In 20201116_noise_int_down we looked at how noise integrates down with the number of detectors for polarization. This post just repeats the same for the \(TT\) spectrum.
The left plot below shows the first 100 realizations when co-adding over 1, 2, 3, 4 or 5 Keck receivers for 2013 (where each receiver contains 500 detectors). The right plot just takes the mean over the realizations. I also add a lines for BICEP2 which is 500 detectors over 3 years. We can see that the \(TT\) \(N_\ell\) spectrum is nearly "all \(1/f\)" - only flattens at the highest \(\ell\).
Fig2:
As in the previous post we can plot the low \(\ell\) and high \(\ell\) values as a function of nominal map depth. The dashed lines just scale the first point by the ratio of the total weights. Clearly 500 is already more than enough simultaneous detectors for measuring T.
Fig3:
Here is a plot showing the pixel layout of all five Keck receivers overlaid:
Fig4:
