IUU fishing detection inside exclusive economic zones
Combining SAR dark-vessel detection with spaceborne AIS and VIIRS night-light data lets coastal states identify fishing activity inside their EEZs that never appears in official catch records. Each method has hard limits; used together, they narrow the ambiguity considerably.
Sensors
- Sentinel-1 SAR (C-band, ESA): 5 x 20 m resolution in Interferometric Wide Swath mode, 250 km swath, roughly 6-day revisit at the equator (shorter at higher latitudes). Detects vessels as bright radar cross-section returns independent of cloud or daylight. Doppler centroid anomaly in the azimuth direction provides a line-of-sight velocity estimate for moving vessels, useful for distinguishing active gear deployment from drifting.
- RADARSAT Constellation Mission (C-band, CSA): Three identical satellites providing daily revisit over Canada's EEZ and sub-daily passes at high latitudes. Medium-resolution ScanSAR modes reach 50 m; spotlight modes reach 3 m. The constellation's daily cadence is operationally significant for tracking vessel movements across a 24-hour window.
- Spire Global spaceborne AIS: Over 100 LEO satellites receiving Class A and Class B AIS transponder messages globally. Message latency to ground can be under 30 minutes. Detection rates in congested areas are imperfect due to signal collision, but Spire's published revisit is sub-hourly over most ocean regions. Vessels without an AIS signal that appear in SAR are the primary IUU indicator.
- VIIRS Day/Night Band (NOAA/NASA, Suomi-NPP and NOAA-20): 750 m nadir resolution, single nightly overpass per satellite. Detects vessel deck lighting at sea with a published minimum radiance threshold around 2 x 10⁻⁹ W/cm²/sr/µm under low-moonlight conditions. Particularly effective for squid and other light-luring fisheries; less useful for vessels running dark. Cloud cover is a hard limit: a single overpass with heavy cloud cover yields nothing.
What the FAO framework actually asks of a coastal state
The FAO International Plan of Action to Prevent, Deter and Eliminate Illegal, Unreported and Unregulated Fishing (IPOA-IUU), adopted in 2001, places the primary burden of EEZ monitoring on the coastal state. A state must be able to demonstrate that it is actively monitoring its waters, that it has a system for recording vessel activity, and that it can produce evidence sufficient to support a boarding, citation or port-state denial. Satellite data does not replace a patrol vessel or a legal process, but it creates the documented trail of positions, behaviours and timestamps that makes enforcement credible.
IUU fishing is not a single offence. It includes fishing without a licence, fishing in closed seasons or areas, exceeding catch quotas, using prohibited gear, and failing to report catches. Each sub-category has a different satellite signature. Gear type matters: longliners behave differently from purse-seiners, and both differ from trawlers. A detection workflow that treats all slow-moving vessels as equivalent will generate too many false positives to be operationally useful.
What a SAR image actually reveals about a fishing vessel
A synthetic aperture radar image does not show you a photograph of a vessel. It shows a bright return caused by the vessel's metal structure reflecting C-band microwave energy back to the sensor. The intensity of that return depends on the vessel's radar cross-section, which correlates loosely with size and construction material. A 60-metre steel-hulled trawler is reliably detectable in Sentinel-1 IW mode. A 10-metre fibreglass artisanal pirogue in a moderate sea state is frequently not. This is the single most important honest limit in EEZ IUU work: the vessels most likely to be fishing illegally in small island developing states are often the vessels least likely to appear in the data.
Where SAR adds real discrimination beyond simple detection is in the Doppler centroid anomaly. Because SAR is a coherent sensor, a vessel moving along the satellite's line of sight introduces a measurable frequency shift in the returned signal. This shift can be inverted to estimate the vessel's radial velocity component. Fishing vessels deploying or hauling gear typically move at 1 to 4 knots; transiting vessels move at 8 to 14 knots. A vessel sitting at 2 knots with no AIS signal, 80 nautical miles inside an EEZ boundary, is a different analytical problem from a vessel at 12 knots on a straight heading. The Doppler method is published in peer-reviewed literature and implemented in ESA's SNAP toolbox, though it requires careful calibration and is not a plug-and-play product.
The dark-vessel problem and why AIS correlation is the first filter
A vessel is 'dark' in the AIS sense when no transponder signal is received for a period inconsistent with normal transmission intervals. Class A transponders, mandatory on vessels over 300 gross tonnes in international waters, are required to transmit every 2 to 10 seconds when under way. A gap of more than a few minutes in open ocean, away from known AIS collision zones, is anomalous. Spire's constellation collects these gaps systematically.
The analytical workflow runs in two stages. First, SAR detections are extracted from the image using a constant false alarm rate (CFAR) detector, which adapts its threshold to local clutter and sea state. Second, each detection is matched against the AIS position record for the same time window, with a position uncertainty buffer that accounts for AIS message latency and vessel speed. Detections with no AIS match are flagged as dark. Not all dark vessels are IUU: naval vessels, some research ships and vessels with faulty equipment also go dark. Behaviour over time, specifically repeated appearances in the same fishing ground, loitering patterns and proximity to known fishing areas, is what converts a dark detection into a credible IUU indicator.
