Pre-IPO physical asset verification from satellite imagery
Satellite imagery lets analysts independently verify whether declared factories, warehouses and land holdings exist, match their stated scale, and show signs of genuine operation before an equity listing or acquisition closes.
Sensors
- Maxar WorldView-3: 0.31 m panchromatic resolution, 1.24 m multispectral, 8-band VNIR plus SWIR. Revisit roughly 1 day at mid-latitudes due to off-nadir agility. The sharpest routinely available commercial optical sensor; can resolve individual vehicles, roof-mounted HVAC units and loading-dock occupancy.
- Airbus Pléiades Neo: 0.30 m panchromatic, 1.2 m multispectral, four-satellite constellation giving same-day or next-day revisit over most targets. Stereo and tri-stereo collection modes allow 3-D building height estimation, useful for cross-checking declared floor-area figures against volumetric footprint.
- Planet SkySat: 0.50 m panchromatic, approximately 1 m multispectral. A fleet of 21 satellites supports multiple collections per day over a single site when tasked in sequence, enabling time-lapse analysis of vehicle presence across a working day rather than a single snapshot.
- ICEYE X-band SAR: Spotlight mode achieves approximately 0.25 m range resolution. SAR penetrates cloud and operates at night, making it useful in tropical or monsoon-affected jurisdictions where optical windows are rare. Coherent change detection between two passes can flag whether a facility has been modified or left idle between visits.
What a prospectus cannot fake at 30 centimetres
A listing document can describe a factory in precise, audited-sounding language and still be wrong. The building may be leased to a third party, partially constructed, or simply not the size stated. Very-high-resolution optical imagery from WorldView-3 or Pléiades Neo resolves rooftop area to within a few percent when orthorectified against a reference digital elevation model. Cross-referencing declared gross floor area against measured rooftop footprint, combined with building height from stereo imagery, produces an independent volume estimate that either corroborates or contradicts the prospectus figure.
The check is not infallible. A multi-storey facility with a small footprint can be misread if stereo collection is not commissioned. Covered or underground storage is invisible to optical sensors entirely. The honest position is that imagery confirms what is visible from above; it does not confirm what is inside, and it cannot audit lease arrangements or ownership structures. Used correctly, it narrows the range of plausible interpretations rather than delivering a verdict.
Activity signals: what parked lorries do and do not prove
Vehicle presence at loading docks is the most commonly cited proxy for operational activity, and it is genuinely informative. A facility with 40 heavy goods vehicles queued at bays at 10:00 local time is almost certainly processing throughput. An identical facility with an empty car park and no vehicles on a weekday suggests something different.
The risk is snapshot bias. A single image taken at an atypical moment, a public holiday, a maintenance shutdown, or simply a slow Tuesday, can make an active site look idle and vice versa. Planet SkySat's multi-pass capability partially addresses this: tasking the same site three or four times across a single day gives a time-series of vehicle counts that is harder to game and harder to misread. Even so, vehicle presence measures activity on the day of collection, not average throughput over a quarter. Analysts should treat it as a consistency check against declared revenues, not as a direct revenue estimate.
Thermal infrared adds another layer. Operational industrial processes, kilns, furnaces, large refrigeration plant, emit heat signatures detectable by sensors such as Landsat 8/9 TIRS (100 m resolution) or, at finer scale, by airborne thermal where regulations permit. A declared food-processing plant that shows no thermal anomaly during operating hours warrants a question. The limit here is resolution: TIRS at 100 m cannot isolate a single building within a dense industrial estate.
Change detection as a timeline audit
Archive imagery stretching back to 2009 in the WorldView constellation, and earlier in Landsat's public record, allows analysts to reconstruct a facility's construction and operational history. A company claiming a factory has been operating for eight years can be tested against eight years of imagery. This matters because some prospectus misrepresentations involve backdating: presenting a recently built or recently activated facility as a long-established asset.
SAR change detection is particularly useful here. ICEYE coherent change detection, or Sentinel-1 interferometric analysis at coarser 5 m resolution, can identify whether surface materials at a site have changed between two passes separated by days or weeks. A warehouse floor that has been disturbed, a yard that has been resurfaced, or a roof that has been replaced all produce a measurable coherence drop. This is not conclusive evidence of fraud, but it is a flag that warrants ground-truthing.
Jurisdictions where the camera is legally limited
Commercial imaging resolution is not uniformly available worldwide. The United States historically restricted commercial sales of sub-0.25 m imagery under the NOAA licensing framework, though Maxar received approval to sell 0.25 m products commercially in 2014 and 0.31 m products subsequently. Other countries impose their own shutter-control regimes: India restricts foreign satellite imaging of certain areas, and several Gulf states have bilateral agreements that affect what operators will task and deliver.
