Every serious land-use decision — flood modelling, road alignment, military route planning, dam safety — rests on an accurate digital elevation model (DEM). The problem is that commercially available global DEMs such as SRTM or Copernicus GLO-30 carry horizontal errors of 5–30 m and vertical errors that compound badly in steep or forested terrain. A nation relying on third-party data is, in effect, planning critical infrastructure on someone else's measurement, collected at someone else's schedule, with someone else's quality-control standards.
A sovereign constellation built around repeat-pass InSAR (interferometric synthetic aperture radar) closes this gap systematically. Two passes of an X-band or L-band SAR separated by a controlled baseline produce coherent phase differences that, after processing, yield digital surface models with vertical accuracy better than 0.5 m RMS over open terrain and 1–2 m under forest canopy. Fusing those radar-derived models with stereo optical imagery from the same constellation sharpens feature edges — road cuts, river banks, coastal cliffs — where radar phase decorrelates. The result is a living national DEM that can be updated quarterly or after any significant geomorphic event.
The operational payoff is immediate and cross-sectoral. Defence engineers get accurate cross-country trafficability maps without importing foreign-controlled data. Hydrologists can run credible 100-year flood simulations. Mining and energy regulators hold a reference surface against which every future survey can be checked for subsidence or stockpile change. Owning the sensor means owning the update cadence: after an earthquake, landslide or volcanic eruption the constellation is retasked within hours, not weeks.
Frequently asked
What is the practical difference between a DSM, a DEM, and a DTM — and which does a satellite produce?
A Digital Surface Model (DSM) captures the top of everything — buildings, trees, infrastructure. A Digital Terrain Model (DTM) is the bare-earth surface after those features are removed. A Digital Elevation Model (DEM) is often used loosely for either. Satellites — whether optical stereo or SAR — natively produce DSMs; converting to a DTM requires post-processing to filter out non-ground returns, typically using algorithms such as progressive TIN densification or, best of all, airborne LiDAR reference data.
Why would a nation bother running its own DEM programme when commercial providers like Planet or ICEYE will sell data?
Three reasons: data sovereignty, continuity, and cost at scale. A government that purchases DEMs from a commercial vendor has no guarantee of access during a conflict, trade dispute, or vendor insolvency. It also pays per-kilometre fees indefinitely rather than amortising a one-off constellation investment. For a country of 500,000 km², a sovereign six-satellite SAR constellation typically breaks even within seven to ten years against equivalent commercial subscriptions — and delivers data classified at whatever sensitivity level the government requires.
How accurate can a nanosatellite or microsatellite SAR constellation realistically be?
Modern 100–150 kg SAR microsatellites operating in stripmap mode at X-band routinely achieve 1–3 m spatial resolution and, with careful InSAR processing and GCP support, ±0.5 m relative vertical accuracy — sufficient for flood modelling, infrastructure monitoring, and land-use planning. Centimetre-level accuracy requires bistatic configurations or differential InSAR with very high coherence, which demands either two closely-flying spacecraft or extremely stable atmospheric conditions.
Can optical stereo satellites replace radar for terrain modelling?
Optical stereo (as used by Planet's SkySat or legacy SPOT) is cost-effective in cloud-free regions and can reach 0.5 m GSD, yielding DEMs accurate to 1–2 m vertically. However, it fails entirely under persistent cloud cover — which affects 60–80% of tropical landmasses — and cannot penetrate vegetation canopy. SAR is therefore the backbone technology for a comprehensive national DEM programme, with optical stereo as a complementary layer for urban and arid areas.
What ground infrastructure does a nation need to operate a terrain-mapping satellite programme?
At minimum: one or two ground receive stations (ideally at high latitude to maximise contact time with LEO satellites), a processing cluster capable of SAR focusing and DEM generation, a geodetic ground control network with centimetre-level GNSS receivers, and a national data archive conforming to ISO 19115 metadata standards. Nations without existing geodetic networks should treat that ground investment as a prerequisite — a satellite that delivers 0.5 m accuracy over a country with only 30 m GCPs cannot realise its full potential.
How does terrain data underpin disaster-risk management?
Flood inundation models, landslide susceptibility maps, tsunami run-up projections, and earthquake fault-rupture assessments all begin with a DEM. FEMA in the United States and the Copernicus Emergency Management Service in Europe have both demonstrated that upgrading from 30 m to 1 m DEMs reduces false-positive evacuation zones by 30–50%, saving both lives and economic disruption. A sovereign DEM capability means a government can run these models itself, in real time, without waiting for a commercial data licence to clear.
What ITU coordination is required before launching a SAR terrain-mapping constellation?
A nation must file frequency coordination with the ITU Radiocommunication Bureau under ITU-R RS.577-8, covering the active sensor bands used (typically X-band at 9.3–9.9 GHz or C-band at 5.25–5.57 GHz). It must also register the orbital slots through its national telecommunications authority and observe coordination timelines — which can run to 24–36 months for contested bands. Early filing is strongly advised; ITU registration confers legal priority that protects a sovereign operator against later commercial entrants in the same frequency band.
How often does a national DEM need to be refreshed?
It depends on land-use dynamics. Urban and coastal zones change fast enough to need annual or sub-annual updates; stable highland terrain may need only a 5-year refresh cycle. WMO and FAO guidance on land-monitoring recommends a baseline refresh no longer than every 3 years for agricultural and hydrological applications. A constellation with 24–48 h revisit can be tasked to re-survey priority areas continuously, making a sovereign programme inherently more responsive than any static commercial dataset purchased on a one-off licence.