InSAR mission configurations
InSAR turns repeat SAR passes into millimetre-scale ground-motion maps, but the measurement lives or dies on orbit-tube control, baseline geometry and coherence preservation. This page covers the mission-design layer, not the radar itself.
What InSAR actually measures, and what it does not
Interferometric SAR does not measure displacement directly. It measures the difference in two-way travel time between two radar passes over the same patch of ground. When that phase difference is unwrapped and projected onto a line-of-sight vector, you recover surface displacement to roughly one-tenth of the radar wavelength, which is 5–6 mm for C-band (5.6 cm wavelength, as on Sentinel-1) and 12–15 mm for L-band (24 cm, as on ALOS-2). X-band systems such as COSMO-SkyMed achieve sub-millimetre precision in ideal conditions because the shorter wavelength amplifies phase sensitivity, but that same sensitivity makes them more vulnerable to atmospheric delay and temporal decorrelation.
The critical word is line-of-sight. InSAR sees the component of motion along the radar look vector, typically 20–45 degrees from vertical. Pure horizontal motion perpendicular to the flight track is nearly invisible. Separating vertical from horizontal requires either ascending and descending pass combinations, or a companion GNSS network. Any mission design that skips this geometry discussion is selling a partial product.
Repeat-pass geometry: the baseline budget
Interferometric coherence requires that the two passes view the ground from nearly the same angle. The perpendicular baseline, the cross-track separation of the two orbital positions projected perpendicular to the look direction, must stay below a critical threshold or phase noise overwhelms the signal. For Sentinel-1 C-band that critical baseline is roughly 5,000 m; for X-band systems it falls to 1,500–2,000 m. In practice, baselines below 200 m are preferred for high-coherence urban monitoring, and below 500 m for agricultural or vegetated terrain.
Achieving this in a single-satellite constellation means flying a tight orbit tube. ESA's Sentinel-1 mission holds its ground track repeat to within ±50 m of the reference orbit, a requirement that drives propulsion sizing and station-keeping fuel budget across the satellite's design life. A sovereign InSAR mission must specify the orbit-tube tolerance before the propulsion subsystem is sized; the two are coupled. Relaxing from ±50 m to ±200 m can save propellant mass, but it introduces baseline scatter that degrades coherence over vegetated or rough terrain.
Temporal baseline: how fast the ground forgets the radar
Coherence decays with time. Vegetation grows, soil moisture changes, wind moves leaves. C-band coherence over dense forest can drop to near-zero within six days; over dry bare rock it persists for months. L-band penetrates canopy and retains coherence over vegetated terrain at 14–46 day intervals, which is why ALOS-2 and the forthcoming NISAR mission use L-band for subsidence and landslide monitoring in tropical regions. X-band decorrelates fastest of all over anything other than built structures or bare soil.
The revisit interval is therefore not just an operational convenience; it is a physical constraint on what terrain types are measurable. A government procuring an InSAR capability for tropical forest monitoring and choosing X-band on cost grounds will find the measurement simply does not exist over the areas of interest. Matching band to application is the first design decision, and it precedes orbit design, not follows it.
Formation flying and tandem configurations
Single-pass interferometry, where two antennas on one pass acquire simultaneously, eliminates temporal decorrelation entirely. The TanDEM-X mission demonstrated this at scale: two X-band satellites flying in a helix formation with cross-track baselines of 150–500 m produced a global 12 m DEM with 2 m height accuracy. The formation required continuous manoeuvring to hold the baseline within specification, consuming propellant at a rate that must be budgeted across the mission life.
For a sovereign programme, a two-satellite tandem formation is a significant step up in complexity and cost. The alternative is a single satellite with a tight repeat orbit and acceptance of temporal decorrelation limits. A pragmatic middle path used by some national programmes is to task a domestic satellite for primary imaging and cross-correlate with Sentinel-1 open data to fill temporal gaps, though this introduces dependency on a third-party archive and is not a sovereign guarantee. The mission architecture document should state explicitly which approach is chosen and what monitoring scenarios each approach cannot support.
Where InSAR configurations fail, and why that matters before contract signature
Atmospheric delay is the dominant error source in most operational InSAR products. Tropospheric water vapour introduces path delays of up to several centimetres, easily masking the millimetre-scale signals of interest. Mitigation requires either external weather model corrections (ERA5 reanalysis is commonly used), GPS-derived zenith delay maps, or temporal averaging across many interferograms. None of these is free: ERA5 corrections add processing complexity and latency; GPS networks require ground infrastructure; temporal averaging requires long time series and cannot detect rapid transient events.
Phase unwrapping fails in areas of steep topography or rapid deformation, where the phase changes by more than half a wavelength between adjacent pixels. C-band at 5.6 cm wavelength means any motion exceeding 2.8 cm in one revisit interval causes wrapping ambiguity that may be unresolvable. The 2015 Nepal earthquake produced surface displacements of several metres; Sentinel-1 interferograms over the rupture zone were largely incoherent in the near-field. InSAR is a precision instrument for slow processes. It is not a rapid-onset disaster monitor in the near-field, and any mission specification that promises otherwise should be questioned.
Specifying an InSAR mission: the decisions that cascade
The specification sequence matters. Band choice sets coherence limits by terrain type. Coherence limits set the maximum useful temporal baseline. Temporal baseline sets the required revisit, which sets the constellation size or the acceptance of gaps. Revisit and orbit altitude together set the orbit-tube tolerance, which sets propulsion mass. Propulsion mass affects launch class and cost. A change to band choice at contract review stage can invalidate the propulsion design entirely.
Ground segment design is equally coupled. InSAR processing pipelines are computationally intensive: a single Sentinel-1 IW-mode scene covers roughly 250 km by 170 km and requires coregistration to sub-pixel accuracy, spectral filtering, phase flattening and unwrapping before any geophysical product emerges. A sovereign programme that owns the satellite but contracts out processing has not achieved data sovereignty in any meaningful sense. The ground segment specification should include the processing chain, the archive policy and the latency requirement for operational products, not just the downlink infrastructure.
Engineering parameters
| Measurement precision (line-of-sight) | 2–15 mm per interferogram depending on band, baseline and atmospheric conditions |
| Orbit-tube tolerance (typical InSAR mission) | ±50–200 m of reference ground track; tighter tolerances require more propellant |
| Critical perpendicular baseline | ~1,500–2,000 m (X-band); ~5,000 m (C-band); scales with wavelength |
| Useful temporal baseline by band | X-band: 1–11 days (built surfaces); C-band: 6–24 days; L-band: up to 46 days over vegetation |
| Minimum revisit for operational subsidence monitoring | 6–12 days preferred; 24 days acceptable for slow (mm/yr) signals with good coherence |
| Atmospheric delay error (uncorrected) | Up to 5–10 cm one-way; dominant error source in most campaigns |
| Tandem formation baseline (TanDEM-X heritage) | 150–500 m cross-track; maintained by continuous manoeuvring over mission life |
| Processing compute class | High: full-scene coregistration and unwrapping typically requires GPU-accelerated infrastructure or cloud burst capacity |
| DEM height accuracy (single-pass, TanDEM-X heritage) | 2 m relative, 4 m absolute at 12 m posting, over low-relief terrain |
One contract, one accountable engineer
Commissioned as one programme, not a stack of contracts: spacecraft, launch, ground segment, mission control, training and handover are priced together. Source-access terms and audit rights are agreed in writing before signature. Review orbit-tube tolerances with our engineers.