Protected area boundary encroachment and buffer-zone clearing detection
Encroachment into protected areas often advances in narrow, incremental strips designed to avoid detection. Spatial analysis combining change-detection outputs with WDPA boundary geometries can expose the pattern, but cloud cover and resolution limits set real constraints on alert latency.
Sensors
- Sentinel-2 MSI: 10 m resolution in visible and near-infrared bands, 5-day revisit at the equator with both satellites. Free and open archive from 2015. Cloud cover in humid tropics routinely reduces usable acquisitions to fewer than one per month at any given site.
- Planet SuperDove: 3 m resolution, daily revisit globally across the SuperDove constellation. Eight spectral bands including red-edge. Provides the temporal density needed to catch incremental buffer-zone clearing between Sentinel-2 acquisitions, though cloud penetration remains an optical limitation.
- Planet SkySat: 50 cm resolution, taskable on demand. Used for confirmation and legal-evidence documentation once a candidate encroachment event is flagged by coarser sensors. Revisit is task-dependent rather than systematic.
- Maxar WorldView Legion: 30 cm native resolution, multiple daily passes over selected latitudes. Provides the spatial detail required to resolve narrow buffer-zone strips of 20 to 50 m width and to produce imagery meeting court-admissible chain-of-custody standards.
Why incremental encroachment is designed to defeat coarse sensors
Encroachment into protected areas rarely arrives as a single large clearing. The more common pattern is a sequence of small cuts, each individually below the detection threshold of moderate-resolution sensors, advancing over months or years toward and eventually across a boundary. A 10 m pixel from Sentinel-2 can resolve a clearing of roughly 0.1 hectares in favourable conditions, but narrow strips cleared along a buffer-zone edge, perhaps 15 to 30 m wide, sit at the ambiguous margin of that sensor's capability. The geometry matters: a strip cleared parallel to a boundary may span only one or two pixels in the across-boundary direction, making spectral change hard to distinguish from shadow variation or phenological shift.
This is precisely why boundary-proximity analysis requires a layered sensor approach rather than reliance on any single constellation. The detection architecture typically runs Sentinel-2 or SuperDove as the systematic monitoring layer, then triggers commercial tasking of SkySat or WorldView Legion when a candidate alert falls within a defined proximity threshold of a WDPA-registered boundary. The threshold is a policy choice, not a technical one: 500 m is common for buffer-zone monitoring, though some national park regulations define narrower legally significant zones.
Intersecting change detection with boundary geometry
Change detection on optical imagery produces candidate polygons. The analytical step that converts those polygons into boundary-encroachment alerts is a spatial intersection with the World Database on Protected Areas geometry, published by UNEP-WCMC and IUCN. WDPA boundaries are not always precise: some are mapped at scales coarser than 1:50,000, which introduces positional uncertainty that can exceed the width of a narrow buffer strip. Any honest encroachment-detection system must propagate that boundary uncertainty into the alert, flagging events as confirmed, probable, or requiring field verification depending on how far inside or outside the boundary the detected change falls relative to the known positional error of the WDPA polygon.
Cumulative encroachment statistics, tracking the total area cleared within successive distance bands from a boundary over a defined period, are more legally and managerially useful than individual event alerts. They reveal whether a site is under sustained pressure and whether the rate is accelerating. A site losing 2 hectares per quarter at its boundary is a different management problem from one that lost 8 hectares in a single event. Both require documentation, but the response and the evidentiary requirements differ.
Cloud cover is not a minor inconvenience
In the Congo Basin, the Amazon, and across much of insular Southeast Asia, cloud-free optical acquisitions over any given 1 km square can be separated by three to six weeks during the wet season. An encroachment event that begins and reaches legally significant scale within that window may be invisible to optical sensors until the clearing is already well established. This is not a solvable problem with faster revisit alone: SuperDove passes daily, but cloud is cloud.
Synthetic aperture radar from Sentinel-1 penetrates cloud and provides 10 m resolution backscatter imagery with a 6 to 12-day repeat. SAR change detection is less spectrally informative than optical, and distinguishing new clearing from other land-surface changes requires careful calibration, but it provides the only reliable cloud-independent signal for tropical sites. A practical monitoring architecture uses SAR to maintain continuity during optical blackouts, then reconciles the two data streams when cloud clears. Alert latency for tropical protected areas should be quoted honestly as days in dry season and potentially weeks in wet season, not as a single figure.
