Urban encroachment monitoring on legally protected archaeological zones
Informal construction within gazetted heritage buffer zones can be tracked quantitatively using time-series change detection in medium- and high-resolution optical imagery. This page covers classification methods, revisit requirements, documented cases in Egypt and Mesopotamia, and how satellite evidence feeds UNESCO and ICOMOS reporting obligations.
Sensors
- Sentinel-2 MSI: 10 m resolution in visible and near-infrared bands, 5-day revisit at mid-latitudes with both satellites. Sufficient to detect building footprints larger than roughly 100 m² and to separate bare-earth construction pads from vegetation using NDVI and NDBI indices. Free archive from 2015.
- Planet Dove: 3–4 m resolution, daily revisit over most land areas. Captures construction events between Sentinel-2 passes and resolves individual structures down to approximately 20–30 m². The high cadence is the main advantage; radiometric consistency across the constellation has improved but remains a calibration consideration for long time series.
- WorldView-2 / WorldView-3: 0.31–0.46 m panchromatic, 1.24–1.85 m multispectral. Provides building-footprint outlines precise enough for legal documentation and court-ready imagery. Tasked on demand; revisit is 1–4.5 days depending on off-nadir tolerance. Cost per km² makes systematic daily coverage impractical over large zones.
- Pléiades Neo: 0.3 m panchromatic, 1.2 m multispectral, 8-band option including red-edge and deep blue. Stereo acquisition in a single pass enables rudimentary volume estimation of new structures. Constellation of two satellites gives a revisit of roughly 1 day at off-nadir angles.
What a construction pad looks like from 500 km up
New informal construction follows a recognisable spectral sequence. Vegetation or agricultural soil is cleared first, exposing bare mineral earth with high reflectance in the red and shortwave-infrared bands and a sharply negative NDVI. Within weeks, concrete-block walls appear, raising the Normalised Difference Built-up Index (NDBI) calculated from SWIR and NIR bands. A finished roof, often corrugated metal or concrete render, produces a spectrally distinct signature that persists across seasons, unlike ploughed soil or harvested crop residue, which revert.
This sequence is detectable in Sentinel-2 at 10 m resolution provided the new structure exceeds roughly one pixel in plan area, which equates to structures larger than about 100 m². Planet Dove at 3–4 m resolves smaller outbuildings and boundary walls. The key classification challenge in heritage buffer zones is separating construction from agricultural change: a newly ploughed field and a cleared building pad can look identical in a single image. Multi-date compositing over 30–90 days resolves the ambiguity, because ploughed soil develops crop cover while a building pad does not.
Revisit frequency determines whether you catch the act or the aftermath
Informal construction in peri-urban areas around heritage sites can progress from cleared ground to a roofed structure in under four weeks. Egypt's Supreme Council of Antiquities has documented cases in the Nile Delta where brick structures were erected over recorded archaeological deposits within a single agricultural off-season. A 5-day Sentinel-2 revisit is borderline for catching the transition before roofing, particularly when cloud cover consumes one or two passes. Planet's daily revisit closes that gap materially.
The practical workflow is a two-tier alert system. Sentinel-2 provides the baseline change-detection layer across the full buffer zone at low cost. Any pixel cluster exceeding a threshold change in NDBI or bare-soil index triggers a tasking request for Planet or Pléiades Neo to confirm and measure the footprint precisely. This avoids the cost of daily commercial tasking over entire buffer zones, which can run to hundreds of square kilometres at sites such as Memphis–Saqqara or the Mesopotamian city mounds of southern Iraq.
Cloud cover is a genuine constraint in the Nile Delta during winter and in the Tigris–Euphrates lowlands during spring. Optical-only pipelines will have blind periods of 2–3 weeks. SAR coherence change detection (covered on a sibling page) can fill those gaps, though it introduces its own interpretation complexity and is outside the scope of this page.
Egypt's Nile Delta and Mesopotamian sites: what the record shows
The Nile Delta is one of the most heavily documented cases of satellite-tracked heritage loss. Studies published in peer-reviewed remote sensing journals using Landsat and later Sentinel-2 time series have mapped the progressive conversion of archaeological sites to brick-kiln operations and residential construction since at least the 1970s. The Delta's low topographic relief means tells are often the only elevated ground in a given area, making them attractive for building. Multitemporal Landsat analysis has shown that dozens of registered sites lost measurable surface area to construction between 1984 and 2010, with rates accelerating after 2000.
In Mesopotamia, the situation documented by researchers using CORONA declassified imagery and later commercial high-resolution satellites shows a different pattern: large unregistered or poorly buffered sites surrounded by expanding agricultural towns. The Iraqi site of Nippur and the broader Dhi Qar governorate sites have been subject to encroachment monitoring by academic teams using WorldView imagery. Buffer zones in Iraq's national heritage register frequently do not correspond to the actual extent of subsurface deposits, which means construction legally outside the buffer can still destroy unexcavated material. Satellite data can quantify this mismatch by overlaying building footprints against geophysical survey extents.
Turning pixel counts into legal evidence
A change-detection map is not, by itself, a legal document. For satellite data to feed into enforcement or UNESCO Periodic Reporting, the analysis must meet several standards. Geometric accuracy must be sufficient to locate a structure within the cadastral or buffer-zone boundary: sub-5 m absolute accuracy is generally required, which Sentinel-2 orthorectified products and Planet's Scene and Ortho Tile products meet in flat terrain. In areas of relief, terrain-corrected products using a local DEM are necessary to avoid positional errors that could misplace a structure inside or outside a boundary.
