Hi, I’ve been looking at DP0.3 to determine the expected number of SSO alerts we will get per visit in LSST. To do this I queried DP0.3 to get the total number of SSO alerts per night and then loaded baseline_v3.0 to get the number of visits per night. I divided the two to get a mean number of alerts per visit (per night). Here’s a notebook showing the results: https://github.com/lsst-uk/adler/blob/main/notebooks/SSO_alerts_per_visit/alerts_per_night.ipynb
I get a value of ~300 SSO alerts per visit and I note that this is 10x lower than the estimate given in https://dmtn-102.lsst.io/.
I assume that this lower value is to be expected, as the numbers in DMTN-102 appear to assume that all asteroids in the field of view of a visit would be detectable. However this does not take into account distance, phase angle or asteroid colour/filter effects that would change the observed brightness, whereas these effects are dealt with in DP0.3. I am asking if this all sounds reasonable, probably best answered by @MelissaGraham?
Thanks for bringing this up. The 3k/visit in DMTN-102 seems like an overestimate/mistke; we’ll look into it.
Early in the planning (~2012 era) we’ve estimated that we’d reach up to ~5,000/visit in the densest areas of the ecliptic. The ballpark number to use would be ~1Bn observations over a 10yr survey (~2.5M visits), giving some ~400/visit on average. This is consistent with the number you’re getting from DP0.3. Note we have a known bug in DP0.3; it underestimates the number of observations by up to ~20-30%, and missing u and y bands – a fix is in the works. This has been in our PCW and LSST@Eu5 talks, but we forgot to update https://dp0-3.lsst.io/ (@MelissaGraham, I’ll have Pedro send you a “Known Issues” paragraph).
I agree that the lower number is to be expected – DP0.3 was a much higher-resolution simulation than the “5.5 million / 18,000 * 10” calculation in DMTN-102. However, there are individual visits in DP0.3 which have >3,000 detections.
I think I will take a closer look at the maximum number of SSO alerts per visit in DP0.3. So far I have just looked at total SSO alerts/visits per night to get an average and therefore my maximum values have been lower than the >3000-5000 detections you have both mentioned.
I will look at some of the ecliptic visits individually to get a better idea of the peak number (knowing now that DP0.3 is an underestimate).
I’ve marked Mario’s reply as the “solution” to this topic, but do let us know if the peak value for number of SSO’s per visit near the ecliptic in DP0.3 doesn’t match expectations.