14–17 Mar 2016
Darmstadtium
Europe/Amsterdam timezone
"Orbiting Towards the Future"

Comparison of the Orekit DSST Short-Periodic Motion Model with the GTDS DSST and the F77 DSST Standalone Models

Not scheduled
20m
Darmstadtium

Darmstadtium

Schloßgraben 1, 64283 Darmstadt, Germany
Oral presentation at the conference

Speaker

Mr Juan Felix San Juan (University of La Rioja)

Description

Development of the DSST started in the mid 1970’s at the Computer Sciences Corporation and continued at the Charles Stark Draper Laboratory and the MIT Lincoln Laboratory. These developments employed the non-singular equinoctial elements. The Draper Semi-analytical Satellite Theory used the GTDS orbit determination system as the development platform. However, users external to the Draper Laboratory wanted access to the Semi-analytical Satellite Theory without the ‘overhead’ of GTDS. The DSST Standalone program was developed in 1983-84. The Standalone included complete models for the mean element motion (based on the conventions then employed in GTDS) and a portion of the short-periodic model. The intent was to provide accuracy for LEO orbits of approximately 200 meters. By 1996, extensive improvements to the GTDS DSST had been made. These included 50 x 50 geopotential fields, solid Earth tides, and J2000 coordinate systems. In 1997, an effort to extensively upgrade the DSST Standalone was undertaken. This effort included improvements to the force modeling and to the maintainability of the source code. While the 1997 upgrade touched large portions of the DSST Standalone source code, testing primarily focused on the mean element equations of motion. In 2010, the first author presented the paper “OPEN SOURCE SOFTWARE SUITE FOR SPACE SITUATIONAL AWARENESS AND SPACE OBJECT CATALOG WORK” at the ICATT meeting in Madrid. This paper proposed the migration of the DSST Standalone Orbit Propagator from Fortran 77 to an Object-Oriented software platform. In 2011, the implementation of the mean element motion portion of the DSST in the Orekit open source library was initiated. Implementation in the Orekit library involved migration of the DSST to the object-oriented java language and to a different functional decomposition strategy. Resolution of the F77 Standalone DSST code and documentation anomalies was an important product. Orekit DSST mean element predictions were compared with those produced with the F77 DSST Standalone. For several test cases involving several thousand day arcs, the Orekit and F77 mean element histories could not be distinguished. The DSST employs Fourier series for the short-periodic motion in the equinoctial elements. These expressions are closed form for the zonal, lunar-solar, and the tesseral m-daily terms. The Fourier coefficients in these expressions are functions of the slowlyvarying mean elements (a, h, k, p, and q) and are slowly varying when plotted over time.1 The Fourier coefficients are computed 'off-grid' via an interpolation process. This Fourier coefficient interpolation must be compatible with both the high order Dormand- Prince and the classical RK integrators employed in Orekit DSST. This paper provides a description of the java classes adopted for the Orekit DSST with emphasis on the short-periodic models and the associated interpolation processes. The Orekit class DSSTForceModel includes the functions getMeanElementRate and getShortPeriodicVariations. This paper provides detailed comparisons of the Fourier coefficients computed by the Orekit DSST, F77 DSST Standalone, and GTDS DSST programs on a perturbation-by-perturbation basis. We also consider the overall accuracy that is possible with the present Orekit DSST implementation.
Applicant type First author

Primary author

Dr Paul Cefola (University at Buffalo)

Co-author

Luc Maisonobe (CS Toulouse)

Presentation materials