Noise power spectrum estimation and fast map making for CMB experiments

Alexandre Amblard,Jean-Christophe Hamilton

DOI: https://doi.org/10.1051/0004-6361%3A20034066

2003-07-10

Abstract:We present a method designed to estimate the noise power spectrum in the time domain for CMB experiments. The noise power spectrum is extracted from the time ordered data avoiding the contamination coming from sky signal and accounting the pixellisation of the signal and the projection of the noise when making intermediate sky projections. This method is simple to implement and relies on Monte-Carlo simulations, it runs on a simple desk computer. We also propose a trick for filtering data before making coadded maps in order to avoid ringing due to the presence of signal in the timelines. These algorithms were succesfully tested on Archeops data.

Astrophysics

What problem does this paper attempt to address?

The problems that this paper attempts to solve are: in cosmic microwave background (CMB) experiments, how to accurately estimate the noise power spectrum in the time domain and quickly generate maps to avoid signal contamination and improve data - processing efficiency. Specifically, the author proposes a method to extract the noise power spectrum from time - ordered data while avoiding contamination from sky signals and taking into account the effects of signal pixelization and noise projection. In addition, a method to avoid the ringing effect caused by filtering when generating co - added maps is also proposed.

### Main problems:
1. **Noise power spectrum estimation**:
- How to accurately estimate the noise power spectrum in CMB experiments in the time domain while avoiding contamination from sky signals (such as galactic dust).
- The traditional maximum - likelihood method requires parallel supercomputers to solve large covariance matrices and is difficult to apply to modern experiments covering large areas of the sky (such as WMAP, Archeops, and Planck).

2. **Fast map generation**:
- How to quickly generate co - added maps after filtering, avoid the ringing effect caused by strong signals (such as galactic plane crossings), and ensure the quality of the maps.

### Solutions:
1. **Noise power spectrum estimation**:
- A method based on Monte Carlo simulation is proposed to provide an unbiased estimate of the noise Fourier power spectrum by correcting the initial guess.
- Using matrix notation, the signal contribution is removed by operating \( n_{\text{raw}} = d - A (A^T A)^{-1} A^T d \), where \( A \) is the pointing matrix.
- The effects of residual noise \( r_n \) and residual signal \( r_s \) are considered, and the magnitudes of these effects are determined by simulation, thereby de - biasing the noise power spectrum.

2. **Fast map generation**:
- A simple co - added map generation method is proposed to reduce the ringing effect by using a smoothed sine filter (instead of direct Fourier truncation).
- By first fitting a low - frequency function template, then removing this template from the initial data, then generating the map, and finally retaining the original data in the CMB region and removing the estimated galactic signal in the contaminated region, the influence of the galactic signal is effectively removed.

### Formula summary:
- The noise power spectrum is defined as:
\[
P(\nu) = \langle \tilde{n}^* \tilde{n} \rangle
\]
- Initial noise estimate:
\[
n_{\text{raw}} = d - A (A^T A)^{-1} A^T d
\]
- De - biased noise power spectrum:
\[
P_{\text{raw}}(\nu) = (1 - \epsilon(\nu))P(\nu) + \epsilon_s(\nu)
\]
where \( \epsilon(\nu) \) is the noise bias and \( \epsilon_s(\nu) \) is the signal bias.

Through these methods, the author has successfully improved the accuracy of noise power spectrum estimation and reduced the ringing effect in map generation, making CMB data analysis more reliable and efficient.