Correlation Analysis between Tibet AS-$\gamma$ TeV Cosmic Ray and WMAP Nine-year Data

TL;DR

This study investigates potential residual foreground contamination in WMAP nine-year data by analyzing correlations with TeV cosmic ray data, finding evidence of residuals with about 95% confidence, but not conclusively.

Contribution

It introduces a correlation analysis method using cosmic ray data to assess foreground residuals in WMAP maps, highlighting the potential presence of residuals and their relation to cosmic ray anisotropy.

Findings

CRs can trace foregrounds in WMAP data

Some TeV CRs originate from the Milky Way

Foreground residuals are likely present with ~95% confidence

Abstract

The Wilkinson Microwave Anisotropy Probe (WMAP) team subtracted template-based foreground models to produce foreground-reduced maps, and masked point sources and uncertain sky regions directly; however, whether foreground residuals exist in the WMAP foreground-reduced maps is still an open question. Here, we use Pearson correlation coefficient (PCC) analysis with AS-$\gamma$ TeV cosmic ray (CR) data to probe possible foreground residuals in the WMAP nine-year data. The correlation results between the CR and foreground-contained maps (WMAP foreground-unreduced maps, WMAP template-based and MEM foreground models) suggest that: (1) CRs can trace foregrounds in the WMAP data; (2) at least some TeV CRs originate from the Milky Way; (3) foregrounds may be related to the existence of CR anisotropy (loss-cone and tail-in structures); (4) there exist differences among different types of foregrounds in the declination range of $< 15^{\circ}$. Then, we generate 10,000 mock CMB sky maps to describe the cosmic variance, which is used to measure the effect of the fluctuations of all possible CMB maps to the correlations between CR and CMB maps. Finally, we do correlation analysis between the CR and WMAP foreground-reduced maps, and find that: (1) there are significant anticorrelations; and (2) the WMAP foreground-reduced maps are credible. However, the significant anticorrelations may be accidental, and the higher signal-to-noise ratio (SNR) Planck SMICA map cannot reject the hypothesis of accidental correlations. We therefore can only conclude that the foreground residuals exist with $\sim$ 95\% probability.