A convolutional neural network for analysis of active ionospheric sounding data from the Mars Express MARSIS instrument

ABSTRACT We present a method for the automatic classification of ionograms using a convolutional neural network which can detect the ionospheric trace. The ionograms are from the Mars Advanced Radar for Sub-surface and Ionosphere Sounding (MARSIS) instrument on board the Mars Express orbiter which has been observing the Martian ionosphere since June 2005, and has produced more than two million ionograms, with more data still being collected as of 2025. The process of designing, training, and testing the classifier is described. We also discuss the results of applying the CNN to the complete set of ionograms from the nominal mission phase. The results show that the final version of the classifier is 97 per cent accurate on a wide range of ionograms. Applying the classifier to the entire nominal mission data set shows the potential to use it for studying plasma transport across the terminator and occasional night side detections of a patchy ionosphere.