The GPS data from Challenging Mini-satellite Payload (CHAMP) is used for its orbit determination for the epoch day of January 1st 2002.  The orbit of CHAMP is computed from the GPS data and ionospheric effects are removed by frequency combination. Further, the orbits of CHAMP for the same epoch day are computed using the satellite tool kit (STK) employing simplified general perturbations (SGP4) and a high precision orbit propagator (HPOP). Results from both techniques (GPS computed orbit and STK computed orbit) are compared. Furthermore, orbits computed using GPS data are also compared with jet propulsion laboratory’s published CHAMP spacecraft orbit and we have found that root mean square difference in ECEF position X component is below 0.01km other than some spikes at poles. The standard deviation of the difference in ECEF position X coordinate is 11.7m. The accuracy of our computed satellite positions (using GPS data) is about 12 metres for other than polar areas. However there are some occasional spikes, especially at poles, having maximum errors (about 0.055 km).

GPS data from Challenging Mini-satellite Payload (CHAMP) is used for its orbit determination for the epoch day of January 1st 2002.  The orbit of CHAMP is computed from the GPS data and ionospheric effects are removed by frequency combination. Further, the orbits of CHAMP for the same epoch day are computed using the satellite tool kit (STK) employing simplified general perturbations (SGP4) and a high precision orbit propagator (HPOP). Furthermore, orbits computed using GPS data are also compared with jet propulsion laboratory’s published CHAMP spacecraft orbit and we have found that root mean square difference in ECEF position X component is below 0.01km other than some spikes at poles. The standard deviation of the difference in ECEF position X coordinate (JPL results – GPS computed results) is 11.7m. Since JPL computed orbits are considered as true orbits of CHAMP with accuracy of centimeter level (https://gipsy-oasis.jpl.nasa.gov/). Therefore this difference can also be referred as observed error in GPS computed orbits. Considering above discussion, we can expect that accuracy of our computed satellite positions (using GPS data) is about 12 metres for other than poles area. However there are some occasional spikes, especially at poles, having maximum errors (about 0.055 km).

This graduate project compares legacy simplified general perturbation model (SGP4) code developed by NASA Johnson Space Center, Orbital Debris Program Office, to a recent public release of SGP4 code by David Vallado. The legacy code is a subroutine in a larger program named PREDICT, which is used to predict the location of orbital debris in GEO. Direct comparison of the codes showed that the new code yields better results for GEO objects, which are more accurate by orders of magnitude (error in meters rather than kilometers). The public release of SGP4 also provides effective results for LEO and MEO objects on a short time scale. The public release code was debugged and modified to provide instant functionality to the Orbital Debris Program Office. Code is provided in an appendix to this paper along with an accompanying CD. A User’s Guide is presented in Chapter 7.

<p>Our modern society is more dependent on its assets in space now more than ever. For<br> example, the Global Positioning System (GPS) many rely on for navigation uses data from a<br> 24-satellite constellation. Additionally, our current infrastructure for gas pumps, cell phones,<br> ATMs, traffic lights, weather data, etc. all depend on satellite data from various constel-<br> lations. As a result, it is increasingly necessary to accurately track and predict the space<br> domain. In this thesis, after discussing how space object tracking and object position pre-<br> diction is currently being done, I propose a machine learning-based approach to improving<br> the space object position prediction over the standard SGP4 method, which is limited in<br> prediction accuracy time to about 24 hours. Using this approach, we are able to show that<br> meaningful improvements over the standard SGP4 model can be achieved using a machine<br> learning model built based on a type of recurrent neural network called a long short term<br> memory model (LSTM). I also provide distance predictions for 4 different space objects over<br> time frames of 15 and 30 days. Future work in this area is likely to include extending and<br> validating this approach on additional satellites to construct a more general model, testing a<br> wider range of models to determine limits on accuracy across a broad range of time horizons,<br> and proposing similar methods less dependent on antiquated data formats like the TLE.</p>