Questions
Frequently Asked Questions
What Seahound is, how the data is produced, and where the methodology stops.
- What is Seahound?
- Seahound is a personal project that processes open Sentinel-1 SAR satellite imagery to estimate vessel congestion at a handful of global maritime chokepoints — Suez, Hormuz, Singapore, Panama, Shanghai, and LA/Long Beach — and traces how that congestion might affect commodity supply chains. It is a work in progress, not a commercial product.
- What does 'semi-automated' mean?
- The processing pipeline — downloading the latest Sentinel-1 scene, running the detector, computing a risk snapshot, and writing it to the database — is triggered manually, not continuously. Updates happen when a new scene is available and the pipeline has been run against it. There will be periods where displayed risk scores are several days stale. This is a known limitation of the current development stage.
- How often is data updated?
- Sentinel-1 revisits each location roughly every 6–12 days depending on latitude and orbital configuration. Not every pass produces a usable scene (cloud cover doesn't affect SAR directly, but rain cells and high wind can). In practice, expect updates at best every few days per zone, and sometimes less frequently.
- What are 'likely vessel detections'?
- The detector finds bright connected components in the SAR image — high-backscatter regions consistent with large metal objects on water. Ships are the most common cause, but offshore platforms, buoys, and processing artefacts can also trigger detections. The algorithm is intentionally conservative: it is designed as an MVP signal to estimate relative congestion, not to identify or count individual vessels with precision.
- What do the risk levels (Low / Moderate / High / Severe) mean?
- Risk levels are derived from a simple formula: the ratio of current detected targets to a static per-zone baseline count. A positive anomaly (more targets than baseline) pushes the score up; a negative one pulls it down. The buckets — Low, Moderate, High, Severe — are thresholds on that 0–100 score. They are not calibrated against historical congestion events and should be read as directional signals, not precision assessments.
- Can a 'Low' risk reading mean conditions are actually bad?
- Yes. SAR vessel detection fails when wind speeds exceed roughly 12 m/s — wave breaking suppresses the contrast between vessels and sea surface. A low detection count under high-wind conditions may reflect a sensor limitation rather than genuine low congestion. Always treat a low reading during known adverse weather with caution.
- What is Sentinel-1?
- Sentinel-1 is a pair of C-band synthetic aperture radar satellites operated by the European Space Agency as part of the Copernicus Earth Observation programme. The imagery is freely available under an open data licence. Seahound uses Sentinel-1 GRD (Ground Range Detected) scenes, accessed via the Element84 Earth Search STAC API on AWS Open Data.
- Is any of this financial or trading advice?
- No. Nothing on Seahound constitutes financial, investment, trading, or operational maritime advice. The data is experimental, frequently stale, and produced by an MVP processing pipeline. Do not make decisions based on it.
- Where can I learn more about the methodology?
- The Kharg Island intelligence brief walks through a real SAR detection run in detail — including the physics of why oil and vessels appear as they do in radar imagery, and an honest account of what two images can and cannot tell you. The About page covers the project status and data limitations.