From CMB-S4 wiki
Foregrounds for lensing: challenges, solutions and open questions
A new regime:
- lensing power spectrum is obtained from a measurement of the four-point function of the CMB, e.g. <EBEB>
- for most applications, most weight comes from polarization, which is probably cleaner (though should also use temperature if we have multifrequency data)
- non-Gaussian galactic or extragalactic foreground can induce biases
- precision requirements are extremely high - need biases below roughly 0.2%
Most significant polarization lensing foregrounds
1. Radio Sources
- contamination was estimated to be moderate
- can further reduce contamination by roughly an order of magnitude by "source hardening" - estimate source statistical anisotropy and project out of lensing
- multifrequency data will help further - sources are probably a solved issue
2. Galactic Dust
- though falls off on smaller scales, not obviously negligible compared to B, perhaps important for EB estimator
- don't know high-ell polarized dust structure. To start, make EXTREMELY rough estimate using Planck 545 GHz maps. This definitely needs to be improved!!!
- not huge (if estimate correct), but not negligible - may need multifrequency data even at high ell.
Open Questions and Tasks
- Simulate high-L polarized dust (how?) and run through lensing pipelines
- How do these foregrounds affect delensing? Do they set a limit on delensing?
- Simulate full delensing procedure with foreground simulations
- How many high-ell bands do we need for S4 lensing?
- Can temperature foregrounds be removed to sufficient accuracy?
- Include bias hardening in forecasts?