The distant, dusty universe: finding star-forming galaxies with the Herschel Space Observatory

The formation and growth of galaxies through star formation is an unsolved problem in astrophysics. Emission from galaxies in the far-infrared is known to be strongly correlated with obscured star formation and makes up a significant fraction of the cosmic infrared background. The Herschel Space Observatory was designed to cover the wavelengths where this background peaks and gave an unprecedented view of the Universe at this part of the electromagnetic spectrum. This thesis uses observations conducted for the Herschel Multi-tiered Extragalactic Survey (HerMES) by the Herschel Space Observatory's SPIRE instrument to catalogue sources of far infrared emission at 250µm, 350µm and 500µm which will allow models of galaxy evolution to be constrained. Source detection and extraction is performed with both blind and prior-driven algorithms on two newly constructed maps and, for the first time, on the largest of the HerMES' maps, HeLMS. Modifications are made to the algorithms to compensate for cirrus contamination. The accuracy and completeness of both algorithms are assessed and compared. In addition, emission from a galaxy across the three SPIRE bands is known to be strongly correlated. The extent of the covariance between the SPIRE images is explored using Principal Component Analysis, finding that up to 95% of the variance in the SPIRE maps can be explained by one linear combination of the three SPIRE bands. This new projection of data could allow previously unfeasible multi-dimensional probability of deflection analyses to be performed through a reduction of the data dimensionality. Looking forward, the Herschel Extragalactic Legacy Project (HELP) will be using Herschel observations from both HerMES and the H-ATLAS collaboration to create homogenous, multi-wavelength data products. In the final chapter, a comparison between datasets to test the homogeneity was undertaken, leading to a preliminary recalculation of the HerMES angular correlation function, a result that has been in contention between collaborations. Catalogues created as part of this thesis are available at hedam.lam.fr/HerMES/.