Non linear impact monitoring: Line Of Variation searches for impactors

When a new near Earth asteroid is discovered, it is important to know whether or not there is the possibility of an impact with the Earth in the near future. In this paper, we describe the technical approaches employed by the two operational second-generation asteroid impact monitoring systems, CLOMON2 and Sentry, paying particular attention to the similarities and differences between these independent systems. The detection and characterization of a potential impact requires the propagation of the orbital probability density function from the time of discovery to the time of hypothetical impact. Since the N-body problem is not integrable, this can be done only by sampling the orbital elements space with a finite number of Virtual Asteroids (VAs), the orbit of each one being propagated numerically. Our methods, illustrated in this paper, use the Line Of Variation (LOV), a unidimensional subspace, to perform this sampling. The primary goal is to detect Virtual Impactors (VIs), which are regions in the initial conditions space leading to dynamically distinct collision solutions; then a probability integral needs to be computed on the volume of the VI. An important issue is how to assure completeness of such a search down to some impact probability threshold. This problem cannot be efficiently solved just by computing more VAs, but requires a geometric description of the behavior of the LOV in order to identify the critical segments of this curve. We have studied these behaviors on the Target Plane (TP) through our analytical theory and the output of many numerical tests. Assuming that the geometry is the simplest compatible with the data available from the sampling, we obtain a classification which allows us to use iterative methods, appropriate for each case, to find the closest approach distance possible along the LOV. After an LOV minimum has been identified, it is possible to use a probability density linearized at this point. However, when the cross section of the Earth is not crossed by the LOV, there is no guarantee that nonlinearity would be negligible in the direction on the TP transversal to the LOV. We describe how to test for such nonlinearity, and thus reduce or eliminate the possibility of spurious VIs. In this way, the performance of our impact monitoring systems has been significantly increased in comparison to the earlier and simpler solitary system. These more advanced systems have identified and then eliminated (through additional observations) more than one hundred cases of asteroids with VIs in the years 2002 2003.