Effective covariance deweighting for orbit estimation

It is noted that sequential orbit determination techniques have failed to produce appreciably better results than the old batch least square techniques. This is thought to derive from the assumption that force modeling errors satisfy the white noise property assumed in Kalman filter treatment. Attention is also given to the techniques based on sequential estimation using time-correlated process noise models that have been developed and shown to improve state and state covariance predictability. The approach to sequential estimation being planned for improving the data system at the U.S. Air Force's Sunnyvale Satellite Test Center is described. In addition, test-bed results using simulated data taken for a low altitude orbiter are presented, indicating the nature and magnitude of the improved performance resulting from the proposed estimator.