LSST Discovery Potential for Solar System Science and Planetary Defense

Transcription

1 LSST Discovery Potential for Solar System Science and Planetary Defense Zeljko Ivezic Mario Juric Lynne Jones Colin Slater Joachim Moeyens (University of Washington) and the LSST Project. SBAG June 14, 2017 NASA REPORT JULY 21, 2016 Name of Meeting Location Date - Change in Slide Master 1

2 Outline 1. A comment about the JPL study of LSST performance 2. Estimates of false positive rates for LSST 3. A general comment about NEO completeness estimates 4. What will LSST do for NEOs? SBAG JUNE 14,

3 1. A comment about the JPL study of LSST performance analogous simulations were also performed by the LSST Project; detailed report is public: reported IOD (Initial Orbit Determination) runtimes for existing codes are from several times to up to an order of magnitude shorter than planned LSST allocation for MOPS therefore, there is quite a robust compute reserve (N.B. the MOPS compute needs are at least an order of magnitude more modest than resources needed for other LSST data processing) these simulations and improved understanding of the algorithms imply that false positive rates several times higher than anticipated could be readily handled (relatively easily up to about 10 times higher rates) most importantly, our results are fully statistically consistent with the results reported by the JPL group SBAG JUNE 14,

4 2. Rates of false positives in image differencing expected for LSST We have empirically estimated the false positive rate using DECam images and LSST software. The full report is publicly available at Summary: - DECam data from DECam NEO survey (PI: Lori Allen, NOAO) - need to be careful about covariant per-pixel noise when computing SNR - for SNR>5 threshold: sources per sq. deg. - after exclusion of static sources (1% loss of fill factor), there are on average 350 sources per sq. deg. (includes false positives, asteroids, and true astrophysical transients) - based on visual inspection, an upper limit for false positives in DECam imaging data is 263 sources per sq. deg., and implies a rate of 460 per sq. deg. for LSST SBAG JUNE 14,

5 2. Rates of false positives in image differencing expected for LSST Need to be careful about covariant noise when computing SNR: - the rate of false positives is a strong function of SNR: False positive rates due to background fluctuations for LSST: SNR FP (per sq.deg) 4.0 4, It is likely that at least some reported high false positive rates are due to this effect. SBAG JUNE 14,

6 3. A general comment about NEO completeness estimates One has to be extremely careful when comparing different simulations because there are a number of important systematics effects, e.g.: - NEO vs. PHA: 3-4% (larger for PHAs) - the contribution of known objects (~10-20%) - the window width: 30 days vs. 15 days is worth 2% - effective detection depth and NEO brightness prediction: 0.2 mag is worth 2% - of course, cadence details matter, too One can easily change numerical estimates from 55% to 85% by changing underlying assumptions! SBAG JUNE 14,

7 4. What will LSST do for NEOs? From the science characterization point of view, LSST is not strongly driven by the NEO completeness. For example, reducing the bias in and understanding the selection function is a much stronger science driver. 6% of observing time is optimized for asteroids (300 visits in griz) The current baseline cadence is optimized for science returns. For detailed analysis, see SBAG JUNE 14,

8 NEO-enhancements to baseline cadence baseline: minion_1016 The completeness for the baseline cadence (left) is still rising after 10 years of surveying - an additional 3-4% can be gained with, e.g. two, additional years of surveying: this motivates astro_lsst_01_1016 1) astro_lsst_01_ increased Ecliptic coverage (24%, with 69% for the main survey; for baseline 6% and 85%) - in 12 years, the main survey gets 98% of the observing time allocated to it in the nominal 10- year survey: for other science programs, this cadence has very similar performance as the baseline cadence astro_lsst_01_1016 years SBAG JUNE 14,

9 Baseline vs. NEO-enhanced cadences: Main conclusions: 1) By improving analysis software and running an NEO-optimized survey (astro_lsst_01_1016) for 12 years, the completeness for PHAs with H<22 can be boosted from 66% to 77% 2) With the contribution of known objects, the total completeness could approach 90% 3) The NEO-optimized cadence obtains as many visits for the main survey in 12 years as the baseline survey in 10 years: therefore, the impact on other science programs is about neutral (e.g. negative impact: it takes 5 years for what is done in 4 years with baseline; positive impact: proper motion errors larger by 20% using baseline cadence) 4) The NEO-optimized cadence astro_lsst_01_1016 could serve as a model for NASA s participation in LSST operations. SBAG JUNE 14,

10 Summary 1. Our simulations agree with the JPL study. 2. Empirically estimated upper limit for the LSST false positive rate in image differencing is 460 per sq. deg MOPS can handle this rate of false positives with about two orders of magnitude less computing capacity than planned for other LSST data processing needs LSST cadence strategy can be optimized to boost the NEO completeness. Assuming a 12-year survey, and modest improvements in software and data management resources, the completeness for PHAs with H<22 can be (optimistically) increased to close to ~90% (assuming a contribution from known objects) 5. Simulation astro_lsst_01_1016 spends 24% of time on NEO-optimized observations and runs 2 years longer: it could serve as a model for NASA s participation in LSST operations. SBAG JUNE 14,