South Korea restricts domestic distribution of imagery below 50 cm resolution of its own territory, which has practical implications for due diligence on Korean manufacturing assets. Israel applies national security review to imagery of its territory regardless of the collecting satellite's nationality. Analysts working across multiple jurisdictions need to map the regulatory constraint before commissioning a tasking campaign, not after. Where high-resolution optical is unavailable or delayed, SAR from a non-US operator such as ICEYE (Finnish) or Capella Space (US, with its own licensing) may offer a partial substitute, though SAR interpretation for facility verification requires more specialist handling than optical.
Structuring a verification campaign for a real transaction
A credible pre-IPO physical asset verification programme typically proceeds in three phases, though the sequencing should be adapted to the transaction timeline. First, open-source and archive screening: Sentinel-2 at 10 m and Landsat at 30 m are free, globally consistent, and sufficient to confirm that a large facility exists and has been present for years. This phase costs almost nothing and eliminates obvious fabrications.
Second, targeted high-resolution tasking of the ten to twenty sites that matter most to the valuation. This is where WorldView-3 or Pléiades Neo collection is commissioned, ideally on multiple dates to capture vehicle presence across different days. Stereo collection should be specified for any site where building height is a material input to floor-area calculations.
Third, anomaly investigation. Any site that fails the first two phases, or produces ambiguous results, is escalated to SAR time-series analysis and, where feasible, cross-referenced against AIS vessel data (for port-adjacent facilities), night-light radiance from VIIRS, or thermal data from Landsat TIRS. Satellize structures analytics campaigns of this kind, drawing on open constellations and commercial tasking under client licence. The output is a site-by-site assessment report with imagery evidence, not a single headline score.
The honest limit of the entire exercise is that satellite verification is a remote-sensing check, not a physical inspection. It answers the question of whether a facility plausibly matches its description. It does not answer questions about title, encumbrance, environmental liability or workforce. It is most powerful when combined with ground-truth visits to a sample of sites, and when the imagery analyst and the financial analyst are in the same room comparing notes.
Typical figures
| Best available optical resolution | 0.30 m panchromatic (Pléiades Neo, WorldView-3) |
| Best available SAR resolution | ~0.25 m range (ICEYE Spotlight mode) |
| Typical revisit for tasked optical | 1 day (WorldView-3 off-nadir); same-day multi-pass possible with SkySat |
| Archive depth (optical) | WorldView from 2009; Landsat public archive from 1972 |
| Minimum detectable vehicle | Heavy goods vehicle reliably resolved at 0.5 m; passenger car distinguishable at 0.3 m |
| Thermal band resolution | Landsat 8/9 TIRS: 100 m; insufficient to isolate a single building in a dense estate |
| SAR coherence change detection | Sentinel-1 at 5–20 m; ICEYE Spotlight at sub-metre; sensitive to surface disturbance between passes |
| Delivery latency (tasked collect to analyst) | Typically 24–72 hours for commercial tasking; archive imagery same day |
| Coverage per tasking pass | WorldView-3: up to 680 km² per pass; SkySat: ~21 km² per collect |
| Jurisdictional resolution caps | India, South Korea, Israel and others impose national restrictions; verify before tasking |
Analytics Satellize can run
| Footprint area vs. declared GFA reconciliation | Rooftop delineation by object-based image analysis; stereo-derived height for volumetric cross-check | Per-site table comparing measured rooftop area and estimated volume against prospectus figures, with imagery excerpts |
| Vehicle presence time-series | Multi-date vehicle detection using template matching or CNN-based object detection on sub-metre optical imagery | Chart of vehicle counts per site per collection date; flagging of sites with anomalously low activity |
| Construction and operational history timeline | Archive change detection across Landsat, Sentinel-2 and commercial imagery; visual chronology of site development | Annotated image timeline per site showing first appearance, expansion events and any periods of apparent inactivity |
| SAR coherence anomaly flag | Interferometric coherence change detection between two SAR passes; coherence loss indicates surface modification | Binary flag with supporting coherence map for each site; narrative explanation of what the change may indicate |
| Thermal activity screening | Landsat TIRS Band 10 thermal anomaly extraction; comparison of site pixel temperature against surrounding background | Pass/flag result per site with caveats on 100 m resolution limits; recommended for large industrial or processing facilities only |
| Regulatory imaging constraint map | Desk review of national shutter-control regimes and operator licence conditions for each target jurisdiction | Jurisdiction-by-jurisdiction table of resolution availability, operator restrictions and recommended sensor alternatives |
| Consolidated site-by-site verification report | Structured synthesis of optical, SAR and thermal evidence against prospectus declarations | PDF report with per-site confidence rating (confirmed / ambiguous / inconsistent), supporting imagery panels and recommended follow-up actions |
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.