Illegal dumping costs municipalities hundreds of millions annually in clean-up, enforcement and public-health remediation, yet most councils still rely on citizen reports and infrequent field inspections to find new sites. By the time a tip-off arrives, a fly-tip can have grown from a mattress to a multi-tonne hazardous pile that requires specialist contractors, legal proceedings and months of remediation. The problem is structurally invisible to ground-level administration: too much land, too few inspectors, too little time.
A constellation of sub-metre optical microsatellites revisiting the same urban fringe, road corridors and industrial periphery every 24–48 hours supplies the persistent watch that ground patrols cannot. Change-detection models trained on confirmed dump sites flag new spectral anomalies — bare disturbed soil, reflective plastic, colour breaks inconsistent with the surrounding land cover — and filter out benign changes such as construction spoil with valid permits. A probability score and geo-stamped imagery tile are pushed to enforcement teams within hours of overpass, enabling same-day site visits while evidence is fresh and perpetrators are more likely to be traceable.
Sovereign operation matters here because the data is inherently forensic. Imagery of waste sites, combined with vehicle-track analysis or RF geolocation of mobile phones present at the time, can become evidence in criminal prosecutions. A national or city authority that routes this intelligence through a commercial third-party cloud has no guaranteed chain of custody, no control over data retention policy and no certainty that access will not be suspended during a contract dispute or geopolitical sanction event. Owning the pipeline from pixel to prosecution is the only way to make the evidence stick.
Frequently asked
What resolution of imagery do you actually need to detect a fly-tip versus clean land?
Peer-reviewed studies and operational programmes consistently point to 0.5 m ground sample distance (GSD) or better as the threshold where automated classifiers achieve usable accuracy (>85%). At 1 m GSD, smaller dumps under roughly 20 m² become unreliable to detect. At 3 m GSD — the Sentinel-2 shortwave infrared bands — you can confirm large, established sites but will miss the fresh incidents you most want to catch early. A sovereign constellation targeting this application should specify at minimum 0.5 m panchromatic / 2 m multispectral as the payload requirement.
Can synthetic aperture radar (SAR) replace optical imagery so clouds are not a problem?
SAR is a valuable complement, not a replacement. It sees through cloud and operates at night, and C-band or X-band backscatter can detect the volumetric roughness change when a dump appears. However, SAR cannot discriminate waste material type (hazardous versus inert), and its false-positive rates for small sites are higher than sub-metre optical in clear conditions. Best-practice architecture combines both modalities: SAR triggers an alert within the cloud-obscured window, and the next clear optical pass confirms and classifies the site.
How does a satellite system actually alert enforcement officers — what does the operational workflow look like?
Imagery is downlinked to a national ground station, ingested into a change-detection pipeline (typically a convolutional neural network comparing yesterday's and today's orthorectified tiles), and candidate detections above a confidence threshold are pushed as GeoJSON alerts to a municipal GIS dashboard or mobile app within two to four hours of acquisition. A duty officer reviews the flagged polygon, cross-checks permit databases, and — if confirmed — dispatches a field team or CCTV-equipped drone. The satellite provides persistent watchfulness; humans close the enforcement loop.
Why should a government own this capability rather than subscribe to Planet, BlackSky, or a similar service?
Commercial services provide excellent imagery, but the data pipeline, the detection algorithm, the alert thresholds, and the retention policy are all controlled by a foreign private entity. A government that owns the constellation sets its own revisit priority over its own territory, retains raw imagery for legal evidentiary purposes under its own data-sovereignty law, and is immune to licence-fee increases or export-control changes. For a municipal function that generates enforcement actions with legal consequences, that control matters.
How many satellites does a viable sovereign constellation require, and what is the approximate cost?
A constellation of 12 microsatellites (each roughly 50–150 kg, 0.5 m optical payload) delivers 4–6 hour revisit over a medium-sized nation, sufficient for next-morning alerting. Scaling to 30 satellites achieves near-real-time 90-minute revisit. Build, launch, and five-year operations costs for a 12-satellite national system are currently in the $180–350 million range depending on technology heritage and launch vehicle choice — substantial, but comparable to what a large city spends on ground-based enforcement in a decade.
Can AI change-detection really keep pace with the volume of imagery a constellation generates?
Yes, and this is now the standard approach. Modern pipelines process thousands of km² per hour on cloud GPU infrastructure, flagging only changed pixels for human review. The bottleneck is not compute but labelled training data and quality-assurance of detections. Nations building sovereign capability should budget for a dedicated geospatial AI team that maintains and retrains classifiers as waste typologies and urban land cover evolve — not just the satellite hardware.
What international or national legal frameworks govern the use of satellite imagery for enforcement?
At the international level, the UN Principles on Remote Sensing (UNGA Resolution 41/65) establish that states may collect imagery over their own territory without restriction. Domestically, admissibility of satellite evidence in enforcement proceedings is governed by national evidence law, which varies widely. The EU's INSPIRE Directive (2007/2/EC) sets metadata and interoperability obligations for spatial data used in public-authority decisions within member states. Nations should audit their evidence legislation before deploying an enforcement-linked detection system.
What happens to detected sites in areas where land tenure is informal or disputed — does satellite evidence create legal risk for residents?
This is a genuine governance risk. In informal settlements, residents may dump on adjacent land out of necessity due to absent municipal collection services; automating enforcement without understanding the socioeconomic context can criminalise poverty rather than organised illegal dumping. Best practice — recommended by the World Bank's urban resilience programme — is to filter detections against informal-settlement boundaries and route those alerts to a social-services triage rather than direct enforcement, while using the data to make the case for expanded waste-collection budgets.