Rock uplift and fluvial incision rate comparison from satellite geodesy
In active orogens, tectonic uplift and fluvial incision compete at rates of fractions of a millimetre to several millimetres per year. Sentinel-1 InSAR, continuous GNSS, and multi-epoch DEMs from SRTM and TanDEM-X can quantify both sides of that balance, with important caveats about atmospheric noise and erosional ambiguity.
Sensors
- Sentinel-1 (C-band SAR, ESA): Provides line-of-sight (LOS) surface velocity fields at 5 x 20 m resolution in Interferometric Wide Swath mode, 6-day repeat at mid-latitudes with both satellites operational. Multi-year stacking yields vertical velocity precision of roughly 1-2 mm/yr over stable ground, degrading significantly in high-relief terrain due to layover, shadow and atmospheric path-length variability.
- TanDEM-X (X-band bistatic SAR, DLR): Global DEM at 0.4 arcsec (~12 m) posting, with absolute vertical accuracy better than 10 m and relative accuracy better than 2 m over most terrain. Acquired primarily 2010-2015, providing a second-epoch elevation surface against which SRTM (2000) differences can detect gorge incision, though signal must exceed combined DEM noise floors.
- SRTM (C-band radar, NASA/NGA, 2000): Single-epoch February 2000 global DEM at 1 arcsec (~30 m) posting. Serves as the baseline epoch for multi-decadal DEM differencing. Absolute vertical accuracy is approximately 16 m at 90th percentile globally, though relative accuracy in stable terrain is considerably better. Voids exist in steep shadowed terrain, complicating gorge-bottom measurements.
- IGS and regional CORS GNSS networks: Continuous GNSS stations from the International GNSS Service and national networks (e.g. UNAVCO, EUREF, regional geodetic arrays) provide three-component velocity vectors with vertical precision of 0.5-1 mm/yr after multi-year averaging. Station spacing in active orogens is typically 20-100 km, insufficient to resolve short-wavelength deformation without interpolation.
- Sentinel-1 ascending and descending pair combination: Combining ascending and descending LOS velocities allows partial decomposition into vertical and east-west components, though north-south sensitivity remains poor for C-band near-polar orbits. GNSS tie-points are essential to resolve the remaining ambiguity and to correct for long-wavelength orbital and atmospheric ramps.
What the balance between uplift and incision actually tells you
In an orogenic belt at steady state, the rate at which tectonic forces push rock upward should roughly equal the rate at which rivers and hillslopes strip material away. When those rates diverge, the landscape is either building relief or losing it. That divergence carries direct implications for seismic hazard: a range that is outpacing erosion is accumulating elastic strain energy; one being rapidly dissected may be exposing deeper, more seismogenic crust.
Quantifying the two sides of this balance independently, using different sensors and different timescales, is the point of this workflow. GNSS and InSAR measure present-day surface velocity, integrating over years to decades. DEM differencing between SRTM (2000) and TanDEM-X (circa 2012) captures net elevation change over roughly a decade, blending incision, landsliding, and deposition. Neither method alone is sufficient. Together, with honest uncertainty accounting, they constrain landscape evolution in ways that field campaigns alone cannot match spatially.
Building the vertical velocity field from InSAR and GNSS
Sentinel-1 delivers LOS velocities, not vertical velocities. A single viewing geometry conflates vertical motion with horizontal motion projected onto the radar look direction (typically 30-46 degrees from vertical in IW mode). Separating them requires at minimum two geometries. Ascending and descending track combination resolves vertical and east-west components; north-south motion remains largely invisible to C-band InSAR and must be supplied by GNSS.
The standard approach is to use continuous GNSS velocities from IGS or regional networks as anchor points, correcting InSAR for long-wavelength biases introduced by orbital errors and tropospheric delay, then interpolating the GNSS velocity field to fill the north-south gap. The result is a spatially dense vertical velocity map, but with important caveats. In high-relief terrain, Sentinel-1 C-band coherence degrades rapidly in vegetated or snow-covered slopes. Atmospheric delay in deep valleys can mimic several millimetres of apparent uplift or subsidence per year. Time-series methods such as SBAS or persistent scatterer analysis reduce but do not eliminate this noise.
Realistic vertical velocity precision over orogenic terrain, after multi-year stacking and atmospheric correction, is probably 1-3 mm/yr at the pixel level, improving to sub-millimetre when averaged over stable reference areas. Uplift rates in active orogens such as the Southern Alps of New Zealand, the Himalayas, or the European Alps range from roughly 1 to more than 5 mm/yr, so the signal is detectable but not always unambiguous.
