Reduction of a VLA uv-dataset

Download (0)

Full text


Reduction of a VLA uv-dataset

Author Frits Sweijen s2364883

Supervisor J. McKean

Version February 2, 2015


In this report we summarize the process of calibrating a uv-dataset taken with the VLA. We focused on one particular source in the dataset. The goal was to then image this source and find out about general properties. First we show the process of calibrating the dataset for the gain and flux-density, followed by the imaging procedure. Finally we shortly discuss the results for the source 4C-01.37.

I Introduction

Radio galaxies have the property that besides emit- ting in the optical and other wavelengths like X-ray they emit radiation in the regime of radio waves.

This radiation usually comes from lobes being pro- duced by jets coming out of the center of the galaxy.

At the center of the galaxy is a super massive black hole. When this black hole starts accreting matter, it can start producing giant jets blowing out parti- cles including electrons. We call these galactic cores active galactic nuclei (AGN). At the end of the jets when the particles are slowed down due to interac- tion with the ISM or IGM, the particles pile up and lobes form. In these regions the magnetic fields are typically of the order of 10−5 Gauss. This causes the electrons to emit synchrotron radiation.[3] Key characteristics of radio galaxies are these luminous radio lobes sometimes extending way beyond the galaxy itself and, depending on the orientation of the source, the jets.

In the reduction of this dataset we focused on one particular source, 4C-01.37. This source is a radio galaxy located at RA 15h26m30.779s and DEC -01d54m09.70s (J2000). It has a redshift of z = 0.93.[1]

II Observations

The VLA is an interferometer with 27 telescopes in a Y shaped configuration. Each telescope is a dish 25 m in diameter and can detect frequencies in the range between 74 MHz and 50 GHz.

The observations were made on the night from October 9 to 10, 1995 with a total integration time of 7030 s, of which 130 s was spent on the source 4C-01.37. The sources were observed with two fre- quency channels with a bandwidth of 50 MHz one at 4.84 GHz and one at 4.89 GHz. This corresponds

to a wavelength of 6 cm. In this frequency range we use the C band detector.

Looking at Fig. 3 we see that the longest base- line is around 10 km. Hence for this observation the VLA was used in the B configuration.[2]

To get close to observing the actual surface brightness distribution we need a good coverage of the uv plane. uv coverage depends on the number of telescopes used, the number of frequency chan- nels and on the total observing time. In Fig. 1 we see the uv coverage of the VLA for all observed sources.

Figure 1: uv plane coverage. The use of two spec- tral channels gives increased radial sampling which can be seen as denser areas.

III Data Reduction

I Calibration

Before processing the data into an image it first needs to calibrated, as the raw data is on an arbi- trary scale. By calibrating the data we can trans- form this into a physical scale like Jansky. To be able to calibrate, several calibrator sources were ob- served. Among these calibrator sources there is one primary calibrator, the flux-density calibrator, and


four secondary or phase calibrators. These can be seen in Fig. 4. The primary calibrator is used to set an amplitude scale. The secondary calibrators are used to correct for atmospheric changes as the source moves through the sky.

The first step in calibration is to set the right flux scale. The signal we detect cannot be inter- preted as actual flux densities. To set the scale for this we use a flux density calibrator. For this calibrator we have a model of what the observed skybrightness would look like. Next we take the Fourier Transform of this model and convolve it with the sampling function S to obtain a model for the observed visibility.

Vobs,model= S(u, v) ∗ Vtrue,model (1) From this we can then determine by how much we need to scale the true observed visibility Vobsto get to the model, Vobs,model.

With the right flux scale, the next step is gain calibration. We need to find solutions for the an- tenna gain relating the observed and true visibili- ties.

Vobs= gi(t)gj(t)Vobs,modelei(θi(t)−θj(t)) (2) We take one antenna as reference and then calcu- late the antenna gains for the other antennas with respect to this reference antenna. For this calibra- tion the reference antenna was VA05.

These calibrations so far have only been calcu- lated for the flux density calibrator. This is be- cause this is a well known source for which a model is available. We can now find the true flux den- sities of the secondary calibrators by scaling these amplitude gains.

II Imaging

Now that the data is properly calibrated, we can produce an image. The simplest image is made by directly convolving the point spread function (PSF) with the observed sky brightness. The PSF has many side lobes besides the main beam causing distortions of the image and is therefore called the dirty beam. This produces a dirty image as seen in Fig. 5. To get rid of the distortions we clean the image.

Cleaning is done through a process called self calibration. First we mark the positions where we believe the source is and then we subtract a small fraction of the image from itself. Then we again se- lect what we think is part of the source and subtract more. This process is repeated iteratively until we are down to only noise. At the locations where we

indicated to be a source we now put a delta func- tion. This delta function is then convolved with the clean beam. The clean beam is the main beam of the PSF, ie. the PSF without the side lobes. For a model of the clean beam we assume a Gaussian that is the same as the PSF without side lobes.

This convolution now gives a new model for the source with lower noise levels. This process is then repeated multiple times to further bring down the noise.

