Advanced Signal Processing Methods for Planetary Radar Sounders Data

Radar sounders are spaceborne electromagnetic sensors specifically designed for subsurface investigations. They operate in the HF/VHF part of the electromagnetic spectrum and are widely employed for applications such as monitoring changes to the polar ice sheets of the Earth and for the study of planetary bodies (e.g. Mars) from satellite. Radar sounding of planetary bodies is a relatively young discipline both in terms of system design and data processing architectures. As a result of the current state of the art in system design, the data recorded by radar sounders are typically affected by artifacts, such as off-nadir surface clutter, which hinders its interpretation by scientists. On top of that, the analysis of the very large of amount of data produced by such systems is typically performed manually by experts thus inherently subjective and time-consuming. Therefore the development of automatic high-level processing strategies for reliable, objective and fast extraction of information is needed. Accordingly, this thesis work deals with different aspects of radar sounding namely system design, low-level and high-level processing. The thesis provides three main novel contributions to the state of the art. First, we present a study on system design, performance assessment and 3D electromagnetic simulations of a radar sounding system specifically tailored for detecting lava tubes under the Moon surface. Lava tubes are considered to be important and useful structures. By having a stable temperature and by providing protection against cosmic ray radiation and micrometeorites impacts, they could potentially serve as natural shelter for human outposts on the Moon. The results presented in this thesis show that a multi-frequency radar sounder is the best option for effectively sound most of the lava tube dimension expected from the literature and that they show unique electromagnetic signature which can be used for their detection. The second novel contribution is focused on low-level processing and consists in a bio-inspired clutter detection model based on bats echolocation. Very relevant analogies occur between a bat and radar sounder such as the nadir acquisition geometry. The mathematical model proposed in this thesis adapts the bats frequency diversity strategy (i.e. multi-frequency approach) to solve clutter ambiguities to the radar sounding case. The proposed bio-inspired clutter detection model has been tested and validated on experimental data acquired over Mars. The experimental results showed that the method is able to discriminate in a precise way the radar echoes arising from subsurface targets with respect to off-nadir surface clutter ones. The third novel contribution of this thesis goes in the direction of high-level processing and in particular of automatic data analysis for accurate and fast extraction of relevant information from radar sounding data. To this extent, we propose a novel automatic method for retrieving the spatial position and radiometric properties of the subsurface layers based on Hidden Markov Models for radar response modeling and the Viterbi Algorithm for the inference step. Furthermore, a novel radargram enhancement and denoising technique has been developed to support the detection step. The effectiveness of the technique has been demonstrated on different radargrams acquired over the North Pole of Mars pointing out its superiority with respect to current state of the art techniques.