What legal-evidence documentation actually requires
Imagery used in legal proceedings, whether for national prosecution, REDD+ compliance, or international environmental law, must meet standards that go beyond what a monitoring alert requires. The key requirements are: spatial resolution sufficient to identify the feature in question without ambiguity (generally accepted as sub-1 m for individual structures, sub-5 m for clearing boundaries); a documented and unbroken chain of custody from sensor to courtroom; metadata establishing the acquisition time, sensor calibration state, and geometric correction applied; and, in many jurisdictions, certification by the imagery provider or an accredited analyst.
Planet SkySat at 50 cm and WorldView Legion at 30 cm both meet the resolution threshold. Chain-of-custody requirements mean that imagery must be ordered and archived under a formal licence with documented provenance, not pulled from a cached tile server. The choice of image product for evidentiary use is therefore a legal procurement decision as much as a technical one. Analysts preparing documentation packages typically deliver georeferenced GeoTIFF files with full metadata, annotated PDF reports showing the boundary intersection geometry, and a change-timeline figure showing the sequence of acquisitions that captured the encroachment progression.
What the numbers can and cannot prove
Satellite-derived encroachment statistics are powerful precisely because they are spatially explicit and time-stamped. A cumulative area figure showing 47 hectares cleared within 200 m of a park boundary over 18 months, supported by a dated image sequence, is a concrete finding. What the imagery cannot prove unaided is intent, ownership, or the identity of the party responsible. Those questions require ground investigation and legal process. Satellite analysis establishes the what, where, and when; it does not replace the who.
Satellize runs boundary-proximity analytics on open Sentinel and commercial Planet imagery for clients requiring systematic WDPA-referenced monitoring, drawing on the same operational approach used in its Tonga crop-estimation programme to combine open-constellation coverage with commercial tasking for high-resolution confirmation. The honest ceiling of this approach is that no optical system resolves encroachment events narrower than its ground sampling distance, and no system eliminates cloud. Those limits should be stated in any monitoring contract.
Typical figures
| Spatial resolution (monitoring layer) | 3 m (Planet SuperDove), 10 m (Sentinel-2 MSI) |
| Spatial resolution (confirmation and legal evidence) | 30 cm (WorldView Legion), 50 cm (Planet SkySat) |
| Revisit cadence (optical) | Daily (SuperDove constellation); 5 days at equator (Sentinel-2 twin satellites) |
| Cloud-independent revisit (SAR) | 6 to 12 days (Sentinel-1, 10 m IW mode) |
| Minimum detectable clearing (Sentinel-2) | Approximately 0.1 ha under favourable conditions; narrow strips below 20 m width are ambiguous |
| Alert latency (dry season, optical) | 1 to 5 days from clearing event to candidate alert |
| Alert latency (wet season, tropical) | Potentially 2 to 6 weeks depending on cloud persistence; SAR maintains continuity |
| Boundary reference dataset | WDPA (UNEP-WCMC / IUCN), updated monthly; positional accuracy varies by site |
| Archive depth | Sentinel-2 from 2015; Landsat from 1972 (30 m); Planet commercial archive from approximately 2016 |
| Deliverable formats | GeoTIFF, GeoJSON, Shapefile, PDF annotated report, API alert feed |
Analytics Satellize can run
| Boundary-proximity change alert | Spectral change detection (NDVI delta or normalised burn ratio) intersected with buffered WDPA polygon geometry | Automated alert feed with candidate polygon, distance to boundary, acquisition date, and confidence class |
| Cumulative encroachment statistics | Time-series aggregation of confirmed change polygons within successive distance bands (100 m, 200 m, 500 m) from boundary | Quarterly GIS layer and tabular report showing area cleared by distance band and cumulative rate of change |
| Legal-evidence documentation package | High-resolution commercial tasking (SkySat or WorldView Legion) with full metadata, geometric correction record, and annotated boundary-intersection figure | Georeferenced GeoTIFF archive, PDF annotated report, chain-of-custody metadata sheet |
| Cloud-gap-filled encroachment timeline | Fusion of Sentinel-1 SAR backscatter change with optical change detections to maintain temporal continuity during cloud-obscured periods | Monthly GIS layer with cloud-gap periods flagged and SAR-derived change polygons distinguished from optical-derived polygons |
| Encroachment rate and trajectory model | Linear and non-linear regression on cumulative cleared-area time series to estimate rate acceleration and projected boundary breach date | Site-level risk report with trend chart and projected encroachment trajectory under current rate assumptions |
| WDPA boundary uncertainty assessment | Comparison of WDPA polygon positional metadata against higher-resolution cadastral or GPS-surveyed boundary where available; uncertainty buffer applied to alert classification | Boundary confidence layer appended to alert outputs, classifying alerts as confirmed, probable, or requiring field verification |
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.