High precision predictions for near-Earth asteroids: the strange case of (3908) Nyx

Abstract

In November 2004 radar delay measurements of near-Earth asteroid (3908) Nyx obtained at the Arecibo radio telescope turned out to be $$7.5\sigma $$ away from the orbital prediction. We prove that this discrepancy was caused by a poor astrometric treatment and an incomplete dynamical model, which did not account for nongravitational perturbations. To improve the astrometric treatment, we remove known star catalog biases, apply suitable weights to the observations, and use an aggressive outlier rejection scheme. The main issue related to the dynamical model is having not accounted for the Yarkovsky effect. Including the Yarkovsky perturbation in the model makes the orbital prediction and the radar measurements statistically consistent by both reducing the offset and increasing the prediction uncertainty to a more realistic level. This analysis shows the sensitivity of high precision predictions to the astrometric treatment and the Yarkovsky effect. By using the full observational dataset we obtain a $$5\sigma $$ detection of the Yarkovsky effect acting on Nyx corresponding to an orbital drift $$da/dt = (142 \pm 29)$$ m/year. In turn, we derive constraints on thermal inertia and bulk density. In particular, we find that the bulk density of Nyx is around 1 g/cm $$^3$$ , possibly less. To make sure that our results are not corrupted by an asteroid impact or a close approach with a perturbing asteroid not included in our dynamical model, we show that the astrometry provides no convincing evidence of an impulsive variation of Nyx’s velocity while crossing the main belt region.