Accurate drag modelling is crucial for satellite operations and re-entries in low-Earth orbit (LEO). A proper treatment of this problem becomes increasingly important as the number of artificial satellites grows. However, atmospheric drag modelling remains as a current major challenge, mostly due to uncertainties related to the atmospheric density (ρ) and the drag coefficient (CD). The fact that the latter magnitudes cannot be disentangled from the product ρCD adds difficulty to the problem. Recent research has led to the development of new models and tools aimed to accurately determining the CD. Simultaneously, high-fidelity satellite geometry determination plus on-board measurements of their accelerations has recently pushed forward our knowledge of the physical properties of the upper thermosphere. In this thesis, we propose and test two methodologies to study satellite aerodynamics using two-line element orbital data (TLE) from real satellites and the NRLMSISE-00 atmospheric model. The KnoCD (”Known” CD) method is applied to data from 56 1U-CubeSats and two ANDE-2 spherical satellites. The CalCD (”Calculated” CD) method, is applied to 12 3U-CubeSats of the Dove constellation of Planet Labs. In the KnoCD approach, the ballistic coefficient BC was estimated by computing the average of the crosssection area of a 1U-CubeSat (or the cross-sectional area of spherical satellites). In the CalCD approach the code SPARTA and the TLEs were used to estimate BC in each period of each satellite. In both cases, the TLEs were used again to propagate the orbits using the simplified perturbation model for orbital space vectors SGP4, and then we integrated the appropriate differential equation to compute the average density between consecutive TLEs. Finally, using KnoCD the yearly average density deviation of the NRLMSISE-00 was computed. With CalCD a statistical analysis was done in order to find the energy accommodation (related to the CD) that best fitted the satellite data. The results from KnoCD show a deviation of the thermospheric density of -5.5% per decade, well in accordance with the literature. With CalCD, our preliminary results showed that not all the satellites adjust equally and filtering should be applied. We performed two simulations with different adjustments in the derived BC and showed that the energy accommodation is in the range of 0.8-0.9, well in accordance with the values found in former studies which used onboard data. Our results, therefore, confirm the potential use of small satellite data to effectively assess the thermospheric density and satellite aerodynamics. In addition, we quantify the relevance of high precision in attitude data in order to obtain reasonably accurate drag information.
gmail.com
