Deeper SAT from Chile

From CMB-S4 wiki
(Redirected from Deeper SAT)
Jump to: navigation, search

This post has been superseded by Deeper SAT from Chile II.

Deeper SAT observations

In this post we build observing schedules that maximize the depth and minimize the sky are observed using the SATs in Chile.

We begin by considering the estimated foreground intensity and polarization at 100 and 150GHz using the Planck sky model. The difference between the two is that one can clearly identify the North Galactic Spur synchrotron emission in the 100GHz maps:

Priority by pixel fwhm10 nbin10.comp.png

All maps are smoothed with a 10-degree Gaussian beam. Instead of plotting the estimated temperature and polarization amplitude, we divide the pixels into 10% quantiles, showing the cleanest and "dirtiest" regions of the sky. Because the foreground polarization is our primary concern, the intensity maps were thresholded so that all pixels that fall in the cleaner 80% of the sky are re-directed into the cleanest bin. We point out that the part of the sky with the least foreground polarization only partially overlaps with the regions where the foreground intensity is minimal (the so-called Southern hole). We combine the two intensity and two polarization amplitude maps by using the maximum quantile over all four maps.

We consider that the minimal throw in azimuth should be at least 20 degrees to resolve large scale noise modes. Observing from Chile, that translates into a minimum 20 degree span in declination of our deep patch. The span in right ascension is not constrained as much, but making the patch wider in RA extends the time the patch overlaps with the allowed observing elevations (50 - 70 degrees in this study).

Here we propose a tiling of the sky using 10x20 degree overlapping tiles that arranged into three tiers:

  1. First tier tiles are outlined in red and are always targeted when available
  2. Second tier is outlined in black and targeted when Tier 1 is not available
  3. Third tier (grey) is targeted when neither of the higher tiers can be acquired.

Priority by pixel fwhm10 nbin10.full.png

Observations in each Tier are balanced to produce maximally uniform hit distributions. That is to say that the scheduler attempts to schedule an equal number of scans over each of the tiles in the same Tier. We also instructed the scheduler to balance rising and setting scans and to prefer higher elevation observations over lower elevation observations.

The resulting yearly observing schedule has 77.5% efficiency with 282.8 total days of observing. 103.6 days (28% of the year) is spent observing Tier 1 tiles leading to two deep patches with well-observed wings.

Here is the daily observing efficiency split between the three tiers:

Sat efficiencies.png

We ran a scanning simulation with a 35-degree wide hexagonal focal plane to mimic actual SAT scanning as was done in "Designs for next generation CMB survey strategies from Chile" [1]. Here is a comparison of the resulting map depth between the earlier study and our observing schedule:

Depth comparison.png

Our hit map is more focused than the earlier map with the deepest parts of the Southern patch observed 10-20% deeper than any pixel in the earlier schedule. Curiously, there is very little difference in the Northern patch. We suspect this to be caused by the Sun and Moon avoidance. Both avoidance radii were set to 45 degrees like in the earlier study. Apart from the deeper scanning, the new schedule also better focuses the observations based on expected foreground emission.


Degree-scale foreground emission

Here are the foreground plots with a 1-degree smoothing instead of the 10-degree smoothing.

Priority by pixel fwhm01 nbin10.comp.png

The tile definitions haven't changed:

Priority by pixel fwhm01 nbin10.full.png

At this resolution it would seem that moving the Southern Tier 1 designation towards the zero meridian would lead to an even cleaner deep patch. However, we found that just a 15-degree shift of the Southern Tier 1 tiles reduced the total time spent observing Tier 1 tiles over the year by 10%. The likely reason is that this configuration increases the time both deep patches are within observing elevation at the same time.