REIT portfolio site verification by satellite
Independent satellite verification of physical asset condition and operational activity gives REIT analysts a data source that does not pass through asset managers. Car-park counts, loading-dock traffic and rooftop equipment condition are observable proxies for tenant health.
Sensors
- Planet SkySat: 0.5 m native resolution (resampled to 0.5 m pan, 1 m multispectral); on-demand tasking with revisit possible multiple times per day over a given site; well-suited to vehicle counts and loading-dock occupancy reads at moderate cost.
- Maxar WorldView-3: 0.31 m panchromatic, 1.24 m multispectral, 3.7 m SWIR; the SWIR bands can distinguish rooftop membrane materials and detect thermal anomalies indicative of HVAC operation or roof-deck moisture ingress. Agile tasking; published collect capacity exceeds 680,000 km² per day across the constellation.
- Airbus Pléiades Neo: 0.3 m panchromatic, 0.75 m multispectral across four satellites; stereo and tri-stereo modes allow 3-D rooftop reconstruction to assess equipment presence and structural change. Revisit of any point on Earth within 24 hours is achievable under favourable geometry.
- BlackSky Global: 0.5 m resolution; the constellation is designed for high-frequency revisit of specific coordinates, making it practical for time-series sampling of the same site across a quarter. Latency from collect to delivery can be under 90 minutes, which matters when an event-driven verification is needed.
What a car park actually tells an analyst
Occupied car spaces are a weak but real signal of tenant activity. Weak, because a full car park does not distinguish between a healthy tenant and a closing-down sale. Real, because a persistently empty car park at a site reported as fully let is a contradiction that demands explanation.
The method is straightforward. At 0.3 to 0.5 m resolution, individual vehicles are resolvable as distinct objects. Automated detection using convolutional neural networks trained on labelled VHR imagery can count vehicles across a car park in seconds, and the published literature (including work using Planet and DigitalGlobe imagery) demonstrates detection precision above 90 percent in well-lit, unobstructed conditions. The honest limit: covered multi-storey car parks are opaque to optical sensors. Shadows cast by adjacent buildings can obscure surface-level spaces in early-morning or late-afternoon collects. A single image is a snapshot at one moment; it cannot distinguish a quiet Tuesday from a structural decline in footfall.
Rooftop equipment as a proxy for operational status
A building that is genuinely operational has active HVAC plant. Rooftop condensing units, cooling towers and exhaust stacks are visible at sub-metre resolution, and their presence, condition and apparent operational state carry information. A unit that has been physically removed, or that shows significant rust and panel damage visible from above, is a signal worth noting in a due-diligence file.
WorldView-3's SWIR bands add a layer of discrimination. Water-saturated roofing membrane has a distinct spectral signature in the 1.2 to 2.4 micron range, which can indicate deferred maintenance or active leakage. This does not replace a physical survey, and it should not be presented as one. What it does is prioritise which assets in a 200-property portfolio are worth sending a surveyor to first.
Pléiades Neo stereo pairs allow a 3-D point cloud of rooftop features to be derived, making it possible to detect whether a rooftop plant room has been added, enlarged or stripped out between two collection dates. For sale-leaseback transactions or lease-renewal negotiations, that kind of change record has direct financial relevance.
Loading docks and logistics yards: the industrial REIT case
For industrial and logistics REITs, loading-dock activity is arguably the most direct observable proxy for tenant throughput. At 0.3 to 0.5 m resolution, docked HGVs are clearly distinguishable from empty bays. Yard truck movements, trailer staging patterns and the presence or absence of palletised goods in open yard areas all contribute to an activity picture.
The sampling challenge is real. A single collect on a Wednesday morning may catch a site at peak despatch. The same site on a bank-holiday Monday would look abandoned. A credible activity assessment for an industrial asset requires at minimum four to six collects spread across different days of the week and times of day over a quarter. That is achievable with BlackSky or SkySat tasking, though the cost scales with collect frequency. The alternative, using archive imagery where it exists, is cheaper but offers less control over collect timing.
Designing a sampling strategy that is honest about its limits
Satellite verification is a sampling exercise, not a monitoring system. Cloud cover is the most obvious constraint: optical sensors cannot see through it, and persistent cloud over a northern European logistics park in November can make a planned quarterly verification programme difficult to execute on schedule. SAR sensors such as Sentinel-1 can detect large vehicles and structural change through cloud, but at 5 to 20 m resolution they cannot resolve individual cars or loading-bay occupancy with the confidence that VHR optical imagery provides.
