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.

What matters

Damage atlases produced from foreign commercial imagery are subject to export-control holds and vendor downlink-priority decisions that can delay delivery by 48–72 hours at exactly the wrong moment.
The Copernicus EMS Rapid Mapping service, the gold standard for foreign-supported response, typically activates in 3–12 hours but covers only areas formally requested by EU member or participating states—non-members get no guarantee.
Building-level damage grading (HAZUS or EMS-98 scale) requires sub-metre optical or coherent SAR; a nanosatellite constellation cannot achieve this—minimum 80 kg microsat class with a 0.5 m optical or X-band SAR payload is required.
A nationally versioned atlas doubles as a legal instrument: insurers, reconstruction authorities and international donors all require a sovereign-endorsed damage record to release funds.

Quick facts

Structures assessed in Turkey–Syria earthquake atlas (Feb 2023): ~180,000 buildings (2023) — Copernicus EMS Activation EMSR648 – Turkey Earthquake
Accuracy of AI-assisted building damage classification vs. field survey (F1 score): 0.87 (2022) — IEEE Transactions on Geoscience and Remote Sensing – Automated Damage Assessment

Sovereignty score: 9/10 — A nation that cannot independently certify its own damage atlas cedes control of the post-disaster narrative, the reconstruction budget and the legal record of destruction to foreign vendors and multilateral bodies.

Commercial imagery providers—Maxar, Airbus, Planet—apply tiered downlink-priority schemes; a country in conflict or with strained diplomatic relations may find its tasking requests deprioritised or withheld entirely during the critical 0–72-hour response window.
Reconstruction financing from the World Bank, IMF and bilateral donors legally requires a nationally endorsed geospatial damage record; dependence on a foreign-produced atlas creates a bottleneck that delays fund disbursement and hands audit authority to a non-sovereign entity.
Damage atlases produced during conflict or civil unrest contain operationally sensitive information about infrastructure status, force disposition and population movement; routing that data through a foreign cloud pipeline creates an unacceptable intelligence exposure.
Insurance and liability claims after industrial or nuclear disasters require a chain-of-custody-clean, court-admissible damage record; a sovereign-operated, cryptographically signed atlas version history satisfies that evidentiary standard in a way a third-party commercial product does not.

Reference architecture

Payload: Primary: 0.5 m panchromatic / 2 m multispectral push-broom optical imager, 12 km swath; secondary: X-band SAR, 3 m stripmap / 1 m spotlight, 30 km swath — both payloads operated in tandem for cloud-penetrating all-weather coverage
Bus class: ESPA-class microsat, 120–180 kg wet mass, 600 W EOL power budget; dual-payload accommodation requires a standard ESPA or ESPA-Grande adapter — nanosats are insufficient for the aperture and power required at sub-metre resolution
Orbit: Sun-synchronous LEO at 500–520 km, 10:30 AM local-time descending node for consistent solar illumination; 8-satellite walker constellation achieving sub-6-hour revisit over any national territory; SAR satellites offset by 45 minutes in LTAN for interleaved day/night coverage
Ground segment: 2 primary X-band / S-band ground stations (capital region + geographically diverse backup site); autonomous emergency tasking uplink triggered by national disaster-declaration API; SatNOGS UHF/VHF network as TT&C contingency for safe-mode recovery
Data pipeline: On-board L0 radiometric correction and lossy-compressed downlink → ground L1 orthorectification against national 1 m DEM → automated co-registration of pre- and post-event image pairs → GPU-accelerated building footprint segmentation (YOLO-v8 or SegFormer) + damage-grade classifier (EMS-98 / HAZUS scale) → PostGIS version-controlled damage atlas updated per overpass → cryptographic hash anchored to national blockchain ledger for legal provenance
End-user delivery: OGC WMS/WFS atlas served from sovereign cloud to civil protection agency GIS consoles, military engineering units and humanitarian cluster coordinators; per-building damage grade downloadable as GeoPackage or GeoJSON; push alerts via national emergency API when damage-pixel count crosses predefined thresholds; classified network feed for defence and intelligence consumers on a separate VLAN
Time to launch: First 2-satellite demonstrator (1 optical + 1 SAR) operational within 28 months of contract award; full 8-satellite constellation with sub-6-hour revisit within 48 months
Caveats: Sub-metre optical payloads and X-band SAR electronics are subject to ITAR/EAR; procure optical sensors from Airbus Defence & Space (France), SSTL (UK) or ISRO commercial partners, and SAR from ICEYE (Finland) or SEOSAT-derived Indian primes to avoid US export-control dependency; on-board AI inference chips (GPU/NPU) should be sourced from non-ITAR supply chains (e.g., Unibap, Nordic Semiconductor) where possible.

