<p>Information on snow depth and its spatial distribution is important for numerous applications such as the assessment of natural hazards, the determination of the available snow water equivalent (SWE) for hydropower, the dispersion and evolution of flora and fauna and the validation of snow-hydrological models. Due to the heterogeneity and complexity of snow depth distribution in alpine terrain, only specific remote sensing tools are able to accurately map the present variability. To cover large areas (> 100 km<sup>2</sup>), airborne laser scanners (ALS) or survey cameras mounted on piloted aircrafts are needed. Applying the active ALS leads to considerably higher costs compared to photogrammetry but also works in forested terrain. The passive photogrammetric method is more economic, but limited due to its dependency on good acquisition conditions (weather, sufficient light, contrast on the snow surface). In this study, we demonstrate the reliable and accurate photogrammetric processing of high spatial resolution (0.5 m) annual snow depth maps during peak of winter over a 5-year period under different acquisition conditions within a study area around Davos, Switzerland. Compared to previously carried out studies, using the new Vexcel Ultracam Eagle M3 survey sensor, improves the average ground sampling distance (GSD) to 0.1 m at similar flight altitudes above ground. This allows for very detailed snow depth maps, calculated by subtracting a snow-free digital terrain model (DTM acquired with ALS) from the snow-on digital surface models (DSMs) processed from the airborne imagery. Despite complex acquisition conditions during the recording of the Ultracam images (clouds, shaded areas and new-snow cover), 99 % of unforested areas were successfully reconstructed. We applied masks (high vegetation, settlements, waters, glaciers) to significantly increase the reliability of the snow depths measurements. An extensive accuracy assessment including the use of check points, the comparison to DSMs derived from unpiloted aerial systems (UAS), and the comparison of snow-free pixels to the ALS-DTM prove the high quality and accuracy of the generated snow depths. We achieve a root mean square error (RMSE) of approximately 0.25 m for the Ultracam X and 0.15 m for the successor sensor Ultracam Eagle M3. By developing an almost automated, consistent and reliable photogrammetric workflow for accurate snow depth distribution mapping over large regions, we provide a new tool for analysing snow in complex terrain. This enables more detailed investigations on seasonal snow dynamics and serves as ground reference for new modelling approaches as well as satellite-based snow depth mapping.</p>