configuration Bower flux (µJy) LTraP flux (µJy)

RT 1984-05-02 C 448_±74 445_{± 82}

RT 1984-06-13 C 566_±81 480_±61

RT 1986-01-15 D 370_±67 370_±79

RT 1986-01-22 D 1586_±248 Image fidelity bad

RT 1992-08-26 D 642_±101 510_±85

RT 1997-05-28 CnB 1731_±232 1744_±394

RT 1999-05-04 D 7042_±963 See discussion

(2007) were only just above 5σ. Bower et al. (2007) reported that they had not used a RMS background map in their search, therefore I then set the RMS map size to the entire image and a number of the transients were then correctly identified.

In Table 5.1, five out of the seven transients are fitted with reasonable agreement when compared with Bower et al. (2007). In Figure 5.3 I show an image within the region of the transient source RT 1984-05-02 produced by the automated pipeline; on the bottom panel I show the lightcurve constructed using the LTraP for all images in C configuration. In Figure 5.4 I give a further example of RT 1986-01-15, the lightcurve is constructed using the LTraP with all images from D-configuration. Both the sources are in good agreement with Bower et al. (2007) and the image fidelity is good.

The observation containing RT 1986-01-22 did not produce a good quality im- age with the automated pipeline. The data were rereduced by hand, but the image quality was still too poor to detect the source. The data were thoroughly flagged and the calibration and imager settings were consistent with those used in previous pipeline runs. It should be noted that at the time this work was conducted, it was the author’s first experience with VLA data. The settings used to make the Bower im- ages were not described in Bower et al. (2007), so potentially a different parameter in the reduction procedure could yield a detection.

The image of the observation containing RT 1999-05-04 did produce a reason- able image, however, the transient was only marginally detected. The source was searched for in the primary beam corrected image of the field, however, it was be- yond the correctable beam radius. A Gaussian was fitted in the uncorrected image yielding a flux of 242_±61 µJy: which is a 4σ detection (see Figure 5.5). As the source was beyond the correctable beam radius, by using the AIPS task PBCOR, I can calculate, by hand, the primary beam correction needed for the VLA using eq.

0 10 20 30 40 50 60 70 80 −1 0 1 2 3 4 5 6 7x 10 −4 _{RT 1984−05−02} Epoch Number (N) Flux (Jy)

Figure 5.3: Top panel: Image of the transient detection RT 1984-05-02 produced by the automated VLA reduction pipeline (not primary beam corrected). Contours are at 3, 4, 5, 6 and 10_{×RMS. Bottom panel: Lightcurve of transient produced} using the LTraP source extraction algorithms.

0 10 20 30 40 50 60 70 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5x 10 −4 _{RT 1986−01−15} Epoch Number (N) Flux (Jy)

Figure 5.4: Top panel: Image of the transient detection RT 1986-01-15 produced by the automated VLA reduction pipeline (not primary beam corrected). Contours are at 3, 4, 5, 6 and 10×RMS. Bottom panel: Lightcurve of transient RT produced using the LTraP source extraction algorithms.

Figure 5.5: Image of the field containing RT 1999-05-04 made using the auto- mated VLA pipeline. Contours are at 3 and 4σ. This image has not been primary beam corrected.

5.1 and eq. 5.2 below (see footnote1for PBCOR reference).

fPBCOR≃ 1 + X.C1 103 + X2.C2 107 + X3.C3 1010 (5.1) X=  ∆ r (arcmins). ν (GHz) 2 (5.2) where X is the distance from the source to phase centre (in arc-minutes),νis the fre- quency in GHz, the constants are defined as C1=-1.372, C2=6.940 and C3=-1.309 at C-band. The corrected flux can then be calculated using SPBCOR= Sv/ fPBCOR. Ap- plying this correction to the flux measured in the uncorrected image yields SPBCOR= −23 mJy. Applying the correction at the maximum correctable beam radius (mea- sured in the VLA primary beam corrected images), I find SPBCOR=32 mJy which uses a correction factor fPBCOR=0.0074. Applying this correction factor to the flux quoted in Bower et al. (2007) i.e. SPBCOR=7042µJy, I find the transient flux in the Bower uncorrected image would be Sv=52.1µJy - assuming it is at the correctable beam radius. A detection of Sv=52.1 µJy would be close to, or below the noise, therefore making it undetectable (typical RMS values are _∼50 µJy). In Bower et al. (2007) they do comment that wide field images were made of all the fields, pos- sibly using a mosaicing mode. Potentially the source was detected in the wide field image and thus the analysis differs from mine. If not, the Bower et al. (2007) RT

1999-05-04 result is incorrect.