The artisanal vessel gap: where satellite data runs out
Pacific island states, West African coastal nations and parts of Southeast Asia face a specific version of this problem. The vessels most likely to be fishing illegally inside their EEZs are small, wooden or fibreglass, unregistered, and entirely below the AIS mandate threshold. They do not appear in SAR with any reliability below roughly 10 to 15 metres in calm conditions, and they carry no transponder. VIIRS can detect them at night if they use bright deck lights, but many artisanal vessels do not.
There is no satellite solution that fully closes this gap today. Very high-resolution commercial SAR (for example, ICEYE or Capella at sub-metre resolution) can detect smaller vessels, but tasking costs and revisit rates make systematic EEZ surveillance at that resolution economically impractical for most small states. The honest position is that satellite data is most powerful for monitoring the larger, licensed or semi-licensed vessels that account for the greatest volume of unreported catch, and for building the pattern-of-life evidence that supports port-state measures against flag-of-convenience fleets.
Building a surveillance product a fisheries ministry can actually use
A raw SAR detection list is not a fisheries intelligence product. Turning it into something a patrol coordinator can act on requires several additional steps: geocoding detections to a common projection, assigning confidence scores based on vessel size estimate and AIS correlation status, cross-referencing against the coastal state's own licensing register, and flagging vessels with a history of prior detections in the same area. The output should be a ranked alert list with supporting imagery, not a map with hundreds of undifferentiated dots.
Satellize structures analytics of this kind around open constellation data, primarily Sentinel-1 and VIIRS, with commercial AIS from providers such as Spire added on client licence. The Tonga crop-estimation programme demonstrated that open satellite data, processed consistently over time, can support government decision-making in a small island state context. The same principle applies to fisheries: consistent methodology, documented uncertainty, and outputs calibrated to what an enforcement agency can realistically act on.
Latency matters operationally. Sentinel-1 data is typically available in the Copernicus Data Space within 1 to 3 hours of acquisition. A workflow that ingests, processes and delivers alerts within 6 hours of overpass gives a patrol vessel a realistic intercept window for a vessel still within the EEZ. Longer pipelines produce historical intelligence, which is still valuable for licensing decisions and diplomatic pressure, but it is not the same as actionable surveillance.
Typical figures
| SAR spatial resolution (Sentinel-1 IW mode) | 5 x 20 m (range x azimuth); multi-looked product ~10 x 10 m |
| SAR revisit (Sentinel-1 at equator) | ~6 days single satellite; ~3 days with both Sentinel-1A and 1C combined |
| SAR revisit (RADARSAT Constellation Mission) | Daily over Canada; sub-daily at latitudes above 50°N |
| Minimum detectable vessel size (Sentinel-1 CFAR) | Approximately 15–20 m length in moderate sea state (Beaufort 3–4); smaller in calm conditions, unreliable below ~10 m |
| AIS revisit (Spire LEO constellation) | Sub-hourly over most ocean regions; message latency to ground under 30 minutes typical |
| VIIRS DNB resolution | 750 m nadir; one overpass per satellite per night |
| VIIRS minimum detectable radiance | ~2 x 10⁻⁹ W/cm²/sr/µm under low-moonlight conditions (published NOAA/NASA figures) |
| Sentinel-1 data latency (Copernicus Data Space) | Typically 1–3 hours post-acquisition |
| SAR archive depth (Sentinel-1) | From April 2014 (Sentinel-1A launch); global coverage not uniform in early years |
| Doppler velocity measurement accuracy (Sentinel-1) | Line-of-sight component; typical uncertainty ±0.5–1.0 m/s depending on sea state and processing |
Analytics Satellize can run
| Dark-vessel alert list | CFAR SAR vessel detection cross-correlated with spaceborne AIS position records; unmatched detections flagged by confidence tier | Ranked alert feed (GeoJSON or CSV) delivered within 6 hours of SAR acquisition, with vessel size estimate, position, and AIS match status |
| Gear-deployment behaviour classification | Doppler centroid anomaly extraction from Sentinel-1 SLC data to estimate vessel radial velocity; slow-loiter signatures (1–4 knots) separated from transit signatures | Per-detection velocity annotation appended to the dark-vessel alert list; vessels with loiter signatures flagged for priority review |
| EEZ incursion pattern-of-life report | Multi-date SAR detection stacking over a rolling 30- or 90-day window; spatial clustering to identify persistent fishing grounds and repeat offenders | Monthly PDF report with annotated maps, vessel frequency histograms by grid cell, and a ranked list of vessels with three or more appearances |
| Licensing register cross-check | AIS MMSI and vessel name matching against the coastal state's own vessel licensing database; detection of vessels operating outside their licensed area or season | Flagged vessel table with licence status, detected position and licensed area boundary, formatted for fisheries ministry case files |
| Night-light fishing activity index | VIIRS DNB radiance extraction over EEZ boundary; vessel-class light sources separated from platform and ship navigation lights using published radiance thresholds | Nightly raster layer (GeoTIFF) of detected light sources with intensity values, plus a monthly time-series chart of fishing effort by EEZ sub-zone |
| Patrol tasking priority grid | Fusion of dark-vessel frequency, Doppler loiter score and VIIRS night-light density into a weighted risk surface by 0.1° grid cell | Weekly priority grid (GeoTIFF and web tile) for patrol vessel routing, with documented weighting methodology for legal defensibility |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.