Road-mobile ballistic missile launcher dispersal tracking
Transporter-erector-launchers are designed to disappear into terrain, but high-revisit optical and SAR imagery can bound their probable locations during dispersal. The method is probabilistic, not continuous, and fails completely under canopy or inside tunnels.
Sensors
- Planet SkySat: Optical imagery at approximately 0.5 m ground sample distance, capable of resolving the elongated chassis of a TEL (typically 12–18 m in length) and distinguishing it from civilian heavy transport. Revisit is non-deterministic but Planet's tasking network can achieve same-day collects over a priority area. Useless in cloud or at night.
- Maxar WorldView-3: 0.31 m panchromatic, 1.24 m multispectral. The best publicly available optical resolution for confirming vehicle type, counting axle groups and reading convoy composition. Single-pass; revisit over a specific point averages one to four days depending on latitude and tasking priority. Cloud remains the dominant operational constraint.
- ICEYE SAR constellation: X-band synthetic aperture radar with spotlight-mode resolution down to approximately 0.25 m (single-look). Operates day and night, through cloud and light precipitation. Large wheeled vehicles produce strong radar returns from corner reflectors formed by their chassis and wheels. Revisit over a tasked area can be under six hours across the constellation. Cannot see inside tunnels or under dense canopy.
- Capella Space SAR: X-band spotlight SAR with published 0.35 m resolution in its highest-resolution mode. Comparable cloud-penetration and night-operation capability to ICEYE. Particularly useful for detecting the characteristic elongated bright return of a TEL erector arm when partially raised. Revisit similar to ICEYE when both constellations are combined.
What a TEL looks like from 500 km up
A road-mobile ballistic missile launcher is a large, slow, distinctive object. North Korean Hwasong-series TELs, for instance, run to roughly 20 m in length on multi-axle chassis. At WorldView-3 resolution, the axle count is legible. At SkySat resolution, the elongated profile is unambiguous against the road surface. In SAR imagery, the metal chassis and erector arm create strong, characteristic double-bounce and specular returns that stand out from civilian lorries.
The public methodology for tracking these vehicles was developed and published openly by 38 North and the Middlebury Institute of International Studies, applied primarily to North Korean TEL movements visible in commercial imagery. Their work established that vehicle geometry, shadow length, convoy composition and the presence of support vehicles (fuel tankers, command vehicles, security escorts) together allow confident type identification even without sub-metre resolution. That published analytical framework is the foundation of everything described on this page.
The dispersal problem: from garrison to unknown
TELs are most detectable at garrison, where they park in known locations and can be counted against baseline imagery. Dispersal is the hard part. A vehicle leaving a garrison at any point between two satellite passes could be anywhere within a radius defined by its road speed and the time elapsed. A TEL travelling at 40 km/h for six hours occupies a search area of roughly 50,000 square kilometres if the road network fans out. That is not a tracking problem; it is a search problem.
High-revisit constellations reduce the uncertainty cone. If SkySat or an ICEYE pass catches a convoy on a specific road segment, the next-pass window constrains the probable onward position to a much smaller arc. The analytic product is therefore a probability distribution over likely locations, updated with each new collect. It is never a continuous track. Analysts at 38 North have been explicit about this limitation in their published work, and intellectual honesty demands the same candour here.
Support convoy signatures help. A TEL rarely travels alone. Fuel tankers, command vehicles and security elements create a convoy footprint longer than the TEL itself, increasing the detectable area and providing additional confirmation of vehicle type. The presence of a convoy on a road that has no civilian reason for heavy military traffic is itself a significant indicator.
Where the method breaks down
North Korea's tunnel network is extensive and purpose-built for exactly this problem. A TEL that enters a tunnel mouth disappears from all spaceborne sensors until it re-emerges. Dense forest canopy has the same effect for optical sensors. X-band SAR can penetrate light vegetation but not a full forest canopy at military-standard concealment depths. Any honest assessment of TEL location must include a probability mass assigned to 'under cover and undetectable'.
Weather is a secondary constraint for optical sensors. Cloud cover over the Korean Peninsula averages above 60 percent in summer months. SAR mitigates this but does not eliminate all gaps. A well-timed dispersal during a multi-day overcast period, combined with movement into forested or tunnelled terrain, can produce a complete loss of track. The method provides bounds on where a TEL probably is; it cannot confirm where it definitely is not.
SAR versus optical: not a competition
The two sensor types answer different questions. Optical imagery at sub-metre resolution gives the analyst the richest information for type confirmation: axle groups, erector arm configuration, vehicle markings, shadow geometry for height estimation. SAR gives coverage at night and through cloud, and the radar cross-section of a large metal vehicle is large enough to be detected reliably even at moderate SAR resolutions.
