Determination des orbites des satellites artificiels

Summary The problem of the determination of the orbit of a celestial body is an old astronomical problem, of which solutions can be found in many classical books on elementary celestial mechanics. However, the introduction of new (radioelectric) means of observations for the artificial satellites have brought up new solutions of this old problem.The author reminds the definitions of the six elliptic elements of an orbit. The problem of their determination is usually divided into two separate steps: a preliminary orbit determination and the improvement of the preliminary elements. Two principal types of preliminary orbit determination exist: the Gaussian type, purely geometrical in which the positions of the body at two different times are determined, and the Laplacian type, of a more dynamical character, in which both position and velocity vectors are found for a given time. The improvement of the preliminary elements is usually obtained by a numerical solution of equations of variations of the elements, minimizing the sum of the squares of the angular distances between the computed and the observed points. The elements on which these variations are applied can be quite various. The coefficients of the equations are usually obtained numerically, although their analytical expressions can also be derived.Most of the modification proposed to the classical methods of orbit determination are more technical improvements in connection with the computation on electronic machines. The most interesting modifications are inspired by the fact that time measurements are less precise than the position (Batrakov, Iszak), the effects of this difference in errors are discussed in this paper. In some cases, using Laplace's method, no preliminary orbit is computed, and all the observations can be used at once (Barlier, Kovalevsky).The determination of orbits from radar measurements is discussed. The proposed methods are quite different in principle from the classical ones. They are actually improvements of a circular orbit whose determination is made by a method of undetermined coefficients (Baker). Some of the elements determined by these methods are quite unstable, and it is always wiser to use other informations together with radar measurements.The determination of orbits from Doppler data alone has been worked out by many investigators. It is the inverse problem of the main problem of all navigational systems. The method proposed by Patton is summarized. The principles of others are quite analogous. The results are improved when more than one receiver are used. The different types of corrections: refraction, perturbations, etc., should be introduced in later stages of the determination, and the whole frequency curve is to be used if a good determination of all elements is desired.Finally, the author quotes a method (Baker) using both Doppler and radar data, and the solution of a new problem: the determination of the orbit of a satellite of another planet from Doppler data alone (Deutsch).