The ESA Meerkat Asteroid Guard: a monitoring service for imminent impactors

TL;DR

The paper presents the Meerkat Asteroid Guard, an automated monitoring service from the ESA NEO Coordination Centre that rapidly assesses imminent impactors by performing systematic ranging on short-arc NEOCP tracklets, computing a posterior grid over topocentric range and range-rate, and translating this into impact probabilities and actionable follow-up guidance. It couples a Gaussian-error orbit-fit with physically informed priors, then uses Monte Carlo sampling to map possible impact locations and to optimize telescope pointing, delivering alert materials including plots and dashboards. The v2.0 implementation leverages the GODOT dynamics library, Python 3.11, CI/CD, and dockerized pipelines to achieve fast alert generation (average ~56 s per alert) and high operational reliability, issuing alerts for notable past events and coordinating with observers to enable rapid confirmation and meteorite searches. Collectively, Meerkat demonstrates a scalable, real-time approach to imminent-impactor warning, complements existing alert systems, and remains poised to exploit upcoming facilities like Flyeye and Rubin for enhanced planetary defense effectiveness.

Abstract

We present the Meerkat Asteroid Guard, an imminent impactor warning service developed and maintained by the European Space Agency's Near-Earth Object Coordination Centre (NEOCC). The software uses the method of systematic ranging to perform orbit determination on tracklets in the Near-Earth Object Confirmation Page (NEOCP), which typically have short observational arcs. Fitted orbits are propagated to determine the likelihood of an impact with Earth. In addition, magnitude fitting and Monte Carlo sampling are performed to estimate the object's size, possible impact locations and times, and suggest a best telescope pointing for object follow-up. A set of object scores are produced from computed posterior probabilities across the grid, giving a statistical description of the object's orbital and physical characteristics. The scores are packaged with several informative plots in an email alert, which is sent to Meerkat subscribers in the event of a significant impact probability, close approach, or other scientifically interesting event. The highlights of the five years of Meerkat's operational service are presented, including the successful warnings for all of the past six imminent impactors discovered before impact and several interesting close approaches.

Figures (11)

• Figure 1: Systematic ranging plot for object P12cRbG, after 3 observations. The WRMSE, 95% confidence region and Earth-impacting regions are indicated. In addition, the borders for unbound-heliocentric solutions and bound-geocentric solutions are shown as dotted and dashed blue lines respectively.
• Figure 2: Schematic of the Meerkat systematic ranging module. An initial attributable $\mathbf{A}_{ij}^{\text{init}}$ is fitted recursively to the input observations for points of fixed topocentric range and range-rate values $\{ (\rho_{i}, \dot{\rho}_{j}) \}$. The grid of best-fit attributables $\mathbf{A}_{ij}$, along with their likelihood $\mathbf{\Delta}_{ij}$ and Keplerian elements $\mathbf{Y}_{ij}$ are then used to compute absolute magnitudes $\mathbf{H}_{ij}$ and impact (or close approach) locations and times $\mathbf{K}_{ij}$. Several scores are computed, estimating the object size and orbital class. Monte Carlo samples are also drawn. Three files are outputted as indicated in yellow, while teal boxes indicate stages where multiprocessing is used.
• Figure 3: Schematic of the Meerkat data pipeline. The NEOCP is continually checked for new or updated observations. Once found, observations are processed and their objects put in a prioritised queue based on their last computed impact scores. Each object in the queue is analysed sequentially using the systematic ranging module (Fig. \ref{['fig:sysranging_fc']}), with the outputs used to compute a best pointing and ephemeris via the sample propagator (Section \ref{['subsec:detprob']}) and generate informative plots (Sections \ref{['subsec:meerkat-dashboard']} and \ref{['subsec:other-plots']}). If a given object meets the threshold criteria, an alert is issued to Meerkat subscribers (Section \ref{['sec:meerkat-alerts']}). If no updated observations are found, object ephemerides may be updated if older than two days. Boxes in teal represent where multiprocessing is used.
• Figure 4: Dashboard of a Meerkat alert for NEOCP object CAQTDL2 after 4 observations (later designated 2024 RW1).
• Figure 5: Station selector plot for the object K08T03C (later designed 2008 TC3) after 8 observations. The full station selector plot includes another column of contour plots to show the observability for several MPC stations.