Most onboard guidance, navigation, and control computers will be programmed using suboptimal filter design in an effort to reduce software size and computation speed. Reduction in the order of the gravity-field model of the moon is one method that can be used to reduce the number of calculations the onboard filter must perform, thus increasing the speed of computation. A result of that reduction is an increase in navigation error, which in turn directly influences the accuracy of the guidance commands. This paper investigates two prevalent lunar gravity-field models-the Ferrari 79 and the Sagitov model. A brief history of their origins and a comparison of the models are given. Particular emphasis is placed on studying their effect on guidance and navigation accuracy as the order of the onboard navigation filter model is reduced, resulting in a suboptimal filter design.
NomenclatureAT, A N = truth and navigation state gradient matrices Alt. = altitude, m q, = semimajor axis, m Cnn = gravitational model constants CR = crossrange, m DR = downrange, m e = eccentricity vector G -sensor model measurements, m Hj, H N = truth and navigation observation update matrices h = angular momentum vector, m 2 /s 2 K F= navigation state Kalman-filter gain P r , P N = truth and navigation covariance matrices before navigation update PT, PN = truth and navigation covariance matrices after navigation update Q = unknown state noise matrix R = navigation measurement noise matrix r -inertial radial vector, m r p = radius of periapsis, m v -velocity magnitude, m/s 2 WT , WN = truth and navigation weighting matrices (identity matrix for this study) XT, X N = truth and navigation state vectors, m // = gravitational parameter, m 2 /s 3 4> r , & N = truth and navigation state transition matrices x = vector cross product