| dc.description.abstract | The errors in current atmospheric drag modeling are the primary source of error for orbit determination for objects in low Earth orbit (LEO) at lower altitudes in periods of high solar activity. This is a direct result of significant advancements in conservative force modeling in the form of high accuracy geopotential models. When these new geopotential models are applied to orbit determination packages, the majority of the error source shifts to non-conservative forces such as solar radiation pressure, Earth albedo, Earth infrared (IR), and atmospheric drag. During periods of high solar activity, the density of the atmosphere is highly variable due to interactions with the Sun and the upper atmosphere. These variations are very difficult for empirical density models to estimate and cause significant errors in deriving precise orbits. For this reason, increasing the accuracy and fidelity of atmospheric density models is crucial in order to further increase the accuracy of orbit determination during these times. If equipped, on-board accelerometers can provide measurements of the non-conservative accelerations that a spacecraft encounters along its orbit. A very accurate approximation of the force on a spacecraft due to atmospheric drag can be found by accounting for all other non-conservative forces and considering the remainder to be drag. Accuracy is reduced when using that force to find the density of the atmosphere due to the nature of the drag equation. The drag coefficient (CD) is used to balance the acceleration due to drag and the density of the atmosphere. Determining the value of the drag coefficient is arduous for most applications. To make the process easier, the projected area (A) and mass (m) terms in the drag equation are often lumped together with CD to form what is called the ballistic coefficient, CDA/m. This is actually the inverse of the traditional definition of the ballistic coefficient. By using this term, the uncertainties in the projected area and mass can be lumped in with the drag coefficient. Approximating the ballistic coefficient using two line element sets (TLE) was one objective of this research. TLEs are based on radar and optical observations and are thus are not nearly as accurate as other tracking methods, but are advantageous because of the multitude of satellites cataloged over the last several decades. The method of ballistic coefficient estimation presented here can be used quickly and without significant resources since TLEs are widely available. The results of this study indicate the ballistic coefficients generated for spacecraft in orbits less than 500 km in the 2001-2004 time period were within 8.2% of ballistic coefficients derived from analytical methods when using the analytical ballistic coefficients as truth. For satellites around 800 km or above, the ballistic coefficients generated were over 100% from those derived using analytical methods. Creating corrections to existing density models has become a popular way of capturing these variations. Several techniques have been devised to generate these corrections in the past few decades. This thesis utilizes corrections to the NRLMSISE-00 empirical model of the atmosphere generated using the dynamic calibration of the atmosphere (DCA) technique. These corrections, along with the NRLMSISE-00 empirical model, are implemented into GEODYN, NASA's precision orbit determination and parameter estimation program, to create three GEODYN versions; an unmodified GEODYN with the MSIS-86 atmospheric model, a version using the NRLMSISE-00 model, and a version using the DCA corrections. Any improvements using these new density routines will provide a direct benefit to orbit estimation which, in turn, improves science data. In this thesis, the GEOSAT Follow-On (GFO), Starlette, Stella, and Geo-Forschungs-Zentrum-1 (GFZ-1) satellites were processed with the three versions of GEODYN in the waning of the most recent solar maximum. The results show that the NRLMSISE-00 empirical model can capture slightly more variations in the atmosphere than the previous MSIS-86 model, especially in high solar activity conditions. The DCA corrections produced results similar to the NRLMSISE-00 model, but after an investigation into the drag coefficient estimated through GEODYN, a more detailed investigation is necessary to determine the validity of these results. This is likely due to the altitude or time period of the satellites chosen for processing being outside the range of the DCA corrections to the NRLMSISE-00 routine. | |
