This thesis reports the calibration, imaging and analysis of one epoch of VLBA observations of the 1612 MHz OH maser emission from the protoplanetary nebula OH 0.9+1.3. These are the first polarisation VLBI observations of this source and the spatial morphology of the OH emission is resolved on this scale. Proto-planetary nebulae represent the transition phase in the evolution of stars between the asymptotic giant branch (AGB) phase and their emergence as planetary nebulae. A long-standing astronomical question is how the predominantly spherical circumstellar envelopes of AGB stars evolve into the bipolar and axisymmetric structures that are commonly observed in planetary nebula. Proto-planetary nebulae offer a unique opportunity to study this transformation process. The high-resolution VLBI maps produced in this thesis were used to investigate the morphology and kinematics of OH 0.9+1.3 with a view to gaining insight into the development of asymmetries in the circumstellar material. The OH maser emission of OH 0.9+1.3 has a double-peaked profile with one peak blue-shifted and the other red-shifted with respect to the stellar velocity. The total intensity maser maps demonstrate a considerable degree of asymmetry with the blue- and red-shifted emission located in spatially distinct regions of the envelope. The blue-shifted emission is distributed preferentially along an axis at a projected position angle of » 135± ( North through East). The morphology of this source is not consistent with the standard symmetric thin-shell model and an attempt to fit the traditional OH/IR kinematic model of a simple expanding shell to the maser components was found to be unsatisfactory. No definitive evidence of a bipolar outflow was observed either. The high degree of asymmetry observed in the source is consistent with its status as a proto-planetary nebula. The source was imaged in all four Stokes parameters and the fractional linear and circular polarisations of the maser components were derived from the Stokes parameter maps. In all except one of the components the total fractional polarisation was found to be low ( < 15%). The mean fractional linear and circular polarisation were calculated to be 5.54% and 7.11% respectively. The absence of an identifiable Zeeman pair in the Stokes V map prohibited the estimation of the magnetic field in the circumstellar envelope of this source.

This thesis reports the results of an investigation into techniques for the calibration and imaging of spectral line polarization observations in Very Long Baseline Interferometry (VLBI). A review is given of the instrumental and propagation effects which need to be removed in the course of calibrating such obervations, with particular reference to their polarization dependence. The removal of amplitude and phase errors and the determination of the instrumental feed response is described. The polarization imaging of such data is discussed with particular reference to the case of poorly sampled cross-polarization data. The software implementation of the algorithms within the Astronomical Image Processing System (AlPS) is discussed and the specific case of spectral line polarization reduction for data observed using the MK3 VLBI system is considered in detail. VLBI observations at two separate epochs of the 1612 MHz OH masers towards the source IRC+ 10420 are reduced as part of this work. Spectral line polarization maps of the source structure are presented, including a discussion of source morphology and variability. The source is sigmficantly circularly polarized at VLBI resolution, but does not display appreciable linear polarization. A proper motion study of the circumstellar envelope is presented, which supports an ellipsoidal kinematic model with anisotropic radial outflow. Kinematic modelling of the measured proper motions suggests a distance to the source of ~ 3 kpc. The cirumstellar magnetic field strength in the masing regions is determined as 1-3 mG, assuming Zeeman splitting as the polarization mechanism.

The aim of this thesis is to perform observational tests of SiO maser polarisation and excitation models, using component-level comparisons of multiple SiO maser transitions in the 43 GHz and 86 GHz bands at milliarcsecond resolution. These observations reqwre very long baseline interferometric imaging with very accurate polarimetric calibration. The supergiant star VY CMa was chosen as the object of this study due to its high SiO maser luminosity, many detected SiO maser lines, and intrinsic scientific interest. Two epochs of full-polarisation VLBA observations of VY CMa were performed. The Epoch 2 observations were reduced using several new data reduction methods developed as part of this work, and designed specifically to improve the accuracy of circular polarisation calibration of spectral-line VLBI observations at millimetre wavelengths. The accuracy is estimated to be better than 1% using these methods. The Epoch 2 images show a concentration of v= l and v=2 J= 1-0 SiO masers to the east and northeast of the assumed stellar position. The v=l J=2-1 masers were more evenly distributed around the star, with a notable lack of emission in the northeast. There is appreciable spatial overlap between these three lines. The nature of the overlap is generally consistent with the predictions of hydrodynamical circumstellar SiO maser simulations. Where the v=l J = 1-0 and J =2-1 features overlap, the v=l J = 2-1 emission is usually considerably weaker. This is not predicted by current hydrodynamical models, but can be explained in the context of collisional pumping in a low density environment. Six observational tests of weak-splitting maser polarisation models were performed, including intercomparisons of linear polarisation in the v=l J=1-0 and J=2-1lines, linear polarisation versus saturation level, linear polarisation versus distance from the star, circular polarisation in the v= l J = 1-0 and J=2-1 lines, circular versus linear polarisation and modeling of ~ 900 electric-vector position angle rotations. The polarisation model tests generally do not support non-Zeeman circular polarisation mechanisms. For the linear polarisation tests, the results are more consistent with models that predict similar linear polarisation across transitions. The scientific importance of these tests is described in detail and avenues for future work are described.

