SAR Vessel Detection and Fixed Infrastructure Pipeline Restored

Update — September 14, 2026: Data feeds for SAR vessel detection and fixed infrastructure have resumed, now powered by Copernicus Sentinel-1C and Sentinel-1D imagery. Data is fully backfilled without gaps from June 30, 2026, onward. 

What changed?

Our Sentinel-1 vessel detection and fixed infrastructure pipelines rely on a CFAR (constant false alarm rate) algorithm, which requires per-satellite sensitivity calibration. Because Sentinel-1C and 1D feature newer imaging systems than Sentinel-1A, existing settings could not be directly reused. We recalibrated the CFAR parameters by leveraging an overlap period when all three satellites were actively acquiring imagery, benchmarking 1C and 1D against 1A across three key performance metrics: 

• • Recall: The percentage of AIS-broadcasting vessels we successfully detect.
  • Noise rate: How often the detector flags something that is not a vessel. We approximate this by counting small detections without a matching AIS signal in quiet stretches of open ocean, where we would expect to find almost none.
  • AIS-match rate: The proportion of global detections matched to AIS data.

Based on these measures, we have set new CFAR parameters for both Sentinel-1C and Sentinel-1D.

What does this mean for your data? 

Compared with Sentinel-1A, the new satellites demonstrate comparable or marginally better detection of small vessels (under 30 meters) and a consistent high-seas noise rate. We found no systematic differences from Sentinel-1A performance across overlapping coverage areas.

An important caveat

This calibration rests on roughly two months of overlapping imagery. Most regions received only a few overpasses during that time window, which may not provide enough data for these statistics to stabilize. While our analysis indicates the current settings are sufficient — we observed no apparent issues at global or regional levels — we will continue to monitor as more Sentinel-1C and 1D data accumulates. We may adjust the CFAR parameters in the future, and any further changes affecting the dataset will be announced here.

Questions?

We are committed to providing the most accurate open-access data available. For technical questions about specific datasets or how this may affect your work, please contact [email protected].