On Mars, mounds and cones can be formed through mud extrusion and periglacial processes such as freezing lenses of water (pingo). The mapping of large populations of mound features on imagery with convolutional neural networks (CNNs) and ML algorithms have already been proven to be effective. However, it has been demonstrated that morphometric parameters that can be derived by Digital Terrain Models (DTMs) help to distinguish such morphologies according to their origins. A definition of an ML pipeline able to detect and categorize large mound populations according to their topographic/morphometric character is relevant for future Mars exploration.

With the ML tools developed for this sciene case, we want to identify the mound populations. Based on their topographic/morphometric parameters we aim at obtaining further classifications. The ultimate achievement would be an automatic mapping and classification of topographic positive relief features according to the likelihood of their origin process.