Search

"Space Navigation and Guidance," taught every fall at the University of Maryland, is required of all space track undergraduate aerospace engineering majors. Every student is required to participate in a group project where real observations are used in the solution of a navigation problem with estimation from observations. In this paper, I discuss two such projects, an observatory project in which the students use a telescope to track a satellite and determine its orbit, and a GPS project in which they analyze GPS receiver data to determine the receiver's position.

The Lie transfer map method may be applied to orbit propagation problems in celestial mechanics. This method, described in another paper, is a perturbation method applicable to Hamiltonian systems. In this paper, it is used to calculate orbits for zonal perturbations to the Kepler (two-body) problem, in both expansion in the eccentricity and closed form. In contrast with a normal form method like that of Deprit, the Lie transformations here are used to effect a propagation of phase space in time, and not to transform one Hamiltonian into another.

Using the elimination of the parallax followed by the Delaunay normalization, we present a procedure for calculating a normal form of the main problem (J2 perturbation only) in satellite theory. This procedure is outlined in such a way that an object-oriented automatic symbolic manipulator based on a hierarchy of algebras can perform this computation. The Hamiltonian after the Delaunay normalization is presented to order six explicitly in closed form, that is, in which there is no expansion in the eccentricity. The corresponding generating function and transformation of coordinates, too lengthy to present here to the same order; the generator is given through order four.

In this paper, we describe the automatic rendering of expressions computed using symbolic manipulation. Computations from astrodynamics frequently can be put in a fixed hierarchy of polynomials and Fourier series. Once in this form, FORTRAN subprograms can be generated automatically in a form that lends itself to numerical evaluation. The goal of the current work is to present an approach for using symbolic manipulation techniques to produce a Fortran representation of the normalized Hamiltonian and other supporting equations representing as many of the actual physical effects on satellites as possible.

The variable-step Stormer-Cowell integrator is a non-summed, double-integration multi-step integrator derived in variable-step form. The method has been implemented with a Shampine-Gordon style error control algorithm that uses an approximation of the local error at each step to choose the step size for the subsequent step. In this paper, the variable-step Stormer-Cowell method is compared to several other multi-step integrators, including the fixed-step Gauss-Jackson method, the Gauss-Jackson method with s-integration, and the variable-step single-integration Shampine- Gordon method, in both orbit propagation and orbit determination. The results show the variable-step Stormer-Cowell method is comparable with Gauss-Jackson using s-integration, except in high drag cases where the variable-step Stormer-Cowell method has an advantage in speed and accuracy.

The Gauss-Jackson multi-step predictor-corrector method is widely used in numerical integration problems for astrodynamics and dynamical astronomy. The U.S. space surveillance centers have used an eighth-order Gauss-Jackson algorithm since the 1960s. In this paper, we explain the algorithm including a derivation from first principals and its relation to other multi-step integration methods. We also study its applicability to satellite orbits including its accuracy and stability.

A variable-step double-integration multi-step integrator is derived using divided differences. The derivation is based upon the derivation of Shampine-Gordon, a single-integration method. Variable-step integrators are useful for propagating elliptical orbits, because larger steps can be taken near apogee. As a double-integration method, the integrator performs only one function evaluation per step, whereas Shampine-Gordon requires two evaluations per step, giving the integrator a significant speed advantage over Shampine-Gordon. Though several implementation issues remain, preliminary results show the integrator to be effective.

In this paper the fixed step Gauss-Jackson method is compared to two variable step integrators. The first is the variable step, variable order Shampine-Gordon method. The second is s-integration, which may be considered an analytical step regulation. Speed tests are performed for orbit propagation with the integrators set to give equivalent accuracy. The integrators are also tested for orbit determination, to determine the speed benefit of the variable step methods. The tests give an indication of the types of orbits where variable step methods are more efficient than fixed step methods.

Two atmospheric density model families that are commonly chosen for orbit determination and propagation, Jacchia and MSIS, are compared for accuracy. The Jacchia 70 model, the MSISE-90 model, and the NRLMSISE-00 model may each be used to determine orbits over fitspans of several days and then to propagate forward. With observations kept over the propagation period, residuals may be computed and the accuracy of each model evaluated. We have performed this analysis for over 4000 cataloged satellites with perigee below 1000km for September-­October 1999, and the 60 HASDM calibration satellites with a large observation set for February 2001. The purpose of this study is to form a picture of the relative merits of the drag models in a comprehensive view, using all satellites in a manner consistent with the operational practice of US space surveillance centers. A further goal is to refine this knowledge to understand the orbital parameter regions where one of the models may be consistently superior.

A generalized Sundman transformation dt = cr^n ds for exponent n >= 1 may be used to accelerate the numerical computation of high-eccentricity orbits, by transforming time t to a new independent variable s. Once transformed, the integration in uniform steps of s effectively gives analytic step variation in t with larger time steps at apogee than at perigee, making errors at each point roughly comparable. In this paper, we develop techniques for assessing accuracy of s-integration in the presence of perturbations, and analyze the effectiveness of regularizing the transformed equations. A computational speed comparison is provided.