Assessing the Vera Rubin Observatory's Ability to Discover Asteroid Impactors Before They Collide with Earth

TL;DR

The paper addresses the challenge of detecting Earth-impacting asteroids across a wide size range and evaluates Rubin LSST’s capability using a novel synthetic impactor population generated from NEOMOD3 and processed with the Sorcha survey simulator. It reveals a strong size dependence in LSST detectability, with 79.7% of >140 m impactors discovered and only 10.5% of 10–20 m objects detected, and shows that long warning times are common for large objects but rare for small ones. A loss-mode analysis attributes LSST incompleteness primarily to photometric sensitivity for small impactors and cadence/linking constraints for larger ones, implying that LSST alone cannot guarantee long-lead warning. The study further demonstrates that a complementary high-cadence survey like Argus can recover impactors missed by LSST due to temporal sampling, suggesting that a coordinated, multi-survey approach will be essential for robust planetary-defense readiness in Rubin’s era.

Abstract

Asteroid impactors larger than ~10 m, from Chelyabinsk-scale airburst and Tunguska-scale events to >300 m continental threats, remain the dominant planetary-defense risk. While the Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST) will transform Solar System science, its observing cadence and survey design were not specifically optimized to discover imminent impactors. To assess its performance, we introduce a new method for efficiently generating synthetic impactor populations by minimally perturbing sampled NEOMOD3 orbits and evaluate their discovery efficiency with the Sorcha survey simulator. Our simulations show that LSST discovers 79.7% of large impactors (>140 m), decreasing to 50.3% for upper mid-sized (50-140 m), 26.8% for lower mid-sized (20 - 50 m), and 10.5% for small objects (10-20 m). Warning times of the discovered impactors show a similar size dependence: small objects are typically discovered only weeks before impact (median:12.4 days), lower mid-sized within a month (median: 21.5 days), and upper mid-sized objects on timescales of a few months (median: 106.2 days). 39.0% of large impactors are discovered more than a year before impact, lacking long-lead warning despite their brightness. A loss-mode analysis reveals the underlying cause that small impactors are limited mainly by photometric sensitivity, whereas mid-sized and large objects are missed primarily due to cadence and linking constraints from LSST and its Solar System Processing (SSP) Pipelines. These results show that LSST excels at discovering faint, small impactors, but cannot by itself guarantee long-lead warning across the hazardous size spectrum. Coordinated multi-survey strategies will therefore be essential in the LSST era to achieve robust planetary-defense capability, and we study a complementary high-cadence, shallow-depth example with the Argus Array.

Figures (10)

• Figure 1: Illustration of the synthetic impactor generation model. The plot compares the geocentric distance of a representative NEO before and after epoch adjustment. In Step 1, candidate NEOs are identified with nodal distances at $\sim$1 au. In Step 2, the object’s orbital epoch is shifted so that Earth and the NEO simultaneously arrive at the same ecliptic longitude, forcing an Earth-crossing geometry. The original orbit (black) has a minimum Earth-approach geocentric distance of 0.23 au, while the adjusted orbit (red) produces an impact configuration of closest geocentric approach at 0.03 au. This procedure preserves the object’s orbital elements ($a$, $e$, $i$) while altering the mean anomaly to generate a dynamically consistent synthetic impactor.
• Figure 2: Size distribution and discovery efficiency of the synthetic impactor population, computed using uniform 10 m diameter bins up to 140 m and a collapsed terminal bin for all objects larger than 140 m. Top: Log-scaled histograms showing the full population (dark brown) and the subset discovered by LSST (red). A break in the x-axis isolates the $>140$ m population corresponding to the PHA hazard regime. Bottom: Discovery fraction (Discovered/All) for the same size-bin setup.
• Figure 3: Fraction of impactors discovered at least $X$ (time) before impact, shown for four size bins (10--20 m, 20--50 m, 50--140 m, and $>140$ m). Values represent cumulative fractions of the full impactor population in each size bin discovered with warning times $\ge X$; the $\ge 0$ bin corresponds to full discovery efficiency. The y axis is logarithmic. Discovery probability and warning time increase strongly with impactor size, with small objects rarely discovered far in advance. Most discovered impactors receive warning times of only weeks to months, while year-scale warning is rare.
• Figure 4: Two representative synthetic-impact cases illustrating how visibility does—or does not—translate into discovery. Panel (a) demonstrates all three loss mechanisms: segments with no blue points represent pointing loss; blue points without red crosses indicate magnitude or trailing loss; and isolated red-cross detections that never form tracklets represent linking loss. Panel (b) shows a case where all conditions align, enabling full linkage and discovery. Together these two examples encapsulate the essential survey failure modes that govern LSST's impactor completeness.
• Figure 5: Fraction of impactors discovered at least X before impact for four diameter bins (10–20 m, 20–50 m, 50–140 m, and $>$140 m), shown for (a) LSST and (b) an Argus-like survey. The y-axis lower limit is clipped at $10^{-3}$ (0.1%) to avoid unnecessary blank space. For the Argus case, several size--time combinations fall at or below this level, indicating discovery fractions consistent with close to zero at the resolution of this plot.