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.

Glossary

SAR: Synthetic Aperture Radar — an active microwave sensor that images the Earth's surface regardless of cloud cover or darkness, making it the primary sensor for rapid post-disaster damage mapping.
Change detection: The computational process of comparing two satellite images of the same area taken at different times to identify pixels or objects that have changed — the foundational technique behind automated damage classification.
Damage grade: A standardised classification label assigned to each mapped building or parcel, typically ranging from 'Destroyed' through 'Heavily/Moderately Damaged' to 'No Damage', following schemas defined by Copernicus EMS or UN-SPIDER.
Copernicus EMS: The European Union's Copernicus Emergency Management Service — a free, on-request satellite mapping service for governments responding to disasters, operated under the broader EU Copernicus Earth observation programme.
UNOSAT: The United Nations Satellite Centre (part of UNITAR) — the UN's primary satellite analysis unit, producing damage atlases, population displacement maps, and geospatial products for humanitarian response globally.
Pre-event baseline: A cloud-free, high-resolution satellite image of an area acquired before a disaster, used as the reference against which post-event imagery is compared to detect and classify damage.
Revisit time: The interval between successive satellite passes over the same ground location; shorter revisit times are critical for rapidly evolving disaster situations where the damage picture changes hour by hour.
HDX: Humanitarian Data Exchange — OCHA's open platform for sharing humanitarian datasets, including satellite-derived damage atlases, accessible at data.humdata.org.
F1 score: A statistical metric combining precision and recall (values 0–1) used to evaluate how accurately an AI classifier detects and labels damaged structures relative to a ground-truth reference dataset.
Cluster system: The OCHA-coordinated humanitarian coordination architecture that organises international response into thematic clusters (Shelter, Logistics, Health, etc.), each of which depends on geospatial damage data to plan and prioritise operations.

References

IEEE TGRS: Deep Learning-Based Building Damage Detection from Satellite Imagery After the 2020 Beirut Explosion — This peer-reviewed study applies convolutional neural network architectures to Maxar and Sentinel imagery of the 2020 Beirut port explosion, achieving F1 scores of 0.87 on building-level damage classification and benchmarking performance against field survey data collected by Lebanese civil authorities.
World Bank: Disaster Risk Finance and Satellite Data — Building National Capacity for Rapid Damage Assessment — The World Bank's Disaster Risk Financing & Insurance Programme documents how satellite-derived damage atlases are being integrated into parametric insurance triggers and sovereign catastrophe bonds for small island developing states and lower-middle-income countries, reducing the information gap that delays post-disaster funding disbursement.
OCHA: Humanitarian Data Exchange – Satellite-Derived Damage Datasets — OCHA's HDX platform hosts damage atlas datasets from Copernicus EMS, UNOSAT, and partner national agencies for over 200 disaster events. The platform's metadata schema and licensing framework set the interoperability baseline for humanitarian satellite data sharing globally.
Sendai Framework for Disaster Risk Reduction 2015–2030: Monitoring Report — The UNDRR monitoring framework for the Sendai targets explicitly requires national governments to report disaster losses at the asset and infrastructure level — a requirement that can realistically only be met at scale through satellite-derived damage assessment programmes rather than ground survey alone.
ESA Earth Observation for Humanitarian Response – Mission Requirements Document — ESA's operational requirements for humanitarian Earth observation missions establish minimum ground resolution (better than 3m for building-level damage), revisit time (under 24 hours for rapid mapping), and geolocation accuracy (sub-5m CE90) as the baseline specification for any constellation intended to support damage atlas production.

Related applications

13.7.1 — OpenStreetMap Field Updates (Humanitarian Mapping)
13.7.5 — Anticipatory Action Triggers (Humanitarian Mapping)
13.1.1 — Tele-Radiology Networks (Telemedicine)
13.1.2 — Mobile Health Camp Operations (Telemedicine)
13.1.3 — Emergency Telemedicine Consultation (Telemedicine)
13.1.4 — Specialty Consultation Networks (Telemedicine)
13.1.5 — Hospital-to-Hospital Tele-ICU (Telemedicine)
13.2.1 — Vector-Borne Disease Risk Mapping (Disease Intelligence)