TL;DR
This paper characterizes the polarization response of the HERA-19 commissioning array, a crucial step for future Epoch of Reionization (EoR) observations. By comparing observational data to simulations, they demonstrate that their calibration method effectively isolates foregrounds in Stokes I, but reveals significant discrepancies in polarized Stokes parameters (Q, U, and especially V), highlighting the need for improved polarized calibration.
What Problem Does This Paper Address?
Detecting the faint, redshifted 21cm hydrogen signal from the Epoch of Reionization (EoR) is a major goal in cosmology. However, this signal is buried under much brighter foreground emissions, primarily synchrotron radiation from our own galaxy and extragalactic sources. These foregrounds are spectrally smooth, while the EoR signal is expected to be spectrally structured. Interferometers like the Hydrogen Epoch of Reionization Array (HERA) exploit this difference by looking at specific regions in the "delay spectrum" (a Fourier transform of the frequency spectrum) to separate foregrounds from the EoR signal.
A significant challenge is polarized foreground emission, which can leak into the unpolarized EoR signal through instrumental effects and miscalibration. Faraday rotation can further complicate polarized foregrounds. For instruments like HERA, which aim to "avoid" foregrounds rather than model and subtract them, understanding and characterizing the instrument's intrinsic polarization response is paramount to avoid contaminating the EoR signal. This paper specifically addresses this by investigating how polarized signals are handled by the HERA-19 commissioning array.
Methods
The authors employ an image-based calibration technique using the unpolarized diffuse emission of the Global Sky Model (GSM) as a reference. This calibration is applied to the HERA-19 commissioning array data. They then construct power spectra for all pseudo-Stokes parameters (I, Q, U, V) to analyze the polarization response.
Key methodological aspects include:
- Image-based calibration: Instead of relying on complex redundant calibration techniques (which will be used in future HERA builds), they use a simpler approach based on the known properties of the sky.
- Global Sky Model (GSM): This is used as a model for the unpolarized diffuse emission to calibrate the instrument.
- Pseudo-Stokes parameters: These are used to represent the observed visibilities, distinguishing them from true Stokes parameters.
- Power spectrum analysis: This technique is used to analyze the spectral structure of the data and compare it to simulations.
- Electromagnetic simulations: Detailed simulations of the HERA dish and feed were used to generate instrumental Mueller matrices.
Datasets
- HERA-19 commissioning array data: Eight nights of observational data were used.
- Global Sky Model (GSM): Used as an unpolarized sky model for calibration.
- Simulated data: Generated using the GSM and detailed electromagnetic simulations of the HERA instrument.
Key Insights and Ideas
- Calibration effectiveness for Stokes I: The simple image-based calibration, when applied to Stokes I (total intensity), successfully achieves qualitative redundancy between baselines and provides reasonable amplitude accuracy. This suggests the calibration is effective at removing instrumental effects that would otherwise distort the primary foreground signal.
- Polarized leakage is a significant issue: The power spectra for the polarized Stokes parameters (Q, U, and especially V) show substantial discrepancies compared to simulations that only account for beam leakage. This indicates that instrumental effects beyond simple beam leakage are significantly impacting the polarized measurements.
- Stokes V is highly problematic: The Stokes V power spectra are found to be highly discrepant from the expected zero power, strongly suggesting a need for more accurate polarized calibration techniques. This is a critical finding for future EoR experiments that rely on the unpolarized nature of the cosmological signal.
- D-terms are likely culprits: The authors point to direction-independent leakage through "D-terms" of the Jones matrix as a likely cause for the excess power in Stokes V. They also suggest that relative amplitude errors in antenna-based gains could explain discrepancies in Q and U.
- Calibration stability: The calibration procedure was found to be stable over the 8 days of observations.
Key Results
- The image-based calibration for Stokes I shows good agreement with simulations, indicating that the calibration does not add significant spectral structure beyond what's expected from the array configuration and primary beam.
- Excess power is observed in the linear polarization (Q and U) power spectra, which is not fully explained by primary beam leakage. This is attributed to a combination of residual calibration errors and actual polarized emission, with an estimated fractional polarization of ~10%.
- Stokes V power spectra show a significant excess compared to simulations, indicating a substantial miscalibration. The authors estimate that D-terms are likely responsible for this, with amplitudes relative to diagonal terms around ~3-5%.
- The calibration achieved an accuracy of about 10% for the Galactic Center observations, with about 20% variation from day-to-day.
Limitations and Future Work
- Incomplete calibration: The calibration used is described as "incomplete," particularly regarding the D-terms, which are crucial for accurate polarized measurements.
- Limited dynamic range: The current noise levels limit the dynamic range of the power spectra, hindering deeper probes of foregrounds and instrument response.
- Future calibration: The authors emphasize the need for more accurate polarized calibration techniques, including the analysis of D-terms. They suggest that future HERA builds with more antennas and improved electronics will allow for deeper integrations and a wider range of k-modes, enabling more thorough characterization of foregrounds and instrument effects.
- Understanding excess power: The origin of the excess power at a delay of ~1000 ns, seen across all polarizations, is not fully understood and requires further investigation.
Summary updated