Subtracting the kinetic Sunyaev-Zeldovich effect from the cosmic microwave background with surveys of large-scale structure

TL;DR

This paper proposes a method called de-kSZing to reduce the kinetic Sunyaev-Zeldovich effect in CMB maps by using large-scale structure surveys, aiming to improve cosmological measurements.

Contribution

It introduces a novel de-kSZing technique that constructs and subtracts kSZ templates from CMB maps using external surveys, with forecasts for various upcoming large-scale structure surveys.

Findings

Forecasts de-kSZing achieves 10-20% efficiency with current formalism.

De-kSZing can improve cosmological constraints from CMB data.

Higher efficiencies are expected with improved methods.

Abstract

The kinetic Sunyaev-Zeldovich (kSZ) effect will be an important source of cosmological and astrophysical information in upcoming surveys of the cosmic microwave background (CMB). However, the kSZ effect will also act as the dominant source of noise for several other measurements that use small angular scales in CMB temperature maps, since its blackbody nature implies that standard component separation techniques cannot be used to remove it from observed maps. In this paper, we explore the idea of "de-kSZing": constructing a template for the late-time kSZ effect using external surveys of large-scale structure, and then subtracting this template from CMB temperature maps in order to remove some portion of the kSZ signal. After building intuition for general aspects of the de-kSZing procedure, we perform forecasts for the de-kSZing efficiency of several large-scale structure surveys, including BOSS, DESI, Roman, MegaMapper, and PUMA. We also highlight potential applications of de-kSZing to cosmological constraints from the CMB temperature power spectrum, CMB lensing reconstruction, and the moving-lens effect. While our forecasts predict achievable de-kSZing efficiencies of 10-20% at best, these results are specific to the de-kSZing formalism adopted in this work, and we expect that higher efficiencies are possible using improved versions of this formalism.