**1. Introduction**

The best systems with the exact impact solutions for dangerous asteroids are presented by the JPL Sentry System: http://neo.jpl.nasa.gov/risk/ and by the NEODyS CLOMON2: http://newton.dm.unipi.it/neodys/index.php?pc=4.1

From many years on the top of these lists were two asteroids: (99942) Apophis (is still up now, October 2011) and (144898) 2004 VD17 – now is removed from the list of the dangerous asteroids. Thanks to the courtesy of those who made free available OrbFit software and its source code at: http://adams.dm.unipi.it/~orbmaint/orbfit/

It is now possible to compute individually dates of possible impacts of selected dangerous asteroids or the energy of impact and others impact factors. In this respect we investigated the motion of these recently discovered minor planets: (99942) Apophis and (144898) 2004 VD17 - the most dangerous for the Earth, according to the Impact Risk Page of NASA: http://neo.jpl.nasa.gov/risk/.

To compute exact impact solutions of asteroids it is necessary to include some additional small effects on the asteroid's motion. The inluence of relativistic effects, the perturbing massive asteroids, the Yarkovsky/YORP effects, solar radiation pressure, different ephemeris of the Solar System were investigated. To compute gravitational forces perturbing the motion of (99942) Apophis and (144898) 2004 VD17 from different massive asteroids, the free software Solex from A. Vitagliano was used: http://chemistry.unina.it/~alvitagl/solex/.

SOLEX computes positions of the Solar System bodies by a method which is entirely based on the numerical integration of the Newton equation of motions (Vitagliano, A. 1997). With the use of Solex it was possible to compute all close approaches between (99942) Apophis and (144898) 2004 VD17 with all nearly 140000 numbering asteroids. Similar work with (15) Eunomia using Solex was done by Vitagliano and Stoss (2006).

Selected orbit solutions for (99942) Apophis and (144898) 2004 VD17 were presented during Meeting on Asteroids and Comets in Europe - May 12-14, 2006 in Vienna, Austria. At that time the new version of OrbFit (3.3.2) was released and gave better results of computations of impact probability mainly with the use of non linear monitoring and multi ple solutions method (Milani et al., 2002, Milani et al., 2005a and Milani et al.,

OrbFit Impact Solutions for Asteroids (99942) Apophis and (144898) 2004 VD17 61

computed solutions by author of this paper; *nr* denotes solution without radar observations and equal to *sigma\_LOV* - approximate location along the LOV in sigma space; values of sigma are usually in the interval [-3,3] which represent 99.7 % probability of occurrence of real asteroid in this confidence region (Milani et al. 2002). The impact probability is not reported if the computed value is less than 1E-11. The presented

of value we observe slightly different impact solutions mainly in the date of possible impact. The differences between the results from the NEODyS (CLOMON2) and the JPL NASA (SENTRY) are evident because they are independent systems as state at: http://neo.jpl.nasa.gov/risk/doc/sentry\_faq.html. For example impact probabilities

Fig. 1. The orbit of (99942) Apophis projected to the ecliptic plane, where *x*-axis is directed to

only the input data in OrbFit software, not the real

different by a factor of ten or so are not extraordinary.

= 3 denotes that the real

LOV. For example

vernal equinox.

are


is between -3 and +3. For different setting

2005b). The main goal of our work was to compare our results generated by OrbFit with the results presented by CLOMON2 system which uses the same OrbFit software and with the results of JPL NASA SENTRY. The second purpose was to prove how differently small effects in motion of asteroid change impact solutions. It was possible thanks to public available source code of the OrbFit software. The orbital uncertainty of an asteroid is viewed as a cloud of possible orbits centered on the nominal solution, where density is greatest. This is represented by the multivariate Gaussian probability density and the use of this probability density relies on the assumption that the observational errors are Gaussian (Milani et al., 2002). Now, August 2011, we have new version of the OrbFit software, v.4.2, implementing the new error model based upon Chesley, Baer and Monet (2010).