From our observations we can determine an ex- pected rms noise using

σrms= SEF D

ηcorrpN (N − 1)∆ντ (3) where N is the number of antennas and ηcorr the correlator efficiency. Using the values N = 27, ηcorr = 0.89, ∆ν = 100 MHz (two channels of 50 MHz) and a system equivalent flux density of 310 Jy[2], we find an expected rms noise level of 1.11 · 10−4 Jy. The rms as measured from the map is 2.06 · 10−4 Jy. This measured noise is higher than the expected value. This can come from er- rors during calibration or simply because we have not cleaned deep enough. Another reason could be because we took the value of the SEF D from data relevant to the VLA as of 2014. It may very well be that the SEF D was different when this observation was made in 1995.

IV Results

After cleaning we can again make a new image. In Fig. 2 we see the dirty image and the clean image side by side. We immediately see that the noise has been reduced by a substantial amount. We can now resolve the structure of the source.

The two most prominent features are the two bright lobes in the lower left and upper right cor- ner. If we look a bit closer we can faintly see a small structure in the center, which is most likely the galaxy, and two jets coming out. If we see the galaxy edge-on the jets would stream out perpen- dicular to our line of sight and we would not see them. The fact that we see them here can mean that the galaxy is somewhat inclined towards us.

This causes the electrons in the jets to be moving slightly towards us, causing relativistic beaming in our direction. This increases the brightness of the jets.

At the end of the jets we see two lobes. In the lobes there are hotspots, regions of high intensity.

The lobes are clearly visible, while the jets are not.

Therefore we classify this source as an FR II radio source.[3]


Figure 2: Comparison between the dirty image and the clean image. Left: the dirty image. The bright lines across the view come from the sidelobes. Right: the clean image after 9 cleaning cycles. The scaling power cycles for both images has been set to -2.

From the created map we can infer a flux density of this source of 220 mJy. We can then calculate the luminosity with

Lν = 4πDL2 Sν (4) where DLis the luminosity distance and S the flux density. 4C-01.37 has a luminosity distance of 5.275 Gpc.[1] This yields a specific luminosity of 6.63·1026 W m−2 Hz−1 or 6.63 · 1033 erg cm−2 Hz−1.

From the lower left lobe to the upper right one the source measures 14.15 arcsec. The scale at this redshift is 7.467 kpc/arcsec. This gives a total size of 105 kpc. The lower left lobe has a diameter of 23.7 kpc and the upper right one a diameter of 22.4 kpc.

V Conclusion

Table 1 shows a summary of the properties found of the radio galaxy 4C-01.37.

Properties of 4C-01.37

Distance (Gpc) 5.275

Luminosity (W m−2 Hz−1) 6.63 · 1026 Luminosity (erg cm−2 Hz−1) 6.63 · 1033

Rms Noise (Jy) 2.06 10−4

Size (kpc) 105

Flux Density (mJy) 220

Table 1: A summary of the properties of 4C-01.37.

The limiting factor is the overall efficiency of the telescope along with the angular resolution. Im- provements could be made by spending more time

on source or by increasing the uv coverage. This can be increased by observing longer or by adding more spectral channels increasing the radial sam- pling. Increasing the angular resolution by chang- ing to the A configuration with a baseline of 36 km may allow for resolving more details of the source like the jets themselves.

VI Discussion

The radio galaxy 4C-01.37 has a size of 105 kpc.

Table 2 lists the sizes of three other sources in the same observation.

Source Size (kpc) 4C+01.42 95

4C-02.64 344 4C+03.38 221

Table 2: A few other sources in the same field with their sizes.

Comparing 4C-01.37 with the other sources in- dicates that this is one of the smaller sources in the observed field.

The main emission mechanism in this source is synchrotron radiation. The electrons in the lobes get accelerated by the magnetic field of the galaxy and therefore emit radiation. This corresponds to a spectral index of roughly α ≈ −0.8[3].

These results make this particular source is interesting for two reasons. First it is a rela- tively compact source compared to others in the same field. This may indicate a relatively dense region in the intergalactic medium, causing the


jets to slow down and form lobes early. Second the luminosity is lower than what would be ex- pected. Typical radio galaxies have luminosities between 1040 erg for starburst galaxies to 1045 erg cm−2 Hz−1 for e.g. quasars.[3]. This ei- ther means that this galaxy is a very weak radio source or an effect of looking at just one frequency.


[1] Nasa extragalactic database. https://ned.

[2] Performance of the vla during the current semsester, vla capabilities february 2014 - september 2014.



[3] Bernard F. Burke; Francis Graham-Smith. An Introduction to Radio Astronomy. Cambridge, 2014.



Figure 3: The distance in the uv-plane is an indicator for the baseline length of the telescope.

Figure 4: Amplitude vs time for all sources. Each color is a different baseline. Top left: the flux-density calibrator. Middle: phase calibrators. Bottom: target sources.


Figure 5: Dirty image of 4C-01.37 clearly showing the distortions from the dirty beam.




Related subjects :