Abstract: Photogrammetry of the snow cover is expected to be difficult, because of its homogenous, low-contrast surface, hindering the automated detection of matching points to generate digital surface models. On the other hand there is rising request for high spatial resolution snow depth and snow surface type mapping as well as snow avalanche event documentation. Unmanned aerial systems (UAS) enable flexible, efficient and economic data acquisition, even within inaccessible alpine terrain. The efficient and economic generation of high-quality digital surface models (DSM) of snow covered surfaces would be a major step forward for many applications in snow hydrology and hydropower generation, avalanche research and warning, for winter tourism as well as for alpine ecology investigations. We investigate the performance of UAS-based structure-from-motion (SfM) photogrammetry on very homogenous snow surfaces at the test site Tschuggen at 2,000 m a.s.l. close to Davos, Switzerland under diffuse illumination conditions, caused by a completely overcast sky, a situation that could be described as worst case for photogrammetry. We investigate the benefit of near infrared information (ë > 830 nm), compared to imagery acquired within the visible part of the electromagnetic spectrum (ë = 400 700 nm) and a combination of these two. We evaluate the accuracy of the different digital surface models (DSMs) qualitatively and quantitatively by applying differential Global Navigation Satellite System measurements with an expected accuracy better than 10 cm in x, y and z directions. The results of this study enable an assessment of the potential and the limitations of SfM photogrammetry for applications on snow-covered terrain.
