A new technology can measure the depth of snow in all locations across Switzerland more rapidly and with greater accuracy than before. To achieve this, a research group at the Swiss Federal Institute of Technology in Zurich (ETH) headed up by Konrad Schindler, Professor of Photogrammetry and Remote Sensing, has set a new standard for measuring the depth of snow in conjunction with the Zurich-based firm ExoLabs, a spin-off from the University of Zurich (UZH) founded in 2016, further details of which can be found in a press release issued by ETH Zurich.
This required two things: firstly, millions of optical images and infrared imaging from the European Space Agency’s Sentinel-2 satellites, which record every location on earth every five days. Secondly, the researchers input comprehensive terrain data from Switzerland to train the AI. For example, the AI learned that the snow on a steep south-facing slope melts more quickly when exposed to sunlight than in a shady hollow.
In a first round of training, the ETH Zurich team used snow maps from ExoLabs, which are based on images from Sentinel-2 in addition to those obtained by other satellite missions. ExoLabs has so far offered this technology to various apps, including Outdooractive, Skitourenguru, Hüttenbuch and swisstopo. A fine-tuning process was carried out with the help of highly detailed snow data collected by the Institute for Snow and Avalanche Research (SLF), which is based in Dischmatal in the canton of Graubünden. The AI is able to extrapolate the spatial connections provided by this dataset across the whole of Switzerland.
This research project is financed by Innosuisse, the Swiss Innovation Agency, with ExoLabs responsible for marketing the technology. As such, the company is now able to “support companies from the insurance, hydropower and tourism sectors and will now offer this product also outside of the Alps, for example in North America, South America and the Himalayan regions”, as Reik Leiterer, CEO of ExoLabs, explains in a video produced by ETH Zurich.
