Space research What Space Reveals from Above

Related Vendors

ETH Zürich Microelectronics Design Center

Wuxi InfiMotion Technology Co., Ltd.

Documenting war crimes, supporting vaccinations, measuring snow depths, and tracking climate change—satellite Earth observations offer immense potential. Researchers at ETH Zurich (Switzerland) are leveraging this technology.

Satellites of the European Space Agency (ESA) send radar beams to Earth from an altitude of 700 kilometers. These are scattered back and provide images of the Earth's surface even when the sky is overcast. Konrad Schindler and his team are particularly interested in the radar images of Ukraine. "The satellites fly over the country every few days, providing a kind of slow-motion video," explains the head of the Institute of Geodesy and Photogrammetry at ETH Zurich. "The signal changes when intact houses turn into a field of rubble." This is because the radar beams are reflected more strongly on smooth surfaces than on rough ones.

However, humans would be overwhelmed by the task of analyzing the freely accessible flood of images. This task is performed by an automatic monitoring system based on artificial intelligence (AI). "This means that the computer system is left to statistically recognize typical patterns in the data," explains Schindler. For this machine learning, the AI system is trained with sample images in which you are sure to see a destroyed building that was previously still intact.

This reference data was provided by the UN Satellite Center (UNOSAT). Experts there painstakingly identify such building damage by hand using high-resolution images from commercial providers in order to document war crimes such as the bombing of civilian facilities. "Our monitoring system is by no means a substitute for these experts," says Schindler. "But because it can cover large areas, it gives them important clues as to where they should take a closer look."

Cards for aid organization

The mapping of war damage is one of two projects that the ETH researchers carried out in collaboration with the International Committee of the Red Cross (ICRC). "The second project was about the seemingly very simple question: where are the people?" says Schindler. If the ICRC is planning a vaccination campaign or a relief operation following a natural disaster, it needs data on population density, which is often lacking in less developed countries. This gap is filled by the population maps that Schindler's group creates using satellite data.

The ETH researchers use satellite observations in combination with machine learning for a wide range of areas. In collaboration with the Swiss start-up ExoLabs, they have developed a system that measures snow depth faster and more precisely than was previously possible. Satellite images and digital elevation maps provided the first estimates of snow cover in Switzerland. From comparisons with precise data collected by an airplane and measuring stations on the ground, the system learned to improve the estimate based on the satellite images. It now provides daily updated snow depth maps, which are due to be launched on the market soon.

A global map of tree heights is already freely available, which is primarily used for ecological purposes and is also based on satellite data. "From space, you can't see where a tree begins and ends, but the reflected signal still contains a surprising amount of information, especially in the infrared range, where vegetation is particularly active," explains Schindler. A computer program learned how these local illumination patterns relate to tree height from examples provided by a laser scanner on board the International Space Station (ISS). Depending on whether the laser pulses are reflected by the treetops or the ground, their transit time shifts minimally. The height of the vegetation can be derived from this.

Early warning system for forests

"The global mapping of tree heights is extremely interesting because it contains information about the age of the stands," says Verena Griess, Head of the Institute of Terrestrial Ecosystems at ETH Zurich. She and her team also work with satellite data to obtain information on the condition of trees. "With climate change, we are in an era of rapid change, also in terms of our forests, and problems caused by pest infestation, windthrow, fire and drought stress are becoming more frequent," says the Professor of Forest Resource Management.

With the help of satellite data, the researchers want to create an early warning system that is designed to trigger a need for action in a stressed forest, even if the signals are not yet visible to the human eye. Such a system could detect small fires in sparsely populated areas or indicate where measures are needed in Switzerland to ensure that the forest continues to protect us from natural hazards. Satellite data could also identify areas that are particularly suitable for the production of wood or areas that need to be protected due to their high biodiversity. "We want to ensure that the right decision is made for each location," explains Griess.

The researchers are currently working on a model that can automatically recognize tree species on satellite images. Based on verified locations of certain species such as beech, spruce or larch, the model learns to recognize the corresponding patterns on the satellite images. Using so-called hyperspectral sensors, satellites also collect information that is not visible to humans and provides information about the health of trees. This is because they can be used to measure the chemical content of certain substances in the leaves and how their concentration changes, which can be the case in the early stages of infestation by pests, for example.

The two-dimensional satellite images can also be linked with 3D data provided by laser scanners from space, from aircraft or drones and from the ground. ETH's Alps supercomputer, which recently went into operation, will speed up the data analyses and improve the training of the models. "With this, the outstanding students and the good cooperation between the professorships and the other institutions of the ETH Domain, Switzerland has an expertise that cannot be found anywhere else in such a concentrated form," says Griess.

GPS technology repurposed

The condition of forest trees can also be investigated using signals emitted by GPS satellites. If a receiving station is placed under a tree and a second one further away, the difference in the signals received can be used to calculate how much biomass is stored in the tree. "It's amazing what you can do with a technology that was actually developed for something completely different—navigation," says Benedikt Soja, Professor of Space Geodesy.

In collaboration with MeteoSwiss, Soja and his team installed low-cost GPS receiving stations at several weather stations throughout Switzerland. This makes it possible to measure how much water vapor is present in the atmosphere, as this slows down the radio signals emitted by the satellites. This makes it possible to calculate the humidity above a station and thus improve rain forecasts.

An analysis of the measurement series over a longer period of time also provides information about climate change. If it gets warmer, the atmosphere can absorb more moisture, i.e. more water vapor. "This in turn intensifies climate change, because water vapor is also a greenhouse gas—in fact the most abundant one," explains Soja. "This makes it particularly easy to determine the impact of climate change on the atmosphere at a local level." The GPS measurements show that it has become wetter on average, but also drier in some regions.

The researchers are using a different measuring technique to track the melting of the ice sheets in Greenland and Antarctica. Two satellites flying in close succession record the Earth's gravitational field. This varies depending on how much mass is present at a particular location. "This allows us to measure mass shifts from space," explains Soja. These measurements can be further refined with the help of other satellites that take altitude measurements. "We see that year after year, huge amounts of ice melt in the polar regions and cause sea levels to rise—hundreds of gigatons, a huge ice cube the size of the city of Zurich and over 3,000 meters (approx. 9,843 ft) high," says Soja. "This observation from space is the best technique for monitoring the consequences of climate change in relation to the ice sheets."