This thesis details work undertaken in the field of radio astronomy instrumentation. Specific components of the data collection and processing systems used by radio astronomers have been implemented using non-traditional approaches. Traditionally, the correlation of radio astronomy data has taken place on dedicated, specific hardware. This thesis deals with the implementation of equivalent correlators using software running on generic clusters of personal computers - the software approach to radio astronomy. Toward this end a system has been developed that records the raw telescope output onto computer hard drives, allowing easy access to the data on cluster supercomputers. Part of this thesis describes the design, construction, testing and utilisation of these data recording systems. The correlator software used to process these data on supercomputers is also fully described, including extensive tests of the software and a detailed comparison between its output and the output of an existing hardware correlator. The software correlator is shown to produce output that agrees extremely well with the hardware correlator, verifying its accuracy and performance. Finally, results of on-going scientific investigations that use the software correlators described in this thesis are outlined, illustrating the flexibility and usefulness of the software approach to radio astronomy.

The mildly active nuclear region in the galaxy M81 (henceforth, M81‹) is one of the nearest low-luminosity active galactic nuclei (LLAGN) whose structure is marginally resolved when probed with Very Long Baseline Interferometry (VLBI). Motivated by the way resolved radio sources usually appear on the smallest scales, i.e., a core with a one- sided jet structure, we developed a strictly one-sided, asymmetric triangular model, which we call ASYM, with brightness distribution along a line segment on the sky, with maximum brightness at one end of the segment fading linearly to zero at the other end. The ASYM model is compared and contrasted with an elliptical Gaussian model (hereafter, GAUS), by fitting existing VLBI data of M81‹ at 39 epochs between 1993 and 2003 at 8.4 and 5.0 GHz with the two models. Contrary to what we envisioned, we find that for 77% of our epochs, a simple GAUS model fits the visibility data of M81‹ at 8.4 GHz better (i.e., has a lower reduced 2) than the ASYM model. We conclude that M81‹ is not strictly a one-sided, asymmetric jetted source; as is thought to be the case for the majority of AGN observed at VLBI scales. Our results imply that M81‹ is mostly symmetrical with a significant jet counterpart which cannot be overlooked. / School of Interdisciplinary Research and Graduate Studies (SIRGS) / M. Sc. (Astronomy)

A considerable number of Very Long Baseline Interferometry (VLBI) surveys have been conducted in the northern hemisphere and very few in the southern hemisphere mostly because of a lack of telescopes and therefore adequate baseline coverage. Thus there is a deficit of calibrator sources in the southern hemisphere. Further, some of the most interesting astronomical objects eg. the galactic centre and the nearest galaxies (the small and large Magellanic Clouds) lie in the southern hemisphere and these require high resolution studies. With a major expansion of radio astronomy observing capability on its way in the southern hemisphere (with the two SKA (Square Kilometre Array) precursors, meerKAT (Karoo Array Telescope) and ASKAP (Australian SKA Pathfinder), leading to the SKA itself) it is clear that interferometry and VLBI in the southern hemisphere need a dense network of calibration sources at different resolutions and a range of frequencies. This work seeks to help redress this problem by presenting an analysis of 31 southern sources to help fill the gaps in the southern hemisphere calibrator distribution. We have developed a multi-parameter method of classifying these sources as calibrators. From our sample of 31 sources, we have 2 class A sources (Excellent calibrators), 16 class B sources (Good calibrators), 9 class C sources (Poor calibrators) and 4 class D sources (Unsuitable calibrators).