Fishing port catch-landing activity and fleet return patterns
Satellite SAR and optical imagery count vessels at berth and in anchorage at small fishing ports, producing fleet-return indices and catch-landing proxies where official landings data is sparse or delayed.
Sensors
- Sentinel-1 SAR (C-band, ESA): 10 m ground range resolution in Interferometric Wide Swath mode, 6-day repeat at mid-latitudes (12-day for a single satellite), cloud-independent. Detects vessels roughly 10 m and longer as bright point targets; shorter wooden or fibreglass hulls near the detection floor and may be missed in cluttered harbour backscatter.
- ICEYE SAR (X-band, commercial): Spotlight mode reaches approximately 1 m resolution, enabling length estimation of individual vessels down to about 5 m. Tasked on demand, so revisit is flexible but per-image cost applies. X-band is more sensitive to small metallic targets than C-band, improving detection of aluminium-hulled skiffs.
- Planet SuperDove (optical, commercial): 3 m resolution, up to daily revisit over most latitudes, eight spectral bands from coastal blue to near-infrared. Excellent for counting vessels in clear conditions and for distinguishing vessel colour and type. Cloud cover is the hard limit; tropical fishing ports can lose many days per month in rainy season.
- Global Fishing Watch VMS and AIS fusion: Provides track histories, flag state, gear type classification and fishing-effort hours for vessels that broadcast. Coverage is patchy for small-scale fleets: many vessels under 12 m carry no transponder at all, and VMS data sharing depends on national authority cooperation.
Why official landings data is not enough
Fisheries managers in most small-island and coastal-developing states receive catch-landings data days or weeks after the fact, if at all. Self-reported logbooks are inconsistent. Port sampling covers a fraction of landings. The result is that stock assessments rest on data that is both thin and stale.
Satellite observation does not replace biological sampling, but it provides something logbooks cannot: an independent, time-stamped count of how many vessels are in port on any given day. Fleet concentration at a berth is a reasonable proxy for the end of a fishing trip. Aggregate that signal over weeks and you get a seasonal pattern that correlates with effort and, indirectly, with catch.
What SAR actually sees in a small fishing harbour
A SAR image of a fishing port produces bright point targets against darker water. In Sentinel-1 IW mode at 10 m resolution, a vessel must present a radar cross-section large enough to exceed local clutter. Steel-hulled trawlers of 15 m and above are reliably detected. Wooden pirogues of 6 to 8 m are borderline: they appear in some images and vanish in others depending on sea state, orientation and hull moisture. This is not a flaw to hide; it is a systematic bias that must be characterised for each port before any count is treated as authoritative.
ICEYE Spotlight imagery at 1 m resolution changes the picture considerably. Individual vessel length can be measured to within roughly one to two metres, allowing a rough tonnage class to be assigned. Even so, vessels packed side by side at a crowded jetty merge into a single bright mass, and manual or semi-automated separation is needed. Analysts should report both a minimum count (distinct resolved targets) and an estimated total that accounts for the occluded fraction.
Fusing SAR counts with AIS and VMS tracks
For the subset of vessels that do broadcast, AIS and VMS tracks confirm port entry and departure times with minute-level precision. Matching a track arrival to the appearance of a new point target in the next SAR pass validates the detection geometry and gives a calibration anchor. Global Fishing Watch publishes vessel track data and fishing-effort layers that can be queried by port polygon.
The majority of small-scale fishing vessels worldwide carry neither AIS nor VMS. In those cases, SAR is the only systematic observation. The honest framing for a fisheries client is this: SAR gives you a fleet-concentration index, not a certified landings figure. Paired with periodic port sampling, that index becomes a multiplier that converts a count into an estimated landed weight, with an uncertainty range that should always be stated explicitly.
Seasonal and weekly rhythms as an analytical signal
Fishing fleets are not random. Most small-scale fisheries have pronounced weekly rhythms driven by market days, and seasonal rhythms driven by monsoon, spawning closures or migratory species. A time series of vessel counts at berth, built from every available Sentinel-1 pass over twelve months, will reveal those rhythms clearly enough to distinguish a normal low-effort week from an anomalous one.
Anomalies matter. A port that is unusually empty during a period when it should be full may indicate a stock collapse, a fuel shortage, a weather event or a regulatory closure. A port that is unusually full during an open season may indicate poor market conditions or a disease event affecting buyer attendance. None of these interpretations can be confirmed from imagery alone, but the signal prompts the right questions at the right time, which is more than a quarterly logbook summary can offer.
