13.7.3 — Humanitarian Mapping — maturity: live
Damage Assessment Atlases
Systematically mapping structural damage after earthquakes, floods, conflicts or industrial disasters using multi-source satellite imagery fused into georeferenced, version-controlled damage atlases.
When a cyclone, earthquake or flood tears through a city, satellite-derived damage atlases tell responders exactly where to go first — and sovereign nations that own that imagery never have to ask permission.
When a major disaster strikes, governments face an immediate and often fatal information gap: they do not know which neighbourhoods are destroyed, which roads are passable, or where survivors are concentrated. Commercial damage assessments arrive slowly, are licensed to single agencies under restrictive terms, and stop updating the moment the media cycle moves on. A nation that cannot produce its own damage atlas within 24 hours of an event is functionally blind during the window when lives can still be saved.
A sovereign constellation closes that gap. Optical microsatellites at 0.5–1 m resolution, combined with a SAR payload that sees through cloud and smoke, generate pre- and post-event image pairs over any point in the country within hours. On-board change-detection algorithms flag candidate damage pixels before the data even hits the ground station. Sovereign GPU infrastructure then runs building-footprint segmentation and damage-grade classification—aligned to the Copernicus Emergency Management Service GRADING scale—and assembles a continuously updated, spatially versioned atlas.
The operational outcome is a living document, not a static PDF. Civil protection agencies, military engineers and humanitarian clusters all read from the same authoritative source. Each new overpass adds a new version layer, so decision-makers can track whether a partially damaged block has deteriorated or been cleared. Because the data pipeline is nationally operated, the atlas can be declassified, redacted or restricted at any sensitivity level the government chooses—without waiting for a foreign vendor's legal team to approve release.
Frequently asked
What exactly is a damage assessment atlas and how does it differ from a regular satellite map?
A damage assessment atlas is a structured, versioned geospatial product that classifies individual structures or land parcels into damage grades — typically following the COPERNICUS EMS or UN-SPIDER schema (e.g. Destroyed, Heavily Damaged, Moderately Damaged, Negligible/No Damage) — derived by comparing pre- and post-event imagery. Unlike a raw satellite image or a simple basemap, it is an analytical derivative with per-feature attributes, uncertainty estimates, and provenance metadata, formatted for direct ingestion into cluster coordination tools and government damage registries.
Why does it matter whether the nation owns the satellite rather than just buying imagery commercially?
Access continuity is the first issue: commercial providers can deprioritise tasking, apply licensing restrictions, or simply be overwhelmed with competing requests after a major event affecting multiple countries simultaneously. The second issue is latency: a sovereign operator can pre-programme systematic acquisition of its own territory on every pass, so a pre-event baseline is always current. Third, price: emergency commercial tasking can cost ten to twenty times standard rates; a national constellation has zero marginal cost per additional pass over national territory.
Can AI fully automate damage classification, or do human analysts still need to be in the loop?
Current best practice, reflected in UNOSAT's operational workflow, is human-machine teaming. Automated classifiers handle the initial triage at speed and scale — flagging candidate damaged structures across thousands of square kilometres — while trained analysts validate, correct, and add contextual interpretation before the product is released for operational use. The IEEE literature puts state-of-the-art F1 scores at roughly 0.85–0.90 on benchmark datasets, but real-world performance drops in informal urban areas, mixed rubble fields, and where pre-event baselines are low-resolution.
Which orbit is best for rapid damage mapping — LEO, MEO, or GEO?
LEO constellations (400–600 km altitude) are the operational standard, offering the resolution and revisit frequency that damage mapping demands. SAR satellites at LEO achieve sub-metre to 3-metre resolution with constellation revisits under 6 hours. Very high resolution optical at LEO (e.g. Planet SuperDove at 3m, or 30cm-class tasking satellites) complements SAR. GEO is unsuitable for the resolution required, and MEO offers no practical advantage for Earth imaging at the scales relevant to building-level damage assessment.
What is Copernicus EMS and should a developing nation rely on it?
The Copernicus Emergency Management Service, operated by the European Commission with JRC and a network of service providers, provides free rapid-mapping activations on request for any government affected by disaster — producing damage grading polygons, reference maps, and situation reports typically within 24–48 hours of activation. It is a genuinely valuable global public good. However, it is a service funded and controlled by the EU, subject to ESA and EC policy decisions, and cannot guarantee priority access when European disasters compete for the same constellation capacity. For a nation experiencing frequent hazard events, relying solely on Copernicus is a strategic risk; owning complementary national capacity is the prudent hedge.
How many satellites does a nation realistically need for an effective national damage-mapping capability?
A practical national SAR constellation for damage mapping typically starts at four satellites in the same orbital plane to achieve near-daily revisit, scaling to eight or more for sub-12-hour revisit over any national territory. If the nation procures dual-use multispectral microsatellites in a complementary plane, a 6–8 satellite mixed constellation (4 SAR + 4 optical) can deliver operationally useful response products within one to two orbital periods of a disaster event. Nations with very small territory can achieve this with fewer satellites; large continental states need more.
How does a damage atlas feed into the humanitarian response beyond emergency mapping?
Damage atlases drive several downstream processes: insurance and government compensation registers, reconstruction planning (which infrastructure to prioritise), population displacement modelling, and anticipatory action trigger calibration for future events. OCHA's cluster coordination system explicitly calls for standardised damage data to guide logistics routing and shelter allocation. Long-term, multi-event atlases covering the same country also become training datasets for improving national AI classifiers, compounding the value of a sovereign programme over time.
What standards should a national damage atlas product conform to so it is usable by international humanitarian organisations?
The minimum interoperability baseline includes: ISO 19115-1 metadata, OGC WMS/WFS services for live data sharing, geometry in EPSG:4326 (WGS 84), and damage classification using the Copernicus EMS grading schema or the UN-SPIDER recommended practice taxonomy. UNOSAT and OCHA additionally recommend publishing products as GeoPackage or GeoJSON with explicit provenance fields. Nations that conform to these from day one can immediately plug their products into OCHA's Humanitarian Data Exchange (HDX) and the Copernicus Emergency Management ecosystem.
