Arctic Vessel Routing and Ice-Condition Surveillance
SAR satellites track vessels and ice conditions simultaneously across the Northern Sea Route and Northwest Passage, where cloud cover and polar night make optical sensors nearly useless. Correlating Sentinel-1 imagery with spaceborne AIS exposes both route accessibility and unauthorised transits through sovereign Arctic waters.
Sensors
- Sentinel-1 SAR (C-band, ESA): 5 x 20 m resolution in Interferometric Wide Swath mode, 250 km swath, 6-day repeat at the equator but 1-2 day revisit above 70°N due to orbital geometry. Detects vessels as bright point targets and discriminates ice types by backscatter signature. Free and open data.
- RADARSAT-2 (C-band, MDA): Flexible imaging modes from 3 m Spotlight to 500 km ScanSAR. The Wide Ultra-Fine mode at 3 m resolves vessel superstructure detail useful for hull-type confirmation. Commercially tasked; used by the Canadian Ice Service for operational ice charting.
- Spire spaceborne AIS: Constellation of over 100 LEO satellites providing global AIS message collection with typical latency under 20 minutes. In high-traffic Arctic corridors, detection probability per vessel per pass is high; in remote stretches of the NSR it degrades due to message collision and sparse traffic density.
- AMSR2 passive microwave (JAXA): Aboard the GCOM-W1 satellite, AMSR2 measures sea-ice concentration at roughly 3-10 km resolution (frequency-dependent) with daily global coverage. Unaffected by cloud or darkness. Provides basin-scale ice-extent context that SAR imagery, with its narrower swaths, cannot supply alone.
Why radar is the only sensor that works here
The Arctic presents optical satellites with two compounding problems. Persistent low cloud covers the Beaufort Sea and the Kara Sea for the majority of the navigation season. In winter, polar night removes reflected sunlight entirely for months at a time. Multispectral sensors, however high their resolution, produce blank or cloud-obscured frames across precisely the stretches of water that matter most to route planners and coastal-state authorities.
C-band SAR at 5.4 GHz penetrates cloud and operates day and night by transmitting its own microwave pulses and recording the return. Sea ice and open water have markedly different dielectric properties: smooth first-year ice returns a weak backscatter signal, multi-year ice with its rougher surface returns a stronger one, and open water returns vary with wind-driven surface roughness. A steel vessel hull produces a very bright return, often ten to twenty decibels above the surrounding ice or water background, making detection tractable even at Sentinel-1's 20 m range resolution. The physics of the interaction is well understood; the interpretation challenge lies in separating vessel returns from ice ridges and icebergs, which can produce similarly bright point-like signatures.
Ice classification: what backscatter actually tells you
Operational ice charting agencies, including the Canadian Ice Service and Norway's MET Norway, have used SAR-based ice classification for decades. The standard approach assigns ice type categories from the World Meteorological Organisation's egg-code system based on backscatter intensity, texture, and spatial context. First-year ice, multi-year ice, new ice, and open water each occupy broadly distinct backscatter ranges in C-band imagery, though overlap between categories is real and seasonal melt complicates signatures between May and September.
AMSR2 passive microwave data fills the spatial-coverage gap. At 10 km resolution it cannot resolve individual floes or leads, but it delivers daily, cloud-free, basin-scale ice concentration that SAR imagery cannot match in coverage. The practical workflow combines both: AMSR2 defines the macro ice edge and concentration zones across the full NSR or Northwest Passage, then SAR imagery is tasked into specific corridors where route decisions are being made, providing the 20-metre-scale detail needed to assess lead width, ice thickness proxies, and ridging.
Vessel detection and the AIS correlation problem
A vessel transmitting AIS is straightforward to locate and identify. Spire's constellation collects AIS messages globally, and in Arctic waters the reduced message-collision problem that plagues dense shipping lanes actually improves per-message detection reliability, though the sparse satellite passes over high latitudes mean latency can reach 30-40 minutes in the most remote segments of the NSR.
The harder problem is the vessel that does not transmit, or transmits a false identity. SAR detects the physical hull regardless of transponder state. Correlating a SAR-detected bright point at a given position and time with the AIS record for that area either confirms a match or flags a discrepancy. A vessel present in SAR imagery but absent from AIS is, by definition, operating without a valid transponder signal. In Arctic sovereign waters, particularly the NSR (where Russia requires transit permits and icebreaker escort under Russian federal law) and the Northwest Passage (where Canada asserts internal waters jurisdiction), an unidentified vessel is a compliance and sovereignty concern, not merely an operational curiosity.
The detection floor is an honest constraint. Sentinel-1 at 20 m range resolution will miss small vessels reliably. A fishing boat under roughly 30-40 metres overall length may not produce a return distinguishable from background clutter, particularly in rough ice conditions. RADARSAT-2 in fine-resolution modes reduces that floor considerably, but commercial tasking has a cost and a planning lead time that open-access Sentinel-1 does not.
Route accessibility assessment: turning imagery into a decision
Ice concentration and ice type alone do not determine whether a vessel can transit. Hull class matters. A Polar Class 3 vessel can operate in medium first-year ice up to roughly 1.2 metres thick; a standard open-water tanker cannot safely enter any significant ice concentration. The analytic product that is actually useful to a maritime authority or route operator is not a raw SAR image but a classified ice chart overlaid with vessel track history, annotated with ice concentration thresholds keyed to the hull classes of vessels known to be operating in the corridor.
