Orbit gravity error covariance

Optimal orbit determination requires a physically connected method to calculate an acceleration error covariance function to drive the filter time-update across each propagation time interval. For gravity modeling errors we derive the orbit covariance function from an acquired covariance matrix on potential function coefficient estimation errors. We employ existing theory, derived from results due to Kaula, Pechenick, and Wright to calculate a double integral in six-dimensions, the gravity orbit covariance function for the filter. Our existing method has been limited to a single potential function, and our algorithm architecture has been limited to a restricted LEO class. Now we have constructed a new method that broadens the algorithm architecture, with applications to EGM96, Lunar Prospector, and GRACE potential functions.