Skip to main content
Open Access Publications from the University of California

Overcoming the Challenges of 21cm Cosmology

  • Author(s): Pober, Jonathan
  • Advisor(s): Parsons, Aaaron
  • et al.

The highly-redshifted 21cm line of neutral hydrogen is one of the most promising and unique probes of cosmology for the next decade and beyond. The past few years have seen a number of dedicated experiments targeting the 21cm signal from the Epoch of Reionization (EoR) begin operation, including the LOw-Frequency ARray (LOFAR), the Murchison Widefield Array (MWA), and the Donald C. Backer Precision Array for Probing the Epoch of Reionization (PAPER). For these experiments to yield cosmological results, they require new calibration and analysis algorithms which will need to achieve unprecedented levels of separation between the 21cm signal and contaminating foreground emission. Although much work has been spent developing these algorithms over the past decade, their success or failure will ultimately depend on their ability to overcome the complications associated with real-world systems and their inherent complications.

The work in this dissertation is closely tied to the late-stage commissioning and early observations with PAPER. The first two chapters focus on developing calibration algorithms to overcome unique problems arising in the PAPER system. To test these algorithms, I rely on not only simulations, but on commissioning observations, ultimately tying the success of the algorithm to its performance on actual, celestial data. The first algorithm works to correct gain-drifts in the PAPER system caused by the heating and cooling of various components (the amplifiers and above ground co-axial cables, in particular). It is shown that a simple measurement of the ambient temperature can remove ∼ 10% gain fluctuations in the observed brightness of calibrator sources. This result is highly encouraging for the ability of PAPER to remove a potentially dominant systematic in its power spectrum and cataloging measurements without resorting to a complicated system overhaul.

The second new algorithm developed in this dissertation solves a major calibration challenge not just for PAPER, but for nearly all of a large class of new wide-field, drift- scanning radio telescopes: primary beam calibration in the presence of a poorly measured sky. Since these telescopes lack the ability to steer their primary beams, while seeing nearly the entire sky at once, a large number of calibrator sources are necessary to probe the entire beam response. However, the catalogs of radio sources at low-frequencies are not reliable enough to achieve the level of primary beam accuraccy needed for 21cm cosmology experiments. I develop, test, and apply a new technique which -- using only the assumption of symmetry around a 180◦ rotation -- simultaneously solves for the primary beam and the flux density of large number of sources.

In this dissertation, I also present the analysis of new observations from PAPER to test theoretical models which predict foreground emission is confined to a "wedge"-like region of cosmological Fourier space, leaving an "EoR window" free from contamination. For the first time in actual observations, these predictions are spectacularly confirmed. In many ways, this result shifts the burden for upcoming PAPER analysis from foreground removal to increased sensitivity. And although increasing sensitivity is no small feat in-and-of-itself, this result is highly encouraging for 21cm studies, as foreground removal was long-viewed as the principal challenge for this field.

The final result in this dissertation is the application of the all the lessons learned building PAPER and the MWA to design a new experiment for 21cm studies at z ~ 1 with the goal of measuring baryon acoustic oscillations (BAO). The design of the BAO Broadband and Broad-beam (BAOBAB) Array is described, and cosmological forecasts are presented. The bottom line is highly encouraging, suggesting that z ~ 1 21cm observations can detect the neutral hydrogen power spectrum with a very modest (16 − 32 element) array, and that still reasonably sized (128 − 256 elements) arrays can produce significant advances in our knowledge of dark energy.

Main Content
Current View