This AI determines the exact snow depth in the Swiss mountains with record speed

Snow data from space: Using artificial intelligence (AI), researchers can read snow depths in Switzerland from satellite images. This method is more accurate and faster than previous methods, ETH Zurich announced on Thursday.

According to the Swiss Federal Institute of Technology in Zurich (ETH Zurich), snow depth data provide an important basis for winter tourism, for hydropower and for assessing the avalanche risk of winter sports enthusiasts.

So far, snow monitoring in Switzerland has been mainly based on data from measuring stations, ETH Zurich said in the statement. Since there are only about 400 stations for the whole of Switzerland, the snow information for many places is quite inaccurate.

For the measurement based on satellite images, researchers from ETH Zurich and the Swiss company Exolabs used images from Sentinel
2 satellites of the European Space Agency (ESA). Every five days they take images of every location on Earth. In addition to the satellite images, the AI ​​researchers fed extensive terrain data from Switzerland.

Using this satellite and terrain data, they trained the AI ​​to infer the snow depth for a given point. To do this, they used a process called supervised learning. To do this, they had the system estimate snow depth and compared the results to real snow measurements.

“We determined at each grid point how far the AI ​​deviated from its estimate and gradually adjusted the system so that the errors became smaller,” Konrad Schindler explains in the ETH statement.

“While the best existing snow maps in Switzerland have an effective resolution of about 250 by 250 meters, on our maps you can zoom in to 10 by 10 meters to read the snow depth,” says Schindler. “We assume that we are setting a new standard for snow depth measurements in Switzerland.”

In addition to Sentinel satellite images
2 satellites, the researchers also use images from other satellite missions, which are spatially less accurate but do provide daily images.

According to a statement from ETH, the research group has a lot of experience with satellite images: they use such images to predict population density in crisis areas, determine war damage to buildings in Ukraine or to measure the height of forests worldwide.

(dsc/sda)