UNESCO's Periodic Reporting cycle, which runs on a six-year schedule for World Heritage properties, now explicitly encourages state parties to include remote-sensing evidence in their conservation reports. ICOMOS monitoring guidelines similarly reference satellite-derived change metrics as acceptable supporting evidence. The deliverable format matters: GIS layers with attribute tables recording structure footprint area, date of first detection, and distance to the nearest registered boundary polygon are more useful to heritage authorities than raster images alone. Annotated before-and-after image pairs at print resolution are required for committee presentations.
One honest caveat: satellite imagery cannot determine whether a structure is occupied, whether construction has a permit, or whether it post-dates the gazettement of the buffer zone. Ground verification remains essential before enforcement action. The satellite record narrows the search area and establishes the timeline; it does not replace the site visit.
Building a monitoring programme that authorities will actually use
The failure mode for satellite heritage monitoring is not technical; it is institutional. Analyses produced by external researchers or consultants frequently do not reach the national authority with the power to act, or arrive too late in a bureaucratic cycle to trigger a response. An effective programme requires three things: a defined alert threshold (for example, any new built-up pixel cluster exceeding 50 m² within 500 m of a registered boundary), a named recipient within the heritage authority, and a reporting cadence aligned with that authority's inspection schedule.
Satellize structures analytics engagements around exactly this kind of operational integration, matching the output format and delivery frequency to the client's workflow rather than to what is technically convenient to produce. The Tonga crop-estimation programme demonstrated that regular, decision-ready outputs outperform occasional comprehensive reports for agencies with limited analytical capacity. The principle transfers directly to heritage monitoring: a monthly encroachment summary with flagged sites is more actionable than an annual atlas.
Archive depth is an underused asset. Sentinel-2 provides consistent global coverage from 2015, and Landsat extends the usable archive to 1984 at 30 m resolution. For sites where encroachment is disputed, the historical record can establish when construction began relative to the buffer zone's legal gazettement date, which is directly relevant to liability and potential remediation orders.
Typical figures
| Spatial resolution (change detection) | 10 m (Sentinel-2), 3–4 m (Planet Dove), 0.3–0.5 m (Pléiades Neo / WorldView) |
| Minimum detectable new footprint | ~100 m² at 10 m resolution; ~20–30 m² at 3–4 m resolution; ~5 m² at sub-metre resolution |
| Revisit frequency | 5 days (Sentinel-2, dual satellite); daily (Planet Dove); 1–4.5 days (WorldView, off-nadir); ~1 day (Pléiades Neo) |
| Spectral bands used | Visible (RGB), NIR, SWIR for NDVI, NDBI, bare-soil indices; red-edge for vegetation discrimination |
| Geometric accuracy (flat terrain) | <5 m absolute (Sentinel-2 L2A orthorectified); <3 m (Planet Ortho Tile); <1 m (Pléiades Neo) |
| Archive depth | 2015–present (Sentinel-2); 1984–present (Landsat 30 m); 2009–present (WorldView commercial) |
| Cloud-cover constraint | Optical methods fail under persistent cloud; 2–3 week blind periods possible in Delta winter or Mesopotamian spring |
| Typical analysis latency | 24–72 hours from satellite pass to change-detection alert, depending on pipeline automation |
| Delivery formats | GeoTIFF change rasters, GeoPackage / Shapefile footprint polygons, PDF annotated image pairs, CSV attribute tables |
| Coverage per analysis run | Full buffer zone per pass; typical heritage buffer zones range from 1 km² to several hundred km² |
Analytics Satellize can run
| Monthly encroachment alert report | Bi-temporal NDBI and bare-soil index differencing on Sentinel-2 10 m composites; threshold-based change flagging | PDF report with flagged sites, footprint areas, and distance-to-boundary measurements; GIS polygon layer |
| New building footprint extraction | Object-based image analysis (OBIA) on Planet or Pléiades Neo imagery; morphological filtering to separate structures from agricultural features | GeoPackage of building footprints with date of first detection, area (m²), and centroid coordinates |
| Time-series construction timeline | Landsat and Sentinel-2 archive stack analysis; per-pixel change-date estimation using monotonic trend detection | Raster layer of first-change date per pixel; CSV table of cumulative built-up area by year for each buffer zone |
| Buffer-zone compliance map | Spatial overlay of extracted building footprints against registered buffer-zone polygons; inside/outside classification | GIS layer with compliance status attribute; summary table of encroachment area by legal zone category |
| Triggered high-resolution tasking confirmation | Sentinel-2 or Planet change flag triggers commercial tasking order for WorldView or Pléiades Neo; sub-metre imagery reviewed for legal documentation | Orthorectified sub-metre image pair (before/after), annotated at print resolution for heritage authority or UNESCO submission |
| UNESCO Periodic Reporting data package | Aggregation of annual encroachment metrics, change maps, and footprint statistics into the format specified by UNESCO's World Heritage reporting guidelines | Structured data annex with maps, tables, and metadata ready for insertion into the state party's Periodic Report |
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.