SOME SPECIAL PROBLEMS IN STATISTICAL INFERENCE ASSOCIATED WITH SATELLITE TRACKING

Abstract

Listen

Translate article

Abstract : Satellite tracking systems (notably optical and doppler) have been developed to such a degree that the accuracy of orbit determination is no longer limited by the accuracy of the tracking data itself but is almost wholly limited by other parameters associated with the satellite motion and tracking system. The principal parameters limiting this accuracy are those associated with the earth's gravitation force field and the locations of the satellite tracking stations. Significant improvement in the values of a few of these parameters has already been made via studies of satellite motion but, in most cases, they have been made without the use of statistical inference techniques. The Navy's doppler tracking system provides ample high quality data for statistically inferring very accurate values of these parameters. However, employing inference techniques to tracking data obtained from such a system gives rise to some unusual problems. In particular the validity of inference techniques depends very strongly on correctly estimating the character of and correlations in the experimental tracking data. This paper will discuss briefly some of the more difficult problems associated with specifying the errors in the data, the consequent problems in determining optimum statistical weighting factors, and finally the difficulties in realistically estimating the probable errors in the inferred values for the parameters.