Processing and scientific analysis of data from BepiColombo’s SERENA space experiment towards an advanced ground system

In the search of new knowledge and understandings of our Solar System and the Universe surrounding us, we send out planetary exploration missions, one of which is BepiColombo - the ESA/JAXA flagship to Mercury, armed with a rainbow of observation instruments. The data provided by these instruments should present the scientific world with reliable observational information and aim to reveal the hidden patterns and mysteries of the closest planet to our Sun. To aid the fulfillment of these scientific goals, this study focuses on the application of practical, theoretical and conceptual methods for data processing and analysis within the Ground System (GS) of SERENA (Search for Exospheric Refilling and Emitted Natural Abundances) - a suite of four particle detectors on-board the Mercury Planetary Orbiter (MPO). The four main aspects of this work are: (1) the design and development of the data processing and archiving pipelines of ELENA (Emitted Low-Energy Neutral Atoms) - one of the SERENA sub-units, and of SCU (System Control Unit) - the common SERENA control processor; (2) the subsequent integration of all four single sub-unit software applications into a common SERENA data processing and archiving package; (3) the elaboration of novel methods for analysis of exospheric data through the utilization of machine learning techniques; and (4) the conceptual design of an advanced architecture of the Ground System of a space experiment. The successful integration of the ELENA/SCU Processing Pipeline and the SERENA Processing Pipeline in both the premises of the SERENA PI (Principal Investigator) and the ESA SGS (Science Ground Segment) is supported by the utilization of techniques for software dependability and portability and by secure automated interfaces between the two. The software’s functionality to translate telemetry (both science and housekeeping) to raw data, and then to calibrated data, and subsequently to prepare them in the valid PDS4 format for ingestion into ESA’s Planetary Science Archive (PSA) is accomplished by robust C++ and Bash procedures that utilize geometry and spacecraft information from the common SPICE information. At the analysis level, a state-of-the-art deep neural networks (DNNs) are devised to relate the exospheric data from the SERENA sub-units to the surface composition and to the interaction with the space environment. Those algorithms show the capability to very accurately reconstruct the mineral and elemental composition and show the potential to constrain the models for the processes that generate the exosphere, when the BepiColombo nominal planetary mission phase begins in 2026 in orbit around Mercury. The DNNs are incorporated in a prototype software for analysis of data from the SERENA sensors - PipeTheFirst (PTF). Finally, a Conceptual SERENA GS is designed, which sets up the future standards for a ground system architecture, through the use of artificial intelligence enhancements, modern visualization tools, support for tools for science planning and telecommanding, complete utilization of external resources, and interfaces between the system and the knowledge analyst. The development of these features would result in an easily maintainable automated system for a full quasi-real-time utilization of the data flow from one of the most important ESA missions in the coming years, which would potentially be reused for scientific knowledge extraction in other future planetary missions.