Revisit frequency is the binding constraint on timeliness. Sentinel-1's 1-2 day revisit above 70°N is adequate for strategic route assessment but not for real-time navigation support, where conditions can change within hours as leads open and close. Commercial SAR constellations with higher revisit rates, or coordinated multi-satellite acquisitions, are required for anything approaching operational navigation guidance. Satellize's Arctic analytics work combines open Sentinel-1 baselines with commercially tasked imagery on client licence, delivering updated ice-condition assessments at cadences matched to the operational tempo of the programme in question.
Sovereignty monitoring: the permit-compliance use case
Russia's Northern Sea Route Administration requires vessels to apply for transit permits and, for certain ice categories, to take on an icebreaker escort. Canada's position on the Northwest Passage asserts that these are internal Canadian waters requiring Canadian consent. Both regimes depend on knowing what is actually transiting. Satellite surveillance is the only scalable means of monitoring thousands of kilometres of remote waterway.
The analytic workflow for compliance monitoring is: generate a SAR-derived vessel detection list for a defined corridor and time window; cross-reference against AIS identity records and known permit databases; flag detections with no AIS match, or AIS identities not appearing on the permit list, for further investigation. False positives from ice features are reduced by requiring detections to persist across at least two SAR passes or to show Doppler-consistent motion signatures. The output is an alert list, not a verdict. Ground-truthing, legal determination, and interdiction remain the responsibility of the coastal state.
Honest limits and what they mean for programme design
Three limits matter most. First, SAR cannot measure ice thickness directly. It infers ice type from surface roughness and dielectric properties, which correlate imperfectly with thickness. Electromagnetic induction sensors on icebreakers or airborne platforms remain the ground truth for thickness. Second, cloud-free optical imagery, when available in summer, provides vessel classification detail that SAR cannot match: hull colour, deck equipment, flag markings. A well-designed Arctic surveillance programme treats SAR as the persistent baseline and optical imagery as the opportunistic supplement, not the reverse. Third, AIS coverage in the most remote NSR segments, particularly east of the Laptev Sea, is still patchy even with Spire's constellation. A vessel running dark in those segments may not be detected until a SAR pass is specifically tasked over that area.
These are not reasons to avoid satellite-based Arctic surveillance. They are reasons to design the programme with realistic expectations, layered sensors, and clear escalation paths when satellite evidence is ambiguous. For governments building or expanding Arctic maritime domain awareness, the satellite layer is the only cost-effective means of persistent wide-area coverage. The question is not whether to use it, but how to integrate it with patrol assets and legal frameworks that can act on what it finds.
Typical figures
| SAR spatial resolution (Sentinel-1 IW mode) | 5 x 20 m (range x azimuth) |
| SAR spatial resolution (RADARSAT-2 fine modes) | 3-8 m depending on mode selected |
| Sentinel-1 revisit above 70°N | 1-2 days (orbital geometry increases pass frequency at high latitudes) |
| AMSR2 ice-concentration resolution | 3-10 km (frequency-dependent); daily global coverage |
| Spire AIS typical latency | Under 20 minutes in well-covered corridors; up to 30-40 minutes in remote NSR segments |
| Minimum detectable vessel (Sentinel-1) | Approximately 30-40 m overall length in typical ice conditions; smaller in calm open water |
| SAR frequency | C-band, 5.4 GHz (both Sentinel-1 and RADARSAT-2) |
| Sentinel-1 archive depth | From April 2014 (Sentinel-1A launch); freely accessible via Copernicus Data Space |
| Delivery formats | GeoTIFF ice-classification rasters, GeoJSON vessel detection lists, PDF route-condition reports, alert feeds (JSON/API) |
Analytics Satellize can run
| SAR vessel detection list | Constant False Alarm Rate (CFAR) detection on calibrated SAR intensity imagery, with ice-clutter masking derived from concurrent ice classification | GeoJSON point layer of detected vessels with position, time, estimated length, and AIS match status |
| Ice-type classification chart | Supervised or threshold-based backscatter classification (first-year ice, multi-year ice, new ice, open water) following WMO egg-code categories; validated against AMSR2 concentration fields | GeoTIFF classified raster and polygon shapefile, updated per Sentinel-1 acquisition cycle |
| Route-accessibility assessment | Ice concentration and type thresholds cross-referenced against Polar Class hull ratings for vessels on the permit or AIS list; lead-width estimation from SAR | PDF corridor report with traffic-light accessibility rating per route segment and hull class |
| Dark-vessel compliance alert | SAR detection cross-correlated with Spire AIS record; unmatched detections flagged; persistence filter applied across consecutive passes to reduce ice-feature false positives | Alert report listing unidentified detections with SAR image chip, position, time, and nearest AIS vessel candidates |
| Historical transit density map | Aggregation of AIS track data and SAR detection records over a defined season or year; kernel density estimation on vessel positions | GeoTIFF density raster and summary statistics by corridor segment and vessel category |
| Ice-edge change time series | AMSR2-derived ice-extent boundary extracted at defined concentration thresholds (typically 15% and 80%) and differenced across dates | Animated GIF or video of seasonal ice-edge progression; CSV of ice-edge latitude by longitude per date |
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.