Data analysis techniques to control systematics in cosmological surveys

UoM administered thesis: Phd

  • Authors:
  • Nialh Mccallum


Future cosmic microwave background (CMB) surveys aim to reach unprecedented levels of sensitivity in upcoming experiments. Consequently an unprecedented control of instrumental systematics will be required. This will require the refinement of existing techniques along with the development of novel methods to characterise and remove the effects of systematics on the observations. This thesis will present several techniques to deal with systematics in CMB surveys and additionally demonstrate their suitability to be extended to the relatively new method of performing cosmological surveys with intensity mapping (IM). After introducing the relevant background material in Chapter 1, in Chapter 2, a formalism is presented that characterises systematic signals which are closely coupled to the scanning strategy in terms of their spin. It is demonstrated how the formalism may be used to describe the mixing of different spin signals analytically and this is validated with time ordered data (TOD) simulations of satellite and ground-based CMB surveys. The formalism is applicable to both full- and partial-sky surveys and may be used to provide a complete elucidation of the leakage occurring, including both intensity-to-polarisation leakage and polarisation mixing; these will both factor crucially in future CMB experiments and this formalism should aid in forecasts of the effects of systematics on CMB surveys accordingly. Additionally it is shown that by utilising the spin-dependence of the systematic signals a simple extension can be made to simple binning map-making that is capable of disentangling spin-coupled systematics from the desired spin signal to be reconstructed in the map-making process. Chapter 3 builds upon the work of Chapter 2 to develop an approach to simulate rapidly systematics that affect CMB observations whilst including the effects of the scanning strategy. Summary properties of the scanning strategy are utilised to capture features of full TOD simulations at much improved rates of computation when simulating the effects of systematics. Using differential gain, pointing, and ellipticity systematics as examples the ability of the map-based approach to capture the salient features of the scan is demonstrated by direct comparison to full TOD simulations, with sub-percent levels of accuracy achieved. The significant speed up afforded by the map-based approach means it should facilitate quick forecasts of systematics for upcoming CMB surveys which include the effects of the scanning strategy. Finally a demonstration of the fast map-based approach is also performed on a full focal plane setup showing its ability to incorporate realistic numbers (thousands) of detectors as seen in modern CMB experiments. In Chapter 4 the effect of the design of the scanning strategy of CMB surveys on certain systematics is elucidated. Most ground-based CMB instruments are designed with the capability to rotate about their boresight pointing direction; it is shown that implementing partial boresight rotation in ground-based scanning strategies can aid in the suppression of a number of systematics. In particular using the formalism outlined in Chapter 2 it is shown that the spin dependence of certain systematics means that they can be almost completely removed by observing at specific pairs of boresight angles in a given sky pixel. This is illustrated using simulations of differential gain and pointing systematics for deep and wide scanning strategies appropriate for next-generation ground-based experiments based in Chile. Chapter 5 builds upon the formalism outlined in the context of CMB experiments in Chapter 2 and expands it to incorporate IM surveys. Using scan strategy information from IM surveys the effects of different instrumental systematics on the recovered cosmological intensity signal are modelled according to their spin symmetry. Full TOD simulations are used to demonstrate that when implementing a straightforward extension of simple binned map-making certain systematics can be disentangled from the intensity signal based on their spin properties. It is shown that through this simple map-making procedure the contamination to the observed intensity signal is reduced significantly and crucially that this approach works for signals that are non-smooth in frequency, e.g. polarized foregrounds leaked via systematics. These map-making approaches are simple to apply and represent a complementary approach to existing techniques for removing systematics from upcoming IM surveys. Lastly, in Chapter 6 the final conclusions are drawn and possible future directions for the research are identified.


Original languageEnglish
Awarding Institution
Award date1 Aug 2022