Sensors
- Maxar WorldView Legion: Sub-30 cm panchromatic resolution; up to 15 revisits per day over a target at mid-latitudes. Resolves individual fence posts, vehicle ruts and roof condition on structures as small as 3 m across.
- Airbus Pléiades Neo: 30 cm native resolution, stereo and tri-stereo collection modes. Stereo pairs yield a digital surface model accurate to roughly 50 cm vertically, useful for confirming whether a watchtower has been built to specified height.
- Planet SkySat: 50 cm resolution, up to 12 tasked collects per day over a site. Lower per-image cost than WorldView Legion makes it practical for frequent revisit across large protected areas where daily coverage is operationally useful.
- Sentinel-1 SAR (C-band, ESA): 10 m resolution in Interferometric Wide Swath mode, 6-day repeat at the equator (3-day with both satellites). Penetrates cloud and smoke. Detects surface roughness changes from new track construction or vegetation clearance even when optical collection is blocked for weeks.
What a satellite can actually see on the ground
A ranger post is a physical object. It has a roof with a detectable spectral signature, a cleared footprint, vehicle access tracks and, in most designs, a water tank or flagpole. At 30 cm resolution, Pléiades Neo and WorldView Legion resolve all of these. A corrugated iron roof is spectrally distinct from surrounding soil and vegetation in the near-infrared band; a water tank casts a shadow of predictable geometry. This is not interpretation. It is measurement.
The practical detection floor for a discrete structure is roughly 3 to 5 metres in plan dimension, provided contrast with the background is sufficient. A small watchtower in open savanna clears this threshold easily. One embedded in dense canopy does not, and that honest limit matters: VHR optical imagery cannot confirm a post inside closed forest without a canopy-penetrating sensor. Sentinel-1 SAR can detect the cleared ground around a post even when the structure itself is obscured, but at 10 m resolution it will not resolve the structure independently.
Unauthorised tracks are the earliest readable signal
Poaching networks do not appear suddenly. They build access. A new vehicle track in a protected area is typically 3 to 6 metres wide and compresses or removes vegetation in a linear pattern that is highly anomalous against undisturbed ground. At 50 cm resolution, a single SkySat collect over a previously imaged baseline will flag this within days of the track's creation. Sentinel-1 backscatter change detection catches the same signal through cloud, though it cannot distinguish a poaching track from a legitimate management track without contextual cross-referencing.
Archive depth is operationally significant here. Planet's SkySat archive and Maxar's archive both extend back several years over many protected areas. A funder auditing a reserve can request a time series showing when a track appeared, how it has been used over subsequent months, and whether it connects to a known boundary crossing point. That is evidence, not inference.
Infrastructure audit: what deterioration looks like from orbit
Conservation funders increasingly require independent verification that capital expenditure on patrol infrastructure has produced the physical assets claimed. Satellite imagery provides this without requiring an auditor to travel to a remote reserve. A ranger post funded in year one should appear in the imagery of year one. Its roof should still be present in year three. If it is not, or if the structure has visibly degraded, the imagery record captures that.
Roof condition is the most reliable proxy for building integrity at VHR resolution. A collapsed or heavily rusted roof changes spectral reflectance measurably in the shortwave infrared. Vegetation encroachment onto a cleared compound is visible as a gradual reduction in bare-soil area over successive collects. Neither indicator is infallible, but together they constitute a credible audit signal that is far harder to fabricate than a field report.
Fence-line monitoring is more demanding. A standard game fence is roughly 2 metres tall but only centimetres wide. At 30 cm resolution, a fence is detectable as a linear feature only when it contrasts with background, typically on bare soil or short grass. Gaps in a fence line, which matter enormously for wildlife containment, are detectable at that resolution if they exceed about 5 metres. Smaller gaps are below the practical detection threshold and require ground verification.