A practical programme for a portfolio of, say, 50 assets might prioritise the 15 sites where covenant quality is weakest or lease events are nearest, task those at higher frequency, and use archive imagery for the remainder as a baseline. Anomalies flagged by the satellite review then trigger targeted on-demand collects. This tiered approach keeps cost proportionate to risk. It also means the output should be presented to investment committees as a probabilistic indicator rather than a definitive audit. A site showing low activity across three collects warrants further investigation; it does not, by itself, prove a tenant is in distress.
Satellize structures verification programmes along these lines for clients who need independent physical intelligence on asset portfolios, drawing on commercial tasking licences and the same change-detection methodology applied in its Tonga crop-estimation work.
Archive depth and the retrospective audit
One underused capability is retrospective analysis. Planet's archive extends back to 2016 for most of the Earth's land surface at 3 to 5 m resolution (PlanetScope), with SkySat tasked collects available from 2014 for selected sites. Maxar's archive, accessible through Maxar ARD and similar products, holds imagery from WorldView-1 and WorldView-2 dating to 2007 and 2009 respectively.
For a REIT conducting pre-acquisition due diligence, this means it is possible to examine how a target asset's car park filled and emptied across multiple years, whether loading-dock activity changed after a reported tenant change, and whether any significant rooftop or yard modifications occurred that were not reflected in the vendor's documentation. The archive does not provide daily coverage for most sites, and collect density varies considerably by geography and historical tasking demand. But for major commercial centres, the historical record is often richer than buyers realise.
Typical figures
| Best available spatial resolution (optical) | 0.3 m panchromatic (Pléiades Neo, WorldView-3) |
| Multispectral resolution | 0.75 m (Pléiades Neo); 1.24 m (WorldView-3); 1 m (SkySat) |
| SWIR bands | WorldView-3: 8 SWIR bands, 3.7 m resolution, 1.195–2.365 µm; useful for roofing membrane and moisture detection |
| On-demand tasking revisit | Sub-24-hour feasible for Pléiades Neo and BlackSky; SkySat multiple collects per day over target sites |
| Delivery latency (on-demand) | BlackSky: under 90 minutes collect-to-delivery published; SkySat and Pléiades Neo: typically same-day to next-day |
| Minimum resolvable target | Individual passenger vehicles at 0.3–0.5 m; HGVs reliably at 0.5 m and coarser |
| Cloud limitation | Optical sensors fully blocked by cloud; SAR (Sentinel-1, 5–20 m) provides cloud-independent fallback at lower resolution |
| Archive depth | PlanetScope from 2016; SkySat from 2014 (selected sites); Maxar WorldView from 2007–2009 |
| Recommended sampling cadence | 4–6 collects per quarter per priority asset, spread across weekdays and times of day |
| Delivery formats | GeoTIFF orthoimage, vector count layers (GeoJSON/Shapefile), PDF site report, change-detection raster |
Analytics Satellize can run
| Vehicle occupancy count per car park | Object detection using convolutional neural network applied to VHR panchromatic imagery; bounding-box detection of individual vehicles | Per-site vehicle count with timestamp, occupancy rate as percentage of marked spaces, GeoJSON point layer |
| Loading-dock activity index | Supervised classification of dock-bay states (occupied, empty, partially obscured) from VHR imagery; manual QA on ambiguous bays | Dock occupancy table per collect date, time-series chart across quarter, flagged anomalies report |
| Rooftop equipment presence and condition assessment | Visual interpretation and change detection on stereo-derived 3-D surface model; SWIR band analysis for membrane moisture anomalies where WorldView-3 is tasked | Per-asset rooftop equipment inventory, condition flag (no change / modified / degraded), PDF site summary |
| Site activity change detection (quarterly comparison) | Multi-date image differencing and object-count comparison across collect series; statistical significance threshold applied to filter noise | Change-detection GeoTIFF, ranked list of assets showing statistically significant activity decline, investment committee briefing slide |
| Pre-acquisition retrospective activity profile | Archive imagery retrieval and vehicle-count time series across available historical collects; gap-filling noted where archive density is low | Multi-year activity chart per asset, annotated with known lease events, PDF due-diligence annex |
| Portfolio priority triage | Scoring model combining occupancy signal, dock activity and rooftop condition flags; weighted by lease expiry proximity and covenant grade supplied by client | Portfolio heatmap ranked by verification priority, recommended tasking schedule for next quarter |
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.