We are a planet-based species: With a few notable (astronaut) exceptions, everyone who has ever lived remained glued to our spinning Earth, which now houses 8.3 billion people.
That’s the gravity of the situation. But we’re also a world of floods, typhoons, earthquakes, heat and cold calamities, volcanic eruptions and even encounters with sun-caused auroras and bruises from incoming asteroids.
It’s quite a world worth watching — and for decades that’s what spacecraft have been doing. But what do they view and who’s on the other end of the space-based telescopes capturing our daily doings, faults and all, down here on terra firma?
Privacy, security and trust
Recently, the Organization for Economic Co-operation and Development (OECD), a group based in Paris, published a report on the expanding access to satellite Earth-observation data, spotlighting what it means for privacy, security and trust.
“Advances in optical systems, photonics, cloud computing and artificial intelligence [AI] have democratized both the quality and accessibility of satellite data. However, this convergence of technologies also carries risks, including for national security and privacy,” states the report, which was authored by the OECD’s Marit Undseth and Claire Jolly.
Broadening access to advanced satellite data also creates new challenges, states the report, such as one of the key characteristics of Earth observation — its dual-use nature. “The same satellites used to track illegal fishing can often also detect military troop movements,” the authors write.
New concerns
The OCED report notes that Earth-observation data and the technological convergence with artificial intelligence raise new concerns about national security, privacy and the ethical use of these data.
In fact, the report points to the growing hazard of fake or incorrectly used satellite imagery. “As part of the growing trend of digital disinformation, the trust in satellite data can be eroded by fake, misinterpreted or intentionally misrepresented imagery,” it states.
Space.com asked leading remote-sensing specialists what’s being done, what’s being learned and what we should worry about.
Sovereignty and cost efficiency
Viktor Stoyanov is the chief operating officer of Smart Solutions at Space42, where he oversees the business unit’s operational and financial performance.
Space42 is a United Arab Emirates-based space technology company that integrates satellite communications, geospatial analytics and AI capabilities. The group runs the Foresight Earth Observation constellation of synthetic aperture radar (SAR) satellites.
Space.com asked Stoyanov about the growing international use of remote-sensing data.
“Governments increasingly expect Earth-observation systems to deliver both sovereignty and cost efficiency, two goals that were long seen as mutually exclusive, as sovereignty typically meant heavy in-country investment in duplicative research and development and capital infrastructure,” he responded.
Dual use
Stoyanov said Space42’s focus is to combine those two outcomes.
“We believe that can only be achieved through international cooperation. Our Foresight constellation reflects this model, as the first sovereign dual-use synthetic aperture radar constellation in the region,” Stoyanov said.
The satellites were manufactured in partnership with ICEYE in Finland, with critical integration and testing completed in the organization’s facility in Abu Dhabi.
“The Abu Dhabi facility was built in cost-cautious and scalable manner, so that it can satisfy current sovereignty requirements commensurate with the constellation size, with the ability to expand the facility and meet demand of future programs,” Stoyanov said.
Through cloud, dust and darkness
What new in-orbit technologies are being applied to Earth observation?
“The biggest shift in recent years has been the move from few, large, general-purpose platforms to constellations of small, specialized satellites and synthetic aperture radar,” said Stoyanov.
“SAR satellites image through cloud, dust and darkness, which matters enormously for our region and for any customer who needs consistency regardless of timing and weather,” he added.
Looking ahead, Stoyanov said that the short lifespan of satellites in low Earth orbit is an opportunity to ensure that each replenishment cycle becomes an upgrade cycle.
“As we plan to maintain and expand our fleet, we will continuously seek to deploy technical improvements across various performance metrics of the system — resolution, latency, satellite life, for example,” said Stoyanov.
Artificial intelligence
A SAR constellation generates far more imagery than analysts can review manually within the allocated time, Stoyanov said. So the value lies in what can be extracted from it, not in the pixels themselves.
AI is therefore becoming increasingly important as a sorting tool for remote-sensing data, Stoyanov said. It is now the only practical way to work, for three reasons: volume, speed and accuracy.
Customers responding to a flood, a maritime incident or a supply-chain disruption need an answer in minutes, not a report next week — “decision-grade intelligence within minutes,” explained Stoyanov.
On the topic of accuracy, while trained analysts are excellent at interpretation and large-object classification — like buildings and sea vessels — there are changes that the eye can easily miss. “Algorithms comparing each new pass against the historical baseline — they catch those consistently, without fatigue,” said Stoyanov.
Hybrid model
Francis Doumet is the CEO and co-founder of Metaspectral, a company based in Vancouver, British Columbia, that builds AI technology to rapidly analyze hyperspectral imagery.
Doumet views collaboration between different national space agencies as “extensive and, somewhat uniquely, very open.” Missions are frequently carried out through “shared funding, instruments, expertise and ground infrastructure,” he said, with the resulting scientific data made available internationally.
Commercial imagery is less openly distributed because it is commercially licensed and may have security implications, Doumet said, but it is increasingly enabling international cooperation in the defense sector.
For example, the U.S. Space Force‘s Tactical Surveillance, Reconnaissance and Tracking (TacSRT) program combines commercial satellite imagery and analytics from American companies and allied nations to rapidly respond to operational requests, Doumet said.
“This has shaped the Earth-observation sector into a hybrid model,” said Doumet. Governments and allies share requirements and intelligence, he explained, while commercial operators supply imagery and analytics “at a speed and scale that would be difficult or impossible for a single nation to produce alone.”
Hyperspectral sensors
The improvement in the spatial resolution of hyperspectral sensors is an important advance in Earth observation, Doumet pointed out.
Better sensors, high-speed compression and improved communications, such as optical links, are now narrowing a “resolution gap.” Onboard AI is equally transformative, he added.
“Satellites can analyze imagery in real time, identify relevant signatures and transmit alerts or compact intelligence products instead of waiting to download an entire hyperspectral data cube,” Doumet explained.
Spectral fingerprints
Angie Crews is a principal research associate with the University of Colorado, Boulder’s Center for National Security Initiatives.
“There has been a tremendous growth in the commercial remote-sensing industry,” Crews told Space.com, noting that this field had previously been government-based. “It’s now possible for these commercial companies to deploy entire constellations of Earth-observing satellites, which fundamentally changes the architecture in place.”
Crews pointed to the hyperspectral nature of Earth remote sensing — taking the “spectral fingerprints” of what’s occurring on Earth.
“It’s impressive how the industry is moving forward with some of these capabilities,” she said.
Balancing act
“There are some tremendous benefits as well as some things that we need to think about” when it comes to the new direction that Earth observation is taking, Crews said. Remote-sensing data can be inherently dual use, she added.
“Overall, I think we’re seeing some enormous benefits from our increasingly capable commercial remote-sensing systems. But it does raise some security and governance questions,” said Crews.
It’s a balancing act to manage, according to Crews. How best to secure the information for national security purposes versus not putting the squeeze on innovation exhibited by the private Earth-observation firms?
Meaningful information
Shifting to her teaching role at the University of Colorado, Boulder, Crews told Space.com that one of the things that’s especially important is to train students in the remote-sensing field to think beyond any single sensor or dataset.
“The next generation needs to understand how different sensing technologies work and also how to work with very large datasets and extract meaningful information using AI and machine learning,” Crews said.
“We also need to train the students on critical thinking and ensure they are aware of the limitations of AI. The underlying physics need to be understood and the results need to be validated so that the students use and interpret the results appropriately,” she concluded.