ARTIFICIAL INTELLIGENCE AND SATELLITE MONITORING FOR THE PLANET

Together with a constellation of other sensors, the galaxy of European satellites within the Copernicus program have collected 34 petabytes (a petabyte is one million GB) of data on land, water, ice and air.

Alongside the satellites, data coming from GPS devices and the Internet of Things (IoT) create even more extensive environmental data sets, which are then made available to researchers, governments and private companies.  Overall they create the "raw" knowledge of EO, an acronym for Earth Observation.

For the European Commission and the European Space Agency, this data must be increasingly accessible and intelligible to generate knowledge and action. For this reason, a single collection point, the Copernicus Data Space Ecosystem, has been created. Here you can find data of all types: radar data to monitor dam failures, air quality surveys that reveal greenhouse gas emissions and measurements which help predict the risk of flooding after heavy rainfall.  Now it's about making it more and more usable and digestible.

“Our goal is to provide decision makers with the most complex information and simulations,” explains to Materia Rinnovabile Giuseppe Borghi, an expert in artificial intelligence and robotics from ESA's Earth Observation Programs Directorate. 
“Today there are programs, such as the Destiny project which the European Commission has entrusted to a consortium that includes ESA, ISS (International Space Station), WWF and EUMETSAT, that strive to create digital twins of the earth to provide complete simulations of all the processes relevant to decision makers, both on climate change and on extreme weather events. This is meant to help non-technical, non-expert people to analyze possible scenarios twenty years from now.”

THE COMING OF AI 

In the architecture of knowledge, immense amounts of EO data are useless without processing, as they can't generate meaning, like letters without semantics and a grammar which should, instead, be as simple and effective as possible.

Artificial intelligence, with its deep learning or foundation models (such as ChatGpt), has created a turning point for the processing of EO data, explains a Briefing Paper of the World Economic Forum from February 2024. According to the authors, not only will it be possible to process and find information in the millions of terabytes of information but it will also be possible to generate "intuitive user interfaces that will make earth observation accessible even to non-experts", encouraging innovation in business models through scalable applications in various sectors, including nature management, energy production and efficiency, transport, climate risk reduction, territorial planning and water management.  Users will include insurance companies, plant managers, farms, urban planners and even regular citizens who will be able to use this information for their own needs without needing a doctorate in analytical statistics.

ENVIRONMENTAL MONITORING AND EARTH OBSERVATION

Discover the exhibition sector of Ecomondo

APPLICATIONS OF SATELLITE DATA & AI

AI and EO have many potential applications in the public and business sectors.

The possibility of making increasingly accurate weather forecasts, for example, leads to detailed information on solar radiation and wind patterns that can help choose the most suitable places for new plants. Google DeepMind has recently developed a tool, GraphCast, that shows that AI can be used to make forecasts of up to ten days “more accurately and much faster than the High Resolution Forecast (HRES), the industry standard weather simulation system, created by the European Center for Medium-Range Weather Forecasts (ECMWF)”.

With systems of this type, risk management during extreme events also becomes simpler, thanks to the ability to create real-time simulations of impacts and to activate the alarm. The ECMWF itself is now using the Google model by running it on its data, to compare it with the method on which weather forecasts in Europe are currently based.

Digital know-how supported by AI is also increasingly necessary in water management. UpstreamTech's HydroForecast project, an end-to-end modeling and decision support service, provides accurate and reliable hydrologic forecasts by modeling predicted future conditions for surface water flows, such as how much water will inflow into a lake over the next 10 days, allowing hydropower plant owners to plan operations more effectively.

Google's Flood Hub, on the other hand, monitors floods in real time and generates maps that can be checked in real time by communities and civil protection, while also sendindìg app notifications or text messages based on the devices' location.

The intersections between satellite data, ground observations and AI promise important breakthroughs in the field of land use analysis. “Analyzing land use change is crucial for the climate and affects the accounting of carbon emissions”, Claire Monteleoni, professor of Computer Science at the University of Colorado Boulder and AI pioneer, explains in an interview with Materia Renewable. “However, the state of the art on this topic is currently very uncertain, we do not have clear models as we do for fossil fuels. That’s why we started working on a new project to build one, based on the amount of data available today thanks also to satellite observations. Scientists have been working on climate models since 1969, but modeling focused on soil, land change and vegetation is actually a much newer field."

SPREADING KNOWLEDGE

In order to raise awareness of the opportunities of using satellite data and AI in relation to the risks of natural and climate disasters, the European Space Agency launched Eurisy, created to promote the use of satellite solutions for risk analysis of natural disasters. “We develop awareness campaigns for public administrations and civil protection,” Annalisa Donati, general secretary of Eurisy said in an interview with Materia Rinnovabile. “Our goal is to create a bridge between space applications and society”. By organizing workshops and meetings, Eurisy introduces municipalities to services available upon request through the Copernicus Emergency Service, including Rapid Mapping which offers geospatial information to map areas at risk of fire or flood, in support of emergency management plans.

ESA deals with innovation, both as research and as marketing of solutions linked to the processed EO data, through the Φ-lab, directed by Giuseppe Borghi, who explained: “With our Invest Office we support entrepreneurs in the development of products and services, while the Explore Office moves along three lines: one dedicated to artificial intelligence and machine learning; one for quantum computing and edge computing; and a third line for blockchain, cognitive space and the Internet of Things. The technologies we are investing in are all linked to the Information and Communications Technology revolution, that is, everything that allows us to extract data from Earth observation and combine it with financial, economic and medical information, to create a digital model of our society". And successfully respond to the challenges of the ecological transition.

Article by Emanuele Bompan