Sentinel-1
- Namrata Kolla
- Karen Farley
Sentinel-1 is a satellite managed by the European Space Agency. The raw data is freely available to the public. Skylight downloads the raw data, uses machine learning to detect vessels in the raw images, and displays the detected vessels in the platform. Skylight also correlates the detected vessels with AIS transmissions in the area to provide vessel data about the known vessels. Sentinel-1 is a great option for regularly monitoring dark vessels. Look here for a video of the Sentinel-1 satellites in action! https://www.esa.int/ESA_Multimedia/Videos/2014/03/Sentinel-1_constellation
Where does Sentinel-1 provide data?
Vessel detections from Sentinel-1 are loading for all of the areas where we have current Skylight users: Western and Eastern Indian Ocean, Caribbean, Gulf of Guinea, Southeast Asia, most of South America, and the Netherlands. Reference the diagram below to see whether your area is covered by Sentinel-1. Note Skylight only processes the areas where we also have Skylight users (listed above).
What is the revisit rate?
At minimum every 12 days, sometimes every 6 days. Reference the map above for how frequently this satellite revisits the same area.
How real-time is Sentinel-1 data?
Like any satellite-based data source, there is a delay - called “latency” - between when the image is collected by the satellite and when it appears in the satellite platform. The largest portion of this delay comes from waiting for the data to become available on the ESA website. There is ~15 minutes of delay added by Skylight downloading this data, processing it for vessels, and correlating it against our AIS feed.
The latency for Sentinel-1 varies by region. Averages can be as low as 2-3 hours in places like West Africa and as long as 8-10 hours in places like the West Pacific.
How do you detect vessels inside Sentinel-1 data?
Skylight uses a computer vision model to look for vessels inside Sentinel-1 data. Because this is a machine learning model, a vessel detection is not a guarantee that the object is a vessel. This can only be confirmed with eyes on the water. The computer vision model was trained by being given a lot of examples of what vessels in satellite radar look like, according to expert annotators who are familiar with maritime imagery. For more technical details about the computer vision model, see this page.
Each chip size (the black box containing the vessel) is 1280 meters by 1280 meters.