In this chapter I have presented the data reduction procedure to create LOFAR im- ages. I have also presented the first algorithmic transient search on wide field LO- FAR images. There are still many issues to resolve in both these procedures. For the calibration and imaging algorithms absolute flux calibration must be stable, and wide field next generation calibration strategies such as direction dependent gains must be applied. The beam properties also need to be better characterised to pro- duce good Gaussian fits. Other issues such as band-pass calibration and primary attenuation are still in a state of commissioning and impact on the quality of the final images.
I have shown in this chapter this it is possible to run the LTraP algorithm over a sequence of LOFAR images, which produces a large number of transient candidates and statistics. The main bottle-neck is the current quality of the images, a large ma- jority of sources within the field show a large index of variability, and this impedes detecting truly variable sources. However, by comparing the images by eye with archival images, some interesting candidates have been identified. The LTraP will soon be capable of automatically detecting (and responding) to the type of attributes that these candidates have i.e. non-LOFAR and archival catalogue detections. The type of logic, testing and re-reduction of images conducted on the LOFAR commis- sioning images may have to become more algorithmic. For example, if a particular setting in the calibration or imaging stages is prone to producing artifacts (in certain situations), the SIP and LTraP must be capable of altering these setting to produce comparison images, to automatically assess the robustness of the result. Simply producing one image and assuming it is always accurate may result in lots of false detections. The flexibility of the LOFAR data reduction system is unique in this sense; all algorithms within the pipeline operate under the same architecture. If the source extraction algorithm detects an intersting source, it can spawn new imaging tasks to produce subsequent images with different settings, or re-task the calibration and flagging etc.
The B0329+54 field will form one of the zenith monitoring fields, therefore de- veloping accurate sky models, and data reduction techniques is essential to com- mission the radio sky monitor. In future observations we can expand the B0329+54 field of view, using multiple LOFAR beams, and using the detailed sky model for this field (plus further sky model sources), we can commission the radio sky mon- itor. We can also use these early datasets and analysis as a benchmark to assess the long term (> year) variability and transient nature of the field. In this chapter I
have only focussed on producing one combined image from 12 hours of data; it is a goal to produce images on much shorter timescales (i.e. one second). Dividing the datasets described in this chapter into smaller chunks and re-imaging could yield further intersting results, because we would be more sensitive to transients with that characteristic timescale. A single 12 hour dataset would produce some 43200 one second images, which in turn would place stringent limits on the rates of short durations transients: if none were detected.
So far ILT J0320.3+5512 remains the most promising candidate and may well turn out to be the first transient detected with LOFAR, however, it must stand up to rigorous testing with different algorithms and data reduction procedures. Potentially many more observations of the B0329+54 field will be obtained, and assuming that each image holds an equal chance of detection, it should hopefully not be long before we have the first confirmed LOFAR transient to report. In the final chapter to this thesis I will comment further on the parameter space we have probed with this survey, and I will relate this to the potential parameter space other SKA pathfinder instruments might explore.
THOMASJEFFERSON(1743 - 1826)
7
Conclusion
In this chapter I will present the main findings of this thesis. I will also discuss future work and broader issues pertaining to each chapter, that were not addressed in the individual chapter conclusions. I will place the broader implications of the transient searches presented in this thesis into context, with an emphasis on the future parameter space SKA pathfinder instruments may explore.
7.1
Summary of thesis and future work
In Chapter 3 of this thesis I presented a detailed variability study of the low lumi- nosity active galactic nuclei NGC 7213. Through this analysis I explored the radio and X-ray correlation using the cross correlation function. I showed that a weak but statistically significant delay existed between the X-ray and radio emitting re- gions. I related this behaviour to that seen in other active galactic nuclei and black hole X-ray binaries. I also used the X-ray and radio luminosities, together with the black hole mass, to show that NGC 7213 was in good agreement with the predicted correlation found on the fundamental plane of black hole activity. Around three years of X-ray and radio data was collected to study NGC 7213 - this is one of the longest campaigns of its type to date. Requesting radio observations over so
many semesters is difficult and time consuming. X-ray data were much more easily obtained with the Rossi X-ray Timing Explorer (RXTE), due to the observatory’s flexible scheduling, and willingness to collect long term datasets. Potentially with ‘all sky monitor’ type properties of the next generation of radio telescope, obtaining large datasets of this type of source (and others) will become easier. It is a goal for these wide field facilities to frequently monitor known (bright) variable sources.
