A study of fundamental limitations to statistical detection of redshifted HI from the epoch of reionization

In this paper we explore for the first time the relative magnitudes of three fundamental sources of uncertainty, namely, foreground contamination, thermal noise and sample variance in detecting the HI power spectrum from the Epoch of Reionization (EoR). We derive limits on the sensitivity of a Fourier synthesis telescope to detect EoR based on its array configuration and a statistical representation of images made by the instrument. We use the Murchison Widefield Array (MWA) configuration for our studies. Using a unified framework for estimating signal and noise components in the HI power spectrum, we derive an expression for and estimate the contamination from extragalactic point-like sources in three-dimensional k-space. Sensitivity for EoR HI power spectrum detection is estimated for different observing modes with MWA. With 1000 hours of observing on a single field using the 128-tile MWA, EoR detection is feasible (S/N > 1 for $k\lesssim 0.8$ Mpc$^{-1}$). Bandpass shaping and refinements to the EoR window are found to be effective in containing foreground contamination, which makes the instrument tolerant to imaging errors. We find that for a given observing time, observing many independent fields of view does not offer an advantage over a single field observation when thermal noise dominates over other uncertainties in the derived power spectrum.