It’s Always Sunny at Orbital Insight: Our Work with SAR Imagery
Satellite imagery gives us a wonderfully unique perspective on our planet. The photos can be breathtaking, with new patterns and geometries emerging from the ground, designs often only visible from the distance of space. At Orbital Insight, we also see satellite imagery as valuable for the data it contains. By analyzing millions of images efficiently through artificial intelligence, we can understand trends like deforestation, consumer activity, or oil stockpiling — essentially, the social and economic pulse of the world.
In fact, this “macroscope” we’ve created can seem like an ideal solution to bring transparency to global socio-economic trends. And with more and more satellites being launched every year, soon we’ll have enough imagery to be able to analyze the whole planet daily.
There’s just one problem, and if you look at the classic “Blue Marble” photo below, you’ll quickly see what it is.
Clouds. They serve an important meteorological function, but they can be frustrating for those of us in the business of analyzing satellite imagery. Some key geographic locations, like certain ports, can be covered by clouds up to 80 percent of the time, meaning it’s very difficult to obtain images that actually allow us to count ships or containers. In geospatial analytics, your predictions are only as good as the frequency of your imagery, and if you can’t get usable photos of a location but once in a blue moon, it will be difficult to make meaningful observations.
Fortunately, there is a solution: synthetic aperture radar (SAR) imagery. Satellites that are SAR-equipped are able to see through clouds, haze, and the dark, making it possible to analyze changes on the ground, regardless of the weather conditions or time of day. The difference lies in the imaging modality. Optical satellites acquire their imagery passively, capturing reflected sunlight from the Earth’s surface, and generating a recognizable image. SAR is different. It’s an active imaging process that beams radio frequencies down to Earth, which hit objects and then bounce back to the satellites, where the data is captured. The optical depth of radio frequencies enables the signals to pass through clouds. Normally, the long wavelength of radio frequencies would make it necessary to have a large antenna to capture the signals, but this of course is impractical on an orbiting satellite. The solution SAR employs is the synthetic aperture, which virtually fabricates a long antenna through the process of the satellite moving from one location to the next.
There are some challenges associated with using SAR imagery for geospatial analytics. If you were to look at a raw SAR image like the bottom image below, you’d see that it doesn’t resemble a traditional, optic satellite image very much. It requires a certain type of data analysis and expertise to tease out the meaning from these images. But for those who have these capabilities, SAR imagery remains an invaluable source of data, especially when it’s cloudy.
That’s why we’re excited to say that, at Orbital Insight, we’re committed to incorporating SAR imagery into our analytics process in order to further improve the accuracy of our data. Today, we’re pleased to announce a partnership with e-GEOS to incorporate SAR data from their COSMO-SkyMed satellite constellation, and last September, we entered a partnership with MacDonald, Dettwiler and Associates to utilize SAR data from their RADARSAT-2 satellite. By expanding our access to SAR-enabled satellites, we’ll be able to make our analytics even more reliable and extensive.
Much of our work with SAR imagery is still in the research and development phase, but we’re excited to start addressing the numerous verticals where this work will be applicable. For instance, retail shopping doesn’t stop when the sun goes down — especially during the busy and, often times, blustery holiday season. With SAR imagery, we’ll be able to count cars in parking lots at night, giving us a more complete picture of overall traffic. We’ll be able to monitor agriculture during monsoon season, a key part of the growing cycle. Disaster relief efforts will also benefit from SAR imagery, which will enable us to assess conditions on the ground more quickly, even if the storm clouds have not yet cleared.
We’ve only just begun exploring the possibilities that SAR imagery opens up, but we’re looking forward to continuing to aggressively pursue this data source. With the insights SAR imagery can provide, we’ll move even closer to our goal of complete socio-economic transparency for the world.