Carlos M. Meléndez is the COO and cofounder of Wovenware, a Maxar company, offering AI and software development services.
According to Deloitte, more than 1,170 Earth observation satellites are currently in orbit, and 51% of them have been launched since 2019.
Thanks to massive advances in satellites as they monitor Earth from their scenic overlook in space, we’re able to capture images of Earth in ways that were previously unimaginable. These satellites contain sensors that increasingly offer incredible advancements in image resolution and clarity, as well as frequency of observation. This can culminate in upwards of a hundred terabytes of data in the form of images per day.
It’s no wonder that something beyond human intelligence is needed to sort through this deluge of imagery and geospatial data. It requires the power of artificial intelligence (AI) to give them life and turn images into insights—namely, geospatial AI (GeoAI).
Understanding Geospatial AI
GeoAI is the application of AI to geographical data about specific areas of the Earth to accelerate real-world understanding of business challenges, environmental issues and the safety and security of nations.
GeoAI allows us to accelerate how we extract meaning from complex datasets to solve these challenges. It does this by applying machine learning, deep learning, computer vision and predictive analytics to the data to uncover context from very detailed patterns and relationships to data.
In addition to satellite imagery, this data can come from aerial photographs; GPS coordinates, tracking the movement of people, vehicles and assets; sensor data, such as weather data, pollution measurements or readings from agricultural sensors; as well as demographic data.
AI (more specifically, computer vision) does the grunt work of identifying how many cars are on a city street, for example, or how many healthy trees are in a forest. If this counting and classifying were done by humans alone, not only would it take months to accomplish, but the rate of human error could mean the difference between safe cities and pollution-clogged highways.
It is the combination of robust data—originating from satellites in space—and the application of AI to tell us what we are seeing, that represents the next frontier in data-driven AI.
Geospatial AI At Work Across Various Sectors
But how exactly does it do it?
In its simplest form, very high-resolution satellites are rocketed into orbit and “tasked” with specific missions, such as taking pictures of specific areas of the Earth at various times throughout the day. The goal of a specific mission, for example, could be to monitor shipping activity at a busy port.
The data collected is sent to ground stations on Earth and then usually uploaded to the cloud in near real time. That data is cleaned up, classified and labeled to fuel AI algorithms, so they’re able to identify what exactly a “ship” may be, how many of them there are and at what times of the day they are in port. By extracting data-driven insights, ports can be kept safer and ensure the most efficient routes for the transportation of goods.
Some other industries where geospatial AI services are being used include:
Agriculture: Understanding soil or crop conditions, the grazing patterns of livestock or storm damage to farmland can be determined with AI algorithms fueled by geospatial data that can help farmers take proactive measures to improve conditions.
Civil Services: Public organizations such as the Federal Emergency Management Agency (FEMA) leverage GeoAI to monitor the impact of regions affected by natural disasters, such as floods, fire or hurricanes. By automating the analysis of data taken from wide swaths of land or sea, it also can help predict the likelihood of disasters occurring in the future to proactively address them.
Environmental Services: Geospatial AI can be used to monitor the impact of deforestation, water quality and air pollution levels. It also could map changes to shorelines or wetlands, or even track the size of glaciers.
Mapping: Most people are well aware of navigation apps, such as Apple Maps or Google Maps, which leverage geospatial data, to provide directions and the best routes for drivers.
The Challenges And Limitations Of Geospatial AI
While geospatial AI offers significant potential, some limitations exist.
The accuracy of AI models depends on the quality and completeness of the geospatial data used. Since the use of satellite imagery as data is fairly nascent, scientists continue to fine-tune their models, tools and techniques to extract the greatest insights out of this form of data.
Additionally, as with all forms of data-hungry AI, there are data privacy concerns. Ensuring the privacy of citizens and other individuals when dealing with location data is crucial. It also requires the highest levels of security, since many uses of Earth intelligence are intended to keep nations safe.
Finally, implementing and using geospatial AI requires specialized skills that are just beginning to be realized as GeoAI moves from a very defense-specific tool to one sought after by commercial enterprises.
With this shift, AI is seeping into organizations of all shapes and sizes and requiring increasingly more data to be effective. Geospatial AI, trained upon massive datasets collected from a satellite’s eye view, is enabling a whole new perspective and actionable insights that play a vital role in protecting and preserving our planet, helping businesses succeed and keeping nations safe.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?