Merging gated frequency-modulated continuous-wave Mars2020 RIMFAX ground-penetrating radar data

Abstract

The integration of ground-penetrating radar data at various frequencies collected with different antennas or with the use of swept-frequency radars opens up interesting perspectives in the study of the subsurface at different resolutions. Our methodology is a semisupervised deep learning algorithm based on bidirectional long short-term memory to automatically merge varying numbers of data sets at different frequencies. Neural network (NN) training is done directly on the inference data by minimizing a custom loss function based on the L2 norm of all the input data, weighted on the custom merging area, and the single output trace. The inference of the trained NN is applied to the same data. Our algorithm is tested on synthetic data simulating the Mars conditions and on Radar Imager for Mars’ Subsurface Exploration radar data collected in the Jezero crater during the Mars2020 mission of the Perseverance rover, showing successful performances and robustness.