In Chapter 4 of this thesis I presented an overview of the prototype LOFAR transients detection pipeline (LTraP). I demonstrated and discussed the main com- ponents that were used to analyse the datasets presented in Chapters 5 and 6. I discussed the broader functionality of the pipeline that will be needed for future operations, and I also presented pitfalls with current approaches to tackling tran- sient searches. Future work which is needed (and is currently underway) to com- mission the full pipeline. This includes: development of quality control metrics, automated classification algorithms, an automated (internal and external) response system, more rigorous transient database queries, and an efficient visualisation sys- tem to view results.
In Chapter 5 of this thesis I presented an automated VLA data reduction pipeline that was used to flag, calibrate and image VLA data. This imaging pipeline was used to produce images for three different datasets. Firstly the datesets containing the transient detections reported in Bower et al. (2007) were imaged. The properties of the transient sources, and the light curves, derived from the LTraP were compared with those reported in Bower et al. (2007). Five out of the seven transient results were in good agreement with those reported in Bower et al. (2007) and this work was an excellent test of the LTraP. Secondly an archival VLA dataset containing the previously unreported outburst of the X-ray binary Swi f t J1753.3-0127, was im- aged and processed through the LTraP. The results were used to explore the X-ray to radio correlation; it was found that the source was less radio luminous then pre- dicted by the relationship derived from the fundamental plane of black hole activity. Finally a blind survey for radio transients was conducted using both the automated VLA reduction script and the LTraP, on archival calibrator fields. This work placed upper limits on the snapshot rate of radio transients at GHz frequencies; it also ex- plored, and made predictions regarding the Log N- Log S of transient radio sources. In conjunction with producing early stage science, this work contributed to com- missioning the LOFAR transient detection system. The VLA achieve potentially contains many hours unsearched data, which could contain radio transients. The imaging algorithms developed for this study could be used to further interrogate the archive in the future.
In Chapter 6 of this thesis I presented the procedure to create images from raw LOFAR data, and also a blind transient search using the LTraP. Many stages of the image production procedure are in a state of commissioning, however, it has been shown that reasonable image fidelity can be achieved, and promising results are ex- pected in the future. In Chapter 6 I only concentrated on producing images with an integration time of 12 (or 6) hour timescales; future work will include reducing the integration time to one second, and also developing and implementing 3rd gener- ation calibration techniques that are specifically tuned for transient studies. There are two key issues which have been addressed in Chapter 6. Firstly, what are the ideal parameters to produce one adequate LOFAR image? Secondly, what are the ideal parameters (and compromises that should be made) to produce a sequence of LOFAR images? So far efforts by other LOFAR imaging commissioners have mainly focused on producing one image, whereas producing multiple images (on short timescales) is clearly a top priority for the Transients Key Project. The tran- sient search using LTraP reflects the current quality of LOFAR data. Until adequate flux calibration is achieved, analysis of the variability of sources within this field is restricted. In this chapter I have however demonstrated that the infrastructure and algorithms are available to do this. Although the flux calibration is unstable, this survey is still sensitive to unique sources not observed in previous images or radio catalogues. I have presented two LOFAR transient candidates, which meet this cri- teria. In the next section, as a conclusion to this thesis, I will explore the parameter space we have probed with this prototype LOFAR survey, and what can potentially be achieved in the future. I will also present the parameter space that other SKA pathfinder facilities may explore.