Tests of AMiBA Data Integrity

TL;DR

This paper details statistical methods for validating AMiBA interferometric data, ensuring data quality for CMB and Sunyaev-Zel'dovich effect measurements through noise analysis and Gaussianity tests.

Contribution

It introduces comprehensive data validation techniques, including power spectrum and Gaussianity tests, specifically tailored for AMiBA's observational data.

Findings

No significant electronic offset variation over 10-minute observations.

Noise correlation between different lags is less than 10%.

Noise variance scales inversely with observation time.

Abstract

We describe methods used to validate data from the Y.T. Lee Array for Microwave Background Anisotropy (AMiBA), an interferometric array designed to measure the Sunyaev-Zel'dovich effect and the anisotropy of the Cosmic Microwave Background (CMB). We perform several statistical tests on data from pointed galaxy cluster observations taken in 2007 and noise data from long-term blank sky observations and measurements with the feeds covered by the absorbers. We apply power spectrum analysis, cross power spectrum analysis among different outputs with different time lags in our analog correlator, and sample variance law tests to noise data. We find that (1) there is no time variation of electronic offsets on the time scale of our two-patch observations (~10 minutes); (2) noise is correlated by less than 10% between different lags; and (3) the variance of noise scales with the inverse of time. To test the Gaussianity of the data, we apply Kolmogorov-Smirnov (K-S) tests to cluster data, and find that a 5% significance level efficiently detects data sets with known hardware problems without rejecting an excess of acceptable data. We also calculate third- and fourth-order moments and cumulants for the noise residual visibilities and find that about 95% of our data are within the 99% confidence regions of Gaussianity.