<b>Complex Wind Regimes on Mars Reconstructed </b><b>from Aeolian Features using Deep Learning: </b><b>A Comparison Against the MGCM Output</b>

This study applied a novel approach using Convolutional Neural Networks (CNNs), trained with Earth's aeolian features and wind data, to reconstruct complex surface wind regimes on Mars. The lack of extensive in-situ Martian wind data led to using this Earth-referenced model, combining CNNs with remote sensing data and atmospheric reanalysis. The findings revealed notable discrepancies between the CNN-derived wind patterns, with a 3-8 km resolution resolution, and the existing 40 km resolution data from a Mars General Circulation Model (MGCM). This highlights the potential limitations of current Martian atmospheric models and suggests a need for more refined simulations, possibly using models like MarsWRF, to better evaluate and understand Martian surface wind patterns.