I take healpix nside=8 pixel centers and pull out a square (sine projection) grid around each of these with side length 30 deg from 1) FDS dust map at 150GHz and 2) WMAP K band (23GHz) Q,U polarization maps. The dust map I multiply by 0.05 to get some crude estimate of the polarized intensity. The WMAP maps I scale by (150/23)^-2.8 to convert to 150GHz.
I then appodize the maps with a sin^2 edge taper mask and take the power spectra. My code converts Q,U to E,B (naively) but since I have interpolated the WMAP maps from galactic to celestial coordinates without applying the proper rotation they aren't really E and B. So I just take the average.
Click here for an animation flipping through the first 50 fields.
Here is a plot showing all the spectra. The thick black line is the spectrum for the BICEP field (ra=0,dec=-57.5). Note that the WMAP spectra are noise dominated over most of the sky from below l=100.
I then read off the value of the spectra at l=100 and take the integral distrbutions. The points show where the BICEP field lies. The problem with this plot is that it makes it look like sync is way more of a problem than it is because the values are noise inflated.
Here are version for l=70 and l=50:
For kicks here is a plot showing the field centers and approx size.
