The number of objects in a geocentric orbit is constantly increasing, and so is the number of objects reentering. While tracking and collision avoidance are crucial in guaranteeing space safety, another significant task is correctly predicting satellites' reentries and overall decay to anticipate potential impacts and intersections with spacecraft trajectories - this knowledge about the state of the reentries strongly correlated with the state of the atmosphere. The European Space Agency (ESA) developed the flight dynamics library godot to optimize and propagate orbits. In this work, we redesign godot's atmospheric core component using state-of-the-art software engineering in modern C++ while interfacing a dynamic plugin that provides the user with an array of new and legacy atmospheric models, such as Jacchia-Bowman 2008 (JB2008), MSISv2, or DTM2020. In the second step, we build upon godot by implementing an atmospheric optimization framework. The framework includes an extensive toolchain for the purpose of atmosphere and reentry optimization consolidating publicly available data sources like Space-Track, Discos, and the HASDM (High Accuracy Satellite Drag Model) SET Database to perform fully automatic reentry prediction for arbitrary satellites by optimizing their ballistic coefficient and the diurnal density coefficients of JB2008.
The optimization framework allows a high degree of freedom in assembling and combining arbitrary optimization building blocks with one another. With a modified version of Picone et al.'s TLE density derivation algorithm to optimize ballistic coefficients, we can achieve median reentry prediction errors as low as ~9% in 2019 for 25 satellites, comparable to the current state-of-the-art errors with only publicly available resources and within runtimes of minutes by exploiting parallelization. The solution will be utilized by the ESA's Space Debris Office as a second toolchain for assessing reentries.