midPointMjdTai in DiaSource DP0.3

Hi, I was wondering how exactly midPointMjdTai is calculated in the DiaSource table of DP0.3? Specifically, when a visit is composed of 2 x 15s exposures how is any overhead time between exposures (e.g. readout) accounted for?

Hi @jrob93, thanks for this question! This is a subtle detail but, as I understand it, since DP0.3 used the ‘baseline version 3’ observing strategy simulation (Section 3 of Rubin Project Science Tech Note 055), and that strategy used 1x30 second exposures, there’s no overhead time to account for.

Special thanks to @Gerenjie for help with this one!

@MelissaGraham I am not sure that’s quite right. All baseline cadences being simulated are 2x15s exposures at the moment.

From Section 3 - "In the 𝑢 band, visits are acquired as a single exposure of 30 seconds to avoid becoming read noise limited; in 𝑔𝑟𝑖𝑧𝑦 bands, visits are acquired as a back-to-back set of two 15-
second exposures. These 2x15s exposures are less efficient than a single 1x30s visit, due to additional shutter and readout time. During commissioning, 1x30s visits for all bands will be evaluated and, if viable, could be adopted; this would lead to an increase in efficiency of ∼ 7%. "

Ah! Thanks @mschwamb. I think we need @Gerenjie and @ljones to comment further on this question then.

So - I don’t know exactly how the moving object sources for DP0.3 were simulated and whether any offset from the times “observationStartMJD” used in the opsim were included. One way you might check is to see if the “midPointMjdTai” values match the “observationStartMJD” values from baseline_v3.0_10yrs (I’m pretty sure the simulation used v3.0).

I would note that we don’t exactly specify whether the “observationStartMJD” is in TAI or UTC, but since the leap seconds for the years of Rubin operation aren’t known either, it’s probably fine to treat them as equivalent. We do consider the times to be the start of observations … but I suspect that treating them as midpoints would be fine as well (which might be what happened, I’m not sure). It’s entirely possible you will find that the midPointMjdTai values are consistently 17 seconds after our observationStartMJD times.
(observationStartMJD - t0 = start of exposure at one side of the focal plane. 15 seconds on-sky time, and then shutter closes for 1 second … by the time it’s fully closed, it is t0 + 16 seconds. There is a 2 second readout, then the second exposure starts at t0+18 seconds. The shutter is fully closed on the second exposure at t0+34seconds.
The midpoint exposure time varies across the focal plane within each single exposure however since the shutters move in different directions in the first vs. second exposure, the final midpoint exposure time across the full focal plane is the same and should be t0+17 seconds.)

I think DP0.3 does not have any solar system observations in u band, so the problem of figuring out exactly what the midPointMjdTai for these single exposure visits is wouldn’t have come up (although it would only vary by 1s across the focal plane).

Hi @jrob93 , just wanted to follow up to see if the responses above have addressed your question about midPointMjdTai. Based on @ljones’s response, it sounds like it’s t0 + 17 seconds (for the non-u-band values since there is no simulated u-band observations in DP0.3).

Hi all, thanks for the really helpful replies. @ljones solution provides the information I need (I just haven’t been in the habit of checking my community forum notifications!)