Sky masks for simulations
Feb 19 2018, Clem Pryke
Up to now for CMB-S4 data challenge sims we have used toy circular relative hits maps with the default being flat to radius=12deg and then rolling to zero as cosine squared over an additional 15deg. This results in f_sky effective of 3%. We centered it at RA=0 and Dec=-45. The degree of edge taper is approximately what one gets with a BICEP3 type instantaneous field-of-view.
We have also considered 1% patch with flat/taper 3.5/15 and 10% patch with flat/taper 30/15.
These were always intended as idealized. Only the 1% could be (wholly) observed from Pole and real experiments do not generate circular patches. On the attached plot these masks are compared to the current BICEP/Keck mask.
For any given mask we can compute the resulting sigma(r). For uniform foreground level this will vary weakly with area/shape/apodization etc. The problem is that when including a foreground model with amplitude which varies across the sky - as we routinely do - it is this variation which is going to drive the result. (It is for this reason that the center of the 10% patch got pushed wrt the 3% patch in the attached plot.)
At ell approx 80 the Planck 353GHz maps hit their noise floor over a good fraction of the sky. We need to a) verify that the existing foreground models do not violate the Planck data (by producing levels higher than reality in the regions where reality is known), b) compare how the models vary in the cleaner regions, and c) accept that there is and will remain significant uncertainty from foregrounds as to the sigma(r) that would be achieved from observing any given patch of sky.
This uncertainty limits the utility of "high accuracy" simulations of scan strategy etc. For example for uniform foreground I would guess that sigma(r) might change by 10% when going from the f_sky=1% to BK14 masks in the attached plot. But with non-uniform foreground it might change by much more - and it might change by factor 2 or more just moving to another place on the sky as dictated by observing from Pole vs. Chile. I will work on quantifying.