High cadence LAT from Chile
March 18, 2019 - Reijo Keskitalo
In this work we explore increasing the cadence of Chile LAT scanning. The target is to observe the entire wide field every one or two days to facilitate finding transient signals. We do not modulate the observation priority based on the level of foreground emission.
For scientific justification for the high cadence, see Nathan Whitehorn's talk on transient science from the 2019 collaboration meeting at Fermilab: Media:CMBS4TransientsFNAL19.pdf
The basic strategy is to scan at the lowest acceptable elevation (45 degrees in this study) and scan continuously as wide as reasonable (140 degrees in this study). Whenever the scan is interrupted by the Sun or Moon, the telescope is rotated 180 degrees and the scanning is continued.
The challenge with the wide throw is that the telescope spends a disproportionate amount of time near the extremes of the visible DEC range. To even out the hit distribution, we add a second class of scans at a higher observing elevation (55 degrees), leading to a narrower range of visible declination. By setting the priority of the 55 deg observations to half of the 45 deg observations, we instruct the scheduler to queue these observations at a rate of 1:2 of the 45 deg observations. By combining 45 and 55 degree observations, we also improve the cross linking.
Furthermore, we use 30 degrees as the Sun and Moon avoidance angle instead of the more conservative 45 degrees. With the wide throws, it seems sensible to make the scanning more continuous while still retaining the possibility of excising samples from the analysis if they are considered suspect.
You can find a movie of the Chile observable sky in Chile Scanning Movie.
The resulting schedule has 88.6% observing efficiency which could be improved by either reducing the Moon avoidance angle or adding a third class of scans that target low declinations. The target of one or two day cadence is reached as can be seen in these daily hit maps through January:
The combined hits after one month look like this:
Even with the missing observations in the total hit map, the sky coverage is 62.7%. The gaps in the maps are caused by the Sun (moving only 30 degrees over the month) and the Moon (traveling around the ecliptic equator over the month).
Here is a full year hit map assuming 1000 pixels and 100Hz sampling rate:
Further work (04/04/2019)
Let us assume that it is possible to modulate the scan rate to balance the integration time. In our high cadence scanning strategy majority of the time is spent observing at fixed elevation and that elevation translates into a fraction of the sky that can be viewed at that elevation. Using Denis Barkats' and Benjamin Racine's elevation noise curves from S4 NET forecasts III, we may try to quantify the relationship between maximal fsky and the elevation penalty in noise.
Here are the predicted NET:s relative to observing at 55 degrees for 100mK bath:
and this is how the minimum observing elevation translates into viewable sky area:
Since the 270 GHz channel is the worst affected by lowering observing elevation, we use it to compare fsky to NET:
We tabulate here some representative numbers for the 270GHz case:
|el_min [deg]||fsky (raw)||NET / NET55|