SAR as the cloud-independent check
Tropical protected areas can be cloud-covered for months at a stretch. The Congo Basin, parts of Southeast Asia and the Amazon wet season routinely defeat optical tasking for six to ten weeks. Sentinel-1's C-band radar is indifferent to cloud. It does not image in the same sense that optical sensors do, but it detects surface roughness and dielectric change. A newly bulldozed track produces a coherence loss in Sentinel-1 interferometric pairs that is detectable even when no optical sensor has seen the ground for weeks.
The practical workflow pairs Sentinel-1 change alerts with optical tasking. When SAR flags an anomaly, a VHR optical collect is ordered to characterise it. This avoids the cost of blanket daily optical coverage while ensuring that cloud does not create a monitoring blind spot. The latency between a SAR anomaly and an optical confirmation collect is typically 24 to 72 hours under normal tasking schedules, which is operationally acceptable for infrastructure audit though not for real-time anti-poaching response.
Turning imagery into an audit deliverable
Raw imagery is not an audit. The analytical steps between a satellite collect and a funder-ready report include orthorectification, baseline comparison, change classification and confidence scoring. Each step introduces uncertainty that should be declared. A change-detection algorithm trained on savanna will underperform in dense woodland; a model calibrated on East African infrastructure may not transfer directly to Southeast Asian forest without retraining.
Satellize structures this as a periodic audit layer: a GIS-delivered polygon set marking each infrastructure element, its observed condition class and a confidence rating, compared against the previous audit period and against the project's own infrastructure inventory. The Kingdom of Tonga crop-estimation programme demonstrated that this kind of structured, inventory-referenced analysis is more useful to a government client than a raw change map, and the principle transfers directly to conservation funders who need to sign off on grant compliance.
The output is not a substitute for ranger presence or community engagement. It is an independent data layer that makes it harder to misreport, and easier to catch deterioration before it becomes irreversible.
Typical figures
| Best optical spatial resolution | Sub-30 cm (Maxar WorldView Legion panchromatic) |
| Typical VHR optical revisit | 1 to 15 times per day over a target (constellation and latitude dependent) |
| SAR spatial resolution | 10 m (Sentinel-1 IW mode); 3 m (Sentinel-1 StripMap) |
| SAR revisit | 3 to 6 days (Sentinel-1A and 1B combined, equator) |
| Minimum detectable structure | ~3 to 5 m plan dimension in open terrain at 30 cm optical resolution |
| Minimum detectable fence gap | ~5 m at 30 cm resolution; smaller gaps require ground verification |
| Archive depth | 2016 to present for Sentinel-1; several years for Planet SkySat and Maxar over many sites |
| Cloud penetration | Full (Sentinel-1 SAR); none (optical sensors) |
| Delivery formats | GeoTIFF, GeoPackage, Shapefile, COG; report PDF for audit deliverables |
| Tasking latency | Optical confirmation collect typically 24 to 72 hours after SAR anomaly flag |
Analytics Satellize can run
| Infrastructure inventory baseline | Object-based image analysis (OBIA) on VHR optical imagery to delineate and classify structures, tracks and fence lines | GIS polygon layer with structure type, dimensions and spectral condition class |
| Periodic condition audit | Multi-date spectral comparison and change classification against baseline inventory | Audit report PDF and updated GIS layer with condition-change flags and confidence ratings |
| Unauthorised track detection alert | Sentinel-1 coherence change detection for new linear disturbance, confirmed by VHR optical tasking | Alert GeoJSON with track geometry, first-detection date and optical confirmation image chip |
| Construction progress verification | Time-series VHR optical collect sequence compared against project infrastructure schedule | Progress report with dated imagery evidence per infrastructure element |
| Vegetation encroachment index | NDVI time series over compound footprints to measure cleared-area reduction as proxy for post abandonment | Per-post encroachment trend chart and GIS polygon update |
| Stereo-derived watchtower height verification | Pléiades Neo stereo pair processed to digital surface model; height extracted at structure centroid | Height measurement table with uncertainty estimate, cross-referenced to design specification |
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.