Abstract
This paper presents the mathematical formulation and implementation techniques used in the GOES orbit and attitude determination (OAD) software. Also presented are some recent results obtained by analyzing data from the operational GOES. The OAD software, a key component of the orbit and attitude tracking subsystems, has been in use for operating the GOES spacecraft successfully since the beginning. Three kinds of observables are ingested by OAD: range, star line-of-sight direction, and landmark line-of- sight direction. Raw observations are preprocessed to rid of known systematic errors. Among the solve-for parameters are six orbital elements, dynamic model constants, range bias, and a number of instrument attitude coefficient. The attitude of either the imager or the sounder instrument is characterized by five angles: roll, pitch, yaw plus two rotation angles specifying the misalignment of the optical axis of the instrument. Each attitude angle is modeled as the sum of Fourier terms and optional polynomial and exponential terms. The original version of the software was implemented as a least squares batch filter, ingesting observations in batches of one to two days at a time. A recent enhancement to the software to include a priori knowledge of the orbit and attitude solution and their statistical properties has been implemented. By reprocessing operational data using this new software, it is shown that the sequential batch solution obtained exhibits better daily stability as well as providing a better model for next day predictions compared to the batch solution.
