Disaster damage triage for telecoms network assets
After an earthquake, cyclone or flood, operators need to know which towers are down, which roads are blocked, and which sites can be restored first. SAR coherence change detection and very-high-resolution optical imagery answer those questions before a single field crew leaves the depot.
Sensors
- Sentinel-1 SAR (C-band, ESA): 20 m ground range resolution in Interferometric Wide Swath mode; 6-day repeat at mid-latitudes with both satellites active. Coherence between pre- and post-event passes drops sharply over disturbed structures, flagging collapse or displacement. Cloud-independent, day/night capable. Free and open archive from 2014.
- ICEYE SAR constellation (X-band): Sub-metre spotlight mode (0.25 m range resolution published); tasked revisit can be as short as a few hours depending on orbital geometry. X-band is more sensitive to small structural changes than C-band, but also more affected by vegetation motion, which can produce false coherence loss.
- Capella Space SAR (X-band): Spotlight imagery at 50 cm resolution; the company publishes tasking-to-delivery windows of roughly 12 to 24 hours for priority orders. Like ICEYE, X-band coherence loss is a strong indicator of surface disturbance, though separating antenna tilt from full tower collapse still requires corroborating data.
- Pleiades Neo (optical, Airbus): 30 cm native resolution, four-satellite constellation enabling same-day revisit over a target. Provides direct visual evidence of structural collapse and road blockage when cloud cover permits. Stereo tasking allows a rough 3-D check on whether a tower mast is still vertical.
What coherence loss actually tells you, and what it does not
SAR coherence is a measure of how similar the radar backscatter phase is between two passes over the same ground. When a structure, a tower mast, a equipment shelter, a road surface, remains physically unchanged between acquisitions, coherence is high. When it collapses, shifts or is buried under debris, the phase relationship breaks down and coherence drops toward zero. That signal is detectable at Sentinel-1's 20 m resolution even for a structure that occupies only a few pixels, because the phase change propagates into the surrounding clutter.
The important caveat: coherence loss is a proxy for surface change, not a confirmed damage assessment. Wet vegetation moving in wind after a storm can suppress coherence over an otherwise intact site. A tower that has lost only its antenna head, leaving the mast standing, may show partial coherence loss that looks similar to a tower that has toppled completely. Distinguishing antenna misalignment from total structural loss from orbit is genuinely difficult. X-band SAR (ICEYE, Capella) is more sensitive to small displacements than C-band Sentinel-1, but that sensitivity cuts both ways: it also responds to minor surface changes that have no operational significance. Coherence maps should be read as a prioritisation tool, not a final damage verdict.
Optical change detection: what the camera confirms
Very-high-resolution optical imagery at 30 to 50 cm provides the visual confirmation that SAR coherence cannot. A collapsed tower leaves a recognisable shadow absence, a debris scatter pattern and often a visible gouge in the ground. Road blockage by landslide or flood debris is unambiguous in plan view at this resolution. Pleiades Neo's stereo capability adds a height dimension: a mast that appears intact in nadir view but leans at 15 degrees will show a foreshortened shadow and a measurable height discrepancy between the two stereo views.
The hard constraint is cloud cover. After a tropical cyclone or a monsoon-season earthquake, the affected area may remain under dense cloud for 24 to 72 hours. Optical data then becomes available only after the most time-critical triage window has passed. This is precisely why SAR coherence, unaffected by cloud, is the first layer to process, with optical confirmation following when skies clear. The two methods are complementary rather than redundant.
Activation timelines and the 12-to-48-hour reality
The Copernicus Emergency Management Service (CEMS) can be activated by EU member states, national civil protection authorities and certain international bodies. Once activated, CEMS typically delivers a first grading product within 24 to 48 hours of the triggering event, though actual latency depends on satellite pass scheduling over the affected area and analyst capacity. The products are publicly documented and include delineation maps and grading maps that classify damage by severity. Telecoms operators in affected regions can access these products directly, though the CEMS spatial focus is on population and infrastructure broadly, not specifically on tower assets.
For commercial tasking of Pleiades Neo, ICEYE or Capella, the published order-to-delivery window for priority disaster tasking is typically 12 to 48 hours. The lower end of that range assumes a favourable orbital pass occurs within a few hours of the order. The upper end reflects situations where no suitable pass geometry exists until the following day, or where cloud rejection forces a repeat attempt. Operators planning disaster response should assume 24 hours as a working estimate for first commercial optical delivery, and 6 to 12 hours for SAR if a commercial X-band pass is available.
