Early orbit determination using U-D covariance propagation filter

Techniques for determining the short-arc orbit of a launch vehicle from single-station tracking data are reviewed, and an algorithm based on the U-D covariance factorization scheme of Bierman and Thornton (1977) is derived and demonstrated. Consideration is given to weighted-least-squares, Kalman-filter, and square-root-filter approaches. The model, algorithm design, input data, and computational steps of the U-D method are listed, and the results of numerical experiments involving 150 s of data (sampled once per second) on a nominally 400-km orbit mission are presented in graphs and tables. Rapid reduction is obtained in the position error, but the velocity-error reduction is found to be slow.