In practice, the most useful monitoring architecture fuses both. SAR passes provide the high-cadence alert layer, flagging anomalous vehicle-sized returns on roads near known garrison sites. Optical tasking is then triggered to confirm type and count convoy elements. ICEYE and Capella together can provide revisit intervals under six hours over a priority zone; WorldView-3 or SkySat then image on demand when cloud permits. The latency from SAR detection to optical confirmation depends entirely on cloud conditions and tasking queue, and can range from hours to days.
What an analyst actually delivers
The output of a TEL dispersal monitoring programme is not a real-time track. It is a series of time-stamped position reports, each with an associated confidence level and a probability envelope for the vehicle's subsequent location given road network constraints. When a vehicle re-enters a known site or is re-acquired on a road, the envelope collapses. When it enters a tunnel or canopy zone, the envelope expands until the next confirmed sighting.
Satellize structures this kind of analysis as a tasked-imagery alert service, pulling from commercial SAR constellations on client licence and running change-detection against a baseline road-network model. For clients requiring a standing watch on a specific garrison complex, the practical starting point is a baseline imagery archive review to establish normal vehicle counts and parking patterns before any monitoring cadence is agreed.
Arms control and the open-source record
The same methodology that supports national intelligence assessments is now partially reproducible from commercial imagery alone. The 38 North and Middlebury work demonstrated this publicly, and the implications are significant for treaty verification. An analyst with access to Planet, ICEYE and Capella can, in principle, monitor declared TEL garrison sites for compliance with any future arms-control agreement that caps vehicle numbers or restricts dispersal zones.
The honest limit is the same as for intelligence use: the method counts and tracks what is visible. It cannot verify what is hidden. A declared count of launchers can be cross-checked against visible vehicles at garrison; it cannot confirm that no additional vehicles exist inside tunnel complexes. That gap is not a failure of the satellite method; it is an inherent property of any verification regime that relies on remote sensing without ground inspection.
Typical figures
| Best optical resolution (WorldView-3 panchromatic) | 0.31 m ground sample distance |
| Best SAR resolution (ICEYE / Capella spotlight) | 0.25–0.35 m single-look |
| Optical revisit (WorldView-3, tasked) | 1–4 days depending on latitude and priority |
| SAR revisit (ICEYE + Capella combined constellation) | Under 6 hours over a priority area |
| Minimum detectable vehicle length (optical) | Approximately 5 m at 0.5 m GSD; TEL chassis (12–20 m) well above threshold |
| Spectral bands (optical) | Panchromatic plus 8-band multispectral (WorldView-3); panchromatic (SkySat) |
| SAR frequency | X-band (9.6 GHz) for both ICEYE and Capella |
| Latency from tasking to delivery | SAR: 2–6 hours post-pass; optical: hours to days depending on cloud |
| Archive depth (commercial optical) | WorldView-3 archive from 2014; Planet archive from 2016 |
| Coverage limitation | No detection under tunnel or dense forest canopy; optical blocked by cloud |
Analytics Satellize can run
| Garrison baseline vehicle count | Object detection on high-resolution optical imagery using template matching against known TEL silhouette geometry; shadow analysis for height confirmation | Annotated imagery report with vehicle count, parking-bay occupancy map and confidence rating; GIS layer of detected positions |
| Dispersal-event alert | SAR change detection against baseline road-network model; anomalous large-vehicle return triggers optical tasking request | Time-stamped alert with SAR chip, road-segment identifier and estimated convoy speed derived from multi-pass interval |
| Probabilistic location envelope | Road-network graph traversal from last confirmed position, weighted by vehicle speed range and tunnel/canopy exclusion zones; updated with each new confirmed sighting | Probability-density map (GeoTIFF or shapefile) showing likely TEL location bounds at 50 percent and 90 percent confidence, with timestamp and assumptions stated |
| Convoy composition analysis | Multi-vehicle detection and classification on optical imagery; support-vehicle type identification (fuel tanker, command vehicle, security escort) using published convoy signature typology from open-source literature | Structured report listing vehicle types, count, order of march and inferred mission phase (transit, pre-launch preparation, return) |
| Historical dispersal pattern analysis | Time-series review of archive imagery at known garrison sites; frequency and seasonality of dispersal events extracted from change-detection log | Trend report with dispersal frequency by month, preferred road corridors identified, and comparison against publicly documented exercise calendars |
| Treaty-compliance garrison count | Periodic optical and SAR collect against declared garrison coordinates; vehicle count compared against declared launcher inventory | Compliance summary table with imagery evidence, discrepancy flags and honest statement of unverifiable tunnel/covered-storage fraction |
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.