Amplitude modulated Gaussian dust sims
August 24 2018, Clem Pryke
We want to start experimenting with more realistic sky masks - see e.g. Sims with nominal Chile and Pole masks.
Planck Intermediate paper XXX showed that the degree scale dust BB power varies strongly across the sky - arxiv/1409.5738.
In Smallfield r-equivalent Maps Kenny showed that the PySM models do not necessarily track this variation faithfully.
Here I try to concoct an amplitude modulation template that can be applied to the existing "00" uniform Gaussian dust realizations to make them vary in brightness across in a realistic manner.
I start with the data file "norm_nHits_SA_35FOV.fits" taken from the end of the posting above. This contains the amplitude at ell=80 of power law fits to the BB power spectra of 400 sq deg circular patches taken from the Planck 353GHz map centered on each of the nside=8 pixel centers. (The half mission data split is used as this produces one of the smoother looking patterns.) Dividing these by 1.098^2 (see arxiv/1801.04945 sec. 3.3) to go from the P353 bandpass to 353GHz exactly (the BK convention for quoting A_d) and plotting on a linear scale we get the plot below. The white asterisk marks the center of the BK field (RA=0, Dec=-57.5). The BK14 constraint on A_d in the BK field is 4.3+1.3-1.0 uK^2. The value of the pixel in which the asterisk lies is 7.6 uK^2. It has long been noted that Planck sees a bit more dust power, and the comparison is not apples-to-apples anyway - so this all seems to check out OK.
I next take the map above, rotate to celestial coords, and re-render at nside=512 with 10deg Gaussian "beam" smoothing. Because of the smoothing the "good region" shrinks in a bit and the value at the location of the asterisk is now 13.6 uK^2.
The "00" Gaussian dust sim maps were generated using A_d=4.25uK^2. To make an amplitude modulation template to apply to them I therefore take the map above, divide by 4.25, and take the square root. This results in the plot below. The value at the asterisk is 1.8 and the minimum value anywhere is 1.1.
Taking realization 0 and multiplying on the template above results in the before/after plot below.
One could go on to do something similar for synchrotron. But maybe it is best to go with just dust to keep the conclusion clearer?