Ripple patterns on Martian barchan dunes in the north: indicators of the flow regime.

Abstract

Aeolian sand transport occurs across the surface of Mars with the north polar region being one of its most active regions, however our understanding of sand transport conditions on the red planet is limited. Sand ripples reflect local flow regimes and are present on both Earth and Mars; but Martian ripples greatly vary in shape and size. Large ripples on Mars have meter-scale wavelengths but seemingly no coarse grains at their crests. Investigating the dynamics of large ripple patterns across Mars would improve our knowledge of local wind regimes and sand transport conditions. In this study, we have selected 40 HiRISE sites with a 25 cm/pixel resolution in the north polar region and cropped out hundreds of barchan dunes overlain by large ripples. The barchan images are filtered to remove the illumination effect, and the surrounding bedrock are masked. From a visual analysis of 20+ dunes, we have identified 3 ripple pattern types, straight, sinuous, and complex, which all reflect different flow regimes. Then, we have applied and compared two methods to automatically map these 3 ripple pattern classes using labels. Our first approach is a deep-learning algorithm based on the U-Net architecture which has been trained to recognise the ripple patterns from the labels and identify them on new data. Our second approach is computing a semi-variogram, using the labels as reference, and extracting the ripple wavelength, direction, and sinuosity. The spatial distribution of these later metrics over the dunes are used to infer the local wind regime around the north polar region of Mars. By doing so, we hope to enhance our understanding of sand transport conditions on the red planet.