Logiciel de restitution de trajectoire de satellite-application au cas de satellites d'observation

This paper presents an orbit determination program based on two algorithms: Kalman filtering and Rauch smoothing. Novel adaptation of these algorithms lets obtain a posteriori statistical estimations of evolution and measurement noises and of their standard deviation. Analysis of results about these noise estimations lets regulate optimally the a priori levels of noise standard deviations and detect eventually model errors. Such use is presented for Earth observation satellites, at first with simulated data, then with real data (TIROS N satellite).