Snow has a major impact on palsa dynamics, yet our understanding of its distribution at the small scale remains limited. We used unoccupied aerial system (UAS) light detection and ranging (lidar) and ground truth data in combination with machine learning to model snow distribution at three palsa sites. We identified extremes in snow depth corresponding to palsa topography, providing insights into the influence of the distribution on their dynamics. The results demonstrate the usability of machine learning and UAS lidar for small-scale snow distribution mapping.