Revisit frequency sets the resolution of this time series. Sentinel-1's six-day repeat at mid-latitudes is adequate for weekly pattern detection. For daily rhythm analysis, Planet SuperDove is needed, with the caveat that cloud gaps break the series in tropical climates.
Honest limits of the method
Vessel detection in SAR degrades in high sea states, when wave clutter raises the noise floor. It also degrades when vessels are moored under corrugated-iron roofing structures, common in South-East Asian and West African ports, which attenuate the radar return. Optical imagery resolves the roofing problem but reintroduces cloud dependency.
Length estimation from SAR shadow geometry works best in Spotlight mode with a known incidence angle and a clear target separation. It fails for vessels moored parallel to the range direction, where the shadow falls behind the vessel rather than beside it. Published studies using Sentinel-1 for fishing vessel detection report detection probabilities of around 70 to 90 percent for vessels above 15 m, falling sharply below that threshold. Clients should treat any count as a lower bound and calibrate against at least one ground-truth survey per port per year.
Satellize applies this methodology on open Sentinel-1 and Planet data, with commercial ICEYE tasking added where a client needs sub-weekly revisit or length-class resolution. The Tonga crop-estimation programme demonstrated that index-based proxies, properly calibrated, are actionable for government planning even when direct measurement is impractical.
What a fisheries client actually receives
The practical output is a time-series dashboard showing vessel count at berth, vessel count in anchorage, and a fleet-concentration index for each port of interest, updated on each satellite pass. Alerts flag departures from the seasonal baseline beyond a configurable threshold. A monthly summary report maps the index against any available VMS effort data and notes periods of cloud or SAR degradation that reduce confidence.
This is not a stock-assessment tool. It is an early-warning and monitoring layer that a fisheries authority can use to schedule port inspections, validate self-reported logbooks and brief ministers on fleet activity without waiting for the next quarterly survey. That is a narrow claim, but it is a defensible one.
Typical figures
| SAR spatial resolution (Sentinel-1 IW) | 10 m ground range, 20 m azimuth (multi-looked product) |
| SAR spatial resolution (ICEYE Spotlight) | Approximately 1 m, enabling individual vessel length measurement |
| Optical resolution (Planet SuperDove) | 3 m, 8 spectral bands (coastal blue to near-infrared) |
| Revisit (Sentinel-1, single satellite) | 6 days at mid-latitudes; up to 12 days near equator depending on orbit geometry |
| Revisit (Planet SuperDove) | Up to daily, subject to cloud cover and tasking priority |
| Minimum detectable vessel (Sentinel-1 IW) | Approximately 10 to 15 m hull length in calm sea state; smaller vessels unreliable |
| Cloud impact | SAR: none. Optical: tropical ports may lose 30 to 60 percent of passes in wet season |
| Archive depth (Sentinel-1) | From 2014 for most regions; enables multi-year seasonal baseline construction |
| AIS/VMS coverage for small-scale fleets | Typically less than 20 percent of vessels under 12 m carry any transponder |
| Delivery formats | GeoTIFF vessel-detection layers, GeoJSON point features, CSV time-series, PDF monthly summary |
Analytics Satellize can run
| Vessel count at berth and in anchorage | SAR constant false alarm rate (CFAR) detection on Sentinel-1 IW or ICEYE Spotlight; optical object detection on Planet SuperDove where cloud-free | Per-pass GeoJSON point layer with vessel count and confidence flag; time-series CSV |
| Fleet-concentration index | Normalised vessel count relative to port-specific historical baseline derived from Sentinel-1 archive from 2014 onwards | Weekly index value per port, delivered as dashboard feed or CSV; anomaly alerts when index exceeds ±1.5 standard deviations |
| Vessel length-class distribution | SAR shadow-length measurement in ICEYE Spotlight imagery at known incidence angle; vessels binned into small (under 10 m), medium (10 to 20 m) and large (above 20 m) classes | Length-class histogram per image pass; GeoJSON with per-vessel length estimate and uncertainty |
| AIS/VMS-to-SAR match rate | Spatial and temporal join of Global Fishing Watch track arrivals to SAR detection events within port polygon; match rate quantifies the untracked fraction | Monthly calibration report stating matched and unmatched vessel counts; used to scale total fleet estimate |
| Seasonal and weekly rhythm profile | Fourier decomposition of vessel-count time series to extract dominant periodicities; comparison across years to identify trend shifts | Annual seasonality chart per port; anomaly flags for departures from expected pattern |
| Cloud-gap and data-quality log | Automated cloud-mask assessment on each optical pass; SAR sea-state flag based on wind speed from ECMWF ERA5 reanalysis at image acquisition time | Per-pass quality flag appended to all data products; monthly summary of effective observation days |
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.