We discuss the application of unmanned aerial systems (UASs) and UAS-derived data to model glacier topography for detecting and analysing slow and rapid changes in glacier surfaces. The study was conducted in East Antarctica in the austral summer 2016/2017. The surveyed areas included an eastern part of the Larsemann Hills, an Airfield of the Progress Station, an initial section of a Sledge route from the Progress to Vostok Stations, and a north-western margin of the Dålk Glacier. We used a Geoscan 201 Geodesy, a professional-grade flying-wing UAS. For the photogrammetric processing, we utilized the Agisoft PhotoScan software. The direct georeferencing approach was applied. Several high-resolution, multitemporal digital elevation models (DEMs) for surveyed areas were produced. The DEM accuracy was estimated in the absence of ground control points and reference DEMs. We determined the ice flow velocity within the Airfield area and the Sledge route zone, where marked points were installed and moved together with the ice surface. We found that UAS imagery can be used for real-time monitoring of open and some snow-covered crevasses of various sizes. Wind-driven snow microforms are also well recognizable on UAS-derived aerial images. The 2017 Dålk Glacier catastrophic subsidence demonstrated that monitoring and studying such events in glaciers and ice sheets are almost impossible without UASs. Optimal meteorological conditions were empirically determined for conducting unmanned aerial survey to obtain images suitable for subsequent photogrammetric processing and DEM generation. Finally, we discuss adaptation of equipment and software to the Antarctic environment.
For Antarctic research, one of the important support tasks is a monitoring of sledge routes and other visited areas for detection of open crevasses and revealing of hidden, snow-covered ones. We study the possibility of revealing hidden crevasses with geomorphometric treatment of high-resolution, glacier digital elevation models (DEMs) produced from imagery captured by unmanned aerial systems (UASs). The study was conducted in East Antarctica, at a sledge route from the Progress to Vostok Stations. The survey performed with Geoscan 201 Geodesy UAS, resulted in a set of aerial images with a resolution of 6 cm. From these images, we produced glacier DEMs for a typical area with snow-covered crevasses. The glacier DEMs with the resolutions of 25 cm, 50 cm, and 1 m were treated by geomorphometric methods. We derived models of 16 morphometric variables including horizontal and vertical curvatures. For the 1-m gridded DEM, mapping of some variables allowed us to reveal one of the crevasses. The approach is functional, at least, for revealing crevasses with a width of 1.5 m and wider, covered by the snow layer with a thickness of up to 1.2 m.
We describe and interpret a catastrophic subsidence in the Dålk Glacier, East Antarctica, occurred in the morning of 30 January 2017. We analysed a dataset of high-resolution aerial photographs and video records captured ten days before, one hour after, and ten days after the event by unmanned aerial systems (UASs) as well as UAS-derived digital elevation models (DEMs). We reconstructed a subsidence mechanism as follows: An englacial cavern measuring about 885,000 m3 with a thin roof existed before the event. There was a channel of the surface and near-subsurface meltwater flow from the Boulder Lake to the Prydz Bay passing nearby the cavern. About a day before the event, an intensive meltwater flow began from the Boulder Lake. The meltwater accumulated on the cavern rooftop leading to its initial collapse. In several days, the meltwater completed the process of subsidence and collapse of the cavern. The event led to the formation of an ice depression about 40,000 m2 in size and up to 43 m deep. Our results demonstrate that ice-surface hydrological processes can cause rapid, deep subsidences of large areas in glaciers, without involving englacial water outbursts from subglacial lakes as well as subglacial geothermal or volcanic activity.