Effects of close approach dynamics on asteroid orbit determination

Abstract

Asteroid orbit determination is the process of fitting astrometric observations to estimate osculating orbital elements using least squares methods. The degree of quality of the fit is assessed through residuals, which are influenced by factors such as observation accuracy, the applied weighting scheme, and the adopted dynamical model. Reliable asteroid orbits are critical for planetary defence, particularly in assessing impact risks with Earth. An analysis of the orbital database maintained by the Near-Earth Object Coordination Centre (NEOCC) of the European Space Agency (ESA) highlighted a particular subset of orbits with anomalously high residuals, here defined broken orbits. These anomalies may lead to the hypothesis of the presence of unmodelled impulsive events within the observational arc, potentially caused by close encounters with asteroids not included in the dynamical model. The aim of this research is to investigate whether flybys between two asteroids, when not properly accounted for in the dynamical model, contribute to such discrepancies, and whether extending the dynamical model to include additional perturbing asteroids could improve orbit determination. To this aim, a statistical characterisation of the NEOCC residual database is initially performed, followed by identification of non- dynamical and dynamical factors influencing the orbit determination process. The former focuses on revising the weighting scheme by identifying observer-dependent biases linked to asteroid dynamical classes. The latter include improved mass estimates for known perturbers and the study of the effect of a flyby with an unmodelled object on the residuals. This is performed using numerical simulations of asteroid- asteroid flybys, orbital propagation and generation of synthetic observations. Results demonstrate that the unmodelled dynamics of flybys between asteroids shall be discarded as the cause of the broken orbits residual anomalies. However, the dynamical model can still be improved trough proposed refinements in mass estimation and weighting schemes.