Improving the radial velocity precision of the HERMES spectrograph to enable exoplanet detection

The evolution of modern astronomical instruments and their capacity to produce useful data has advanced in tandem with, and benefited from, the development of software to optimally analyse those data. In particular, recent developments in the analysis of high resolution spectroscopic data have yielded unique insights on a wide range of astronomical phenomena. As we develop instruments capable of achieving higher spectral resolution and greater wavelength stability, the techniques necessary to produce successful measurements become increasingly challenging. The goal of this thesis is to develop a wavelength solution with an aim to maximise the instrumental performance. For that purpose, we developed a complete reduction pipeline and demonstrate the improvement that it achieves. We show how it produces results that enable new range of scientific goals that were not achievable using the previous methods. We initially investigated the possibility of modelling instrumental wavelength calibration from physical principles, an effort which we applied to the small high-resolution spectrograph RHEA.However we found that a model based only on optical principles was unable to generate a wavelength solution superior to a purely mathematical approach. Taking the lessons learned from RHEA, we then set out to develop a precision radial velocity pipeline for the high-resolution HERMES spectrograph recently commissioned at the 3.9m Anglo-Australian Telescope (AAT). First we characterised the behaviour of the spectrograph PSF across all four channels. Then we undertook a series of observations of stellar targets - some with known radial velocity variability, others as yet unstudied in the time domain-in order to test the radial velocity precision achievable with HERMES. We found that 2dfdr, the standard spectroscopic data reduction package provided by the AAO, was unable to reduce the uncertainty in radial velocity measurements below 400ms−1, due to a combination of PSF effects and the wavelength solution applied by the software. Tests indicated that the extraction method employed by 2dfdr prevented a significant reduction in radial velocity measurement uncertainties; this fundamental limitation led us to develop HARPY, a new radial velocity determination software package that can independently reduce the data produced by HERMES. It calculates a wavelength solution that is stable across observations and performs radial velocity calculations from extracted, calibrated spectra. The final version of HARPY, presented in this thesis,shows radial velocity uncertainties on the order of 70ms−1 with HERMES data. Applying HARPY to our observations, we demonstrate its capabilities on both a binary star and a hot Jupiter system. Proving that HERMES can reach such radial velocity precision, opens a large range of scientific projects to investigation, previously unable to be considered. Its highly multiplexed capability and wide field of view, in combination with its 4 cameras, can be used for large exoplanet surveys, once the lessons learned from this study are adapted to both observing and reduction procedures.