Skip to Content

AI Flood Defense

Capgemini
2020-05-13

In 1978 NASA launched Seasat, one of the first Earth observation satellites. It revolutionized our understanding of the world’s oceans, and sparked a panic at US national security agencies.

Seasat was designed to conduct the first global survey of the Earth’s oceans from space. To do this it carried an instrument known as a Synthetic Aperture Radar (SAR), which transmitted pulses of microwaves down to the ocean’s surface then recorded the returning echoes.

Seasat’s SAR instruments enabled scientists to see tiny variations in the topography of the oceans’ surface, revealing the complexity of global ocean currents. The US military discovered it could see something else too: the tracks of their nuclear submarines travelling hundreds of meters below the surface.

The incredible accuracy of SAR has only improved in the decades since Seasat, and the range of applications for the data it produces has expanded into many unexpected areas. Today Seasat’s descendants are helping governments and enterprises protect against floods and revolutionize hydropower.

Submerged in the data

Seasat was a triumph for the fledgling science of Earth observation. In a hundred days it discovered more about the oceans than shipborne surveys had managed in a hundred years. Since Seasat, other observation satellites with increasingly sophisticated SAR instruments have vastly improved our understanding of ocean and weather systems, and become a headache for militaries trying to keep their submarines secret.

Data science is the key to Earth observation. Seasat and its successors provide terabytes of raw data. Creating the algorithms to extract information and insights from that data is the difficult part. Some algorithms have allowed us to build weather prediction systems hundreds of times more accurate than previously thought possible, others have become the subject of murky, real-life espionage dramas.

One of the great strengths of Earth observation, or any big data collection, is that you can go back to the data again and again to discover new insights. In recent years, data from ocean observation satellites has been used for an entirely new purpose, to track the rise and fall of major rivers.

Even the biggest rivers are obviously much smaller targets than an ocean, but radar echoes bounce back from them just as well and it’s possible to pick them out of the noise from surrounding land features with the right algorithms. This has created a new range of commercial and public safety applications.

Troubled waters

Rivers are critical to the global economy. They provide transport, food, irrigation, hydropower, and water for industrial and domestic use. The billions of tons of water and sediment they carry from the land to the oceans are also a major part of the environmental big picture we must understand and master. River flow data sets delivered by space-based platforms offer new and wide-ranging insights for the study and management of rivers.

A lot of attention has been given to the risk from sea-level rise but flooding from inland water courses (rivers etc.) is a hugely destructive problem right now. A 2015 report by the World Meteorological Organization estimated that rain storms and floods caused one million deaths between 1970 and 2012. A warming climate is likely to generate more rainfall globally, which can only increase the likelihood and scale of inland flooding.

In the developed world, major rivers are monitored by water-level gauging stations, which build up a picture of seasonal variation and give early warnings of surges. In the developing world on-the-ground monitoring is either sparse or non-existent. This represents a very large risk to the cities and agriculture along their banks.

Large river basins in equatorial Africa and South America are now being monitored using data from space-based platforms, providing a level of early warning for floods, and helping with water management for agriculture, urban use, and hydropower.

Capgemini’s Geospatial AI solution – a powerful method for creating valuable insights from Earth observation data – has been applied to ocean and river observations to help identify flood risks, and to predict surges in water levels critical to the safety and efficiency of hydropower operations.

Author



Robert Engels
CTO Europe I&D

Robert has a long term and deep interest in topics and tangible things related to machine learning and artificial intelligence. His wider interests include topics like Semantics, Knowledge Representation, Reasoning, Machine Learning (in all its different colours & shapes) and putting it together in more (or less) intelligent ways.
Where technology meets people, a background in cognitive psychology comes in handy. That’s where the fun starts, and that’s where he wants to be. Utilizing, explaining, producing and creating scenario’s, solutions and understanding for new challenges and situations where AI & ML come around the corner.
Robert holds a PhD in Machine Learning from the Technical University of Karlsruhe (now KIT). He is a regular key-note speaker and published articles on various topics in artificial intelligence, machine learning, semantic web technology, information representation, knowledge management and computer linguistics.