Building a triage priority list from the data
The practical output of a damage triage workflow is a ranked list of sites requiring physical inspection, ordered by probability and severity of damage. Sites showing high coherence loss in SAR combined with visible structural change in optical imagery go to the top. Sites showing coherence loss without optical confirmation are flagged as probable damage pending field verification. Sites showing no coherence change are provisionally cleared, though this does not rule out equipment-level faults invisible from orbit.
Access-road status is a parallel product. Landslide debris and flood inundation blocking approach roads are detectable in both SAR and optical imagery. A tower that is structurally intact but unreachable for five days has a different restoration priority than a collapsed tower on an open road. Overlaying road-blockage polygons with tower-damage probability scores produces a dispatching matrix that field operations teams can act on directly.
One persistent ambiguity deserves emphasis. An antenna that has rotated on its mount, perhaps 10 to 20 degrees off azimuth by wind loading, will cause significant coverage degradation but may be invisible from orbit. The tower mast remains coherent, the optical image shows no collapse, and the site appears undamaged. Ground truth from a drive-test or a field engineer remains the only way to confirm antenna alignment. Space data narrows the inspection list; it does not eliminate the need for it.
How Satellize structures this for an operator
A telecoms operator holding a pre-event SAR archive over its network footprint is in a materially better position than one starting from scratch after a disaster. Pre-event coherence baselines, established from Sentinel-1's free archive, allow immediate differencing the moment a post-event pass is acquired. Without that baseline, analysts must use generic land-cover coherence expectations, which increases false-positive rates.
Satellize ingests Sentinel-1 coherence pairs, runs change-detection against a client's tower-asset register, and delivers a damage-probability score per site as a GIS layer within hours of a new SAR acquisition becoming available. Commercial SAR tasking through ICEYE or Capella is added on client licence for sites where Sentinel-1 resolution is insufficient to distinguish adjacent structures. The workflow is the same one used in the Tonga crop-estimation programme for rapid change detection over dispersed assets, adapted here for vertical infrastructure rather than agricultural parcels. Operators who want to discuss pre-positioning a baseline archive before the next event can book a scoping call with the analytics team.
Typical figures
| SAR coherence resolution (Sentinel-1 IW mode) | 20 m ground range; coherence computed over multi-look windows, typically 40–80 m effective |
| SAR coherence resolution (ICEYE / Capella spotlight) | 0.25–1 m range resolution; coherence products typically resampled to 1–3 m |
| Optical resolution (Pleiades Neo) | 30 cm panchromatic; 1.2 m multispectral (4 bands) |
| Sentinel-1 revisit (two-satellite) | 6 days at mid-latitudes; up to 12 days at equatorial latitudes depending on swath overlap |
| Commercial SAR tasking latency | 6–24 hours for first post-event acquisition; 12–48 hours to delivered product |
| CEMS activation-to-first-product latency | 24–48 hours (published CEMS service level for rapid mapping) |
| Minimum detectable structural change (SAR coherence) | Surface displacement of a few centimetres sufficient to suppress C-band coherence; X-band detects sub-centimetre displacement but with higher false-positive rate from vegetation |
| Sentinel-1 archive depth | From April 2014 (Sentinel-1A launch); free and open via Copernicus Data Space |
| Delivery formats | GeoTIFF coherence difference rasters, vector damage-probability polygons, CSV site-score tables, optional WMS feed |
| Cloud cover constraint | SAR: unaffected. Optical: total obstruction under dense cloud; tropical events may delay optical acquisition by 24–72 hours |
Analytics Satellize can run
| Tower-site damage probability score | SAR coherence differencing (pre/post event Sentinel-1 or commercial X-band) intersected with asset register | GIS point layer with per-site probability score (0–1) and confidence flag, delivered within hours of SAR acquisition |
| Structural collapse confirmation map | Optical change detection using normalised difference of pre/post VHR imagery at 30–50 cm; shadow geometry analysis for mast verticality | Polygon layer of confirmed collapse footprints with photographic evidence chips, PDF report |
| Access-road blockage assessment | SAR backscatter change and optical debris-detection over road network buffer zones | Line-segment layer classifying each road segment as clear, probable blockage or confirmed blockage; integrated with tower-priority ranking |
| Inspection dispatch priority matrix | Weighted overlay of damage probability, access-road status and estimated restoration complexity | Ranked CSV and map of sites ordered for field crew dispatch, updated as new imagery is acquired |
| Pre-event coherence baseline archive | Time-series Sentinel-1 coherence stack over network footprint, establishing seasonal norms per site class | Per-site baseline coherence statistics stored for immediate post-event differencing; delivered as a standing data product ahead of any disaster |
| Post-event change bulletin | Automated SAR coherence anomaly detection triggered on new Sentinel-1 acquisition over a predefined area of interest | Email or API alert with anomaly map within 3 hours of ESA data release; suitable for network operations centre integration |
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.