Thermokarst lakes, which are common in permafrost regions, play a significant role in greenhouse gas emissions. The amount of macrophyte vegetation within a lake's littoral zone influences its methane emissions. This study investigates the factors influencing the extent of macrophytes in thermokarst lakes with a focus on remote sensing techniques, including the use of Synthetic Aperture Radar data from Sentinel-1 and multispectral optical data from Sentinel-2. The analysis centers around the Circumarctic Landcover Unit (CALU) dataset, which includes a unit specific to shallow water and abundant macrophytes. Two modified versions of the CALU were created by combining the originally used data with Sentinel-1 and -2 data that were acquired at different points of time. The altered versions were compared to the original CALU to assess the impact of different acquisition onto macrophyte classification. The fraction of shallow area per lake was derived from an intermediate product of the CALU creation.The altered versions of the CALU showed a lower occurrence of macrophytes than the original CALU. Key features assumed to influence macrophyte occurrence were related to the fraction of macrophytes per lake, including lake area, shallow lake area, latitude and longitude, groundfast ice fraction and distance to coast. The results of the analysis revealed that macrophyte fractions were most commonly associated with lake area and the fraction of shallow water, with the highest macrophyte coverage  found in lakes with surface areas between 1500 and 20000 m2 with an average coverage of 50 %. Even though the amount of groundfast ice per lake also depends on the shallow area per lake, no direct relationship to the fraction of macrophytes could be found. A Random Forest Model applied across all study areas showed strong predictive capabilities for these factors, with total lake area identified as the most important predictor. When applied to individual study areas, the model showed regional variability, with cross-validated R2 values consistently higher in the single areas compared to the generalized model. This suggests that the generalized model could not fully account for local environmental factors.The findings of this study contribute to a better understanding of the drivers behind macrophyte distribution in Arctic thermokarst lakes. They emphasize the potential of remote sensing to estimate macrophyte vegetation and demonstrate the applicability of Random Forest models for such studies.