What a floating DEM difference gives away about river incision
Subtracting the SRTM 2000 surface from the TanDEM-X circa-2012 surface should, in principle, reveal where elevation has changed. In gorge systems, a deepening channel floor would appear as a negative anomaly. In practice, the measurement is confounded by several factors that cannot be fully separated.
First, both DEMs represent the first-return radar surface, not the bare ground. Vegetation change between epochs introduces spurious elevation differences. Second, SRTM voids and layover artefacts cluster precisely in the steep gorge walls where incision signal is expected. Third, the combined vertical noise of the two DEMs, even after co-registration, is typically 2-5 m in rugged terrain. Over a 12-year epoch, detecting incision rates below roughly 0.2-0.4 m/yr is statistically marginal at best. Rapid incision in tectonically active gorges, such as those documented in the Himalayan syntaxes where rates can exceed 10 mm/yr, is detectable. Slow background incision in less active settings is not.
Deposition complicates the picture further. Landslide-dammed reaches can show apparent aggradation while the river is actually incising upstream. Mass-wasting events between epochs add or remove material independently of fluvial processes. Any honest interpretation of a DEM-difference map in orogenic terrain must acknowledge these ambiguities explicitly.
Atmospheric artefacts: the principal adversary in high-relief InSAR
Tropospheric water vapour delays radar signals by an amount that correlates with topography. In mountainous terrain, this produces a stratified atmospheric signal that can reach 5-10 cm in a single interferogram and, when not fully corrected, introduces apparent velocity biases of several mm/yr in stacked products. The correlation between atmospheric delay and elevation is the central problem: it mimics exactly the tectonic signal one is trying to measure.
Correction strategies include using ERA5 reanalysis weather fields (from ECMWF), GACOS (Generic Atmospheric Correction Online Service) tropospheric delay maps, or empirical phase-elevation relationships estimated from interferogram residuals. None is perfect. ERA5 spatial resolution (~31 km) undersamples valley-to-ridge moisture gradients. GACOS improves on this but still leaves correlated residuals in complex terrain. Ionospheric delay is less severe at C-band than at L-band but is non-negligible at high latitudes. The practical consequence is that InSAR-derived uplift rates in high-relief orogens carry systematic uncertainties that are difficult to quantify precisely and that GNSS tie-points only partially resolve.
Combining the datasets: what the comparison actually constrains
When GNSS vertical velocities, InSAR-derived velocity fields, and DEM-difference incision estimates are placed alongside each other, the comparison constrains several geophysically meaningful quantities. Where uplift exceeds incision, the range is building relief. Where they are approximately equal, the system may be near steady state. Where incision rates are faster than present-day uplift, the range may be responding to a past pulse of faster rock uplift, or the current geodetic signal may be incomplete.
Seismic hazard context emerges from this comparison indirectly. Rapid rock uplift sustained over geological timescales implies sustained strain accumulation on the faults driving that uplift. DEM-based incision rates, when combined with cosmogenic nuclide exposure ages from river terraces (a field method beyond satellite data alone), can extend the record back tens of thousands of years. The satellite component provides the spatial coverage and temporal consistency that field sampling cannot.
Satellize can deliver this combined analysis as a georeferenced velocity and incision-rate stack, with uncertainty layers, for any orogenic study area with adequate Sentinel-1 archive depth and GNSS coverage. The methodology is the same class of open-source geodetic processing used in published academic programmes; what Satellize adds is systematic quality control, atmospheric correction, and integration into a client's existing GIS or hazard-assessment workflow.
Honest limits and what they mean for interpretation
No satellite dataset resolves incision rates below roughly 0.5 mm/yr from DEM differencing alone over a single decade. InSAR vertical velocities in vegetated, high-relief terrain carry systematic uncertainties that can approach the magnitude of the tectonic signal in slowly deforming settings. GNSS networks in many active orogens are sparse, leaving large interpolation gaps in the velocity field.
These are not reasons to avoid the analysis. They are reasons to be explicit about what the data can and cannot say. A well-constructed uncertainty budget, communicated clearly to a government geohazard agency or infrastructure planner, is more useful than a confident-looking map that conceals its assumptions. The value of the satellite geodesy approach is spatial coverage and archival consistency; the value of honest interpretation is that the client does not build policy on artefacts.
Typical figures
| InSAR spatial resolution (Sentinel-1 IW) | 5 x 20 m (range x azimuth); typically multilooked to 40-100 m for velocity products |
| Sentinel-1 revisit (both satellites) | 6 days at mid-latitudes; 12 days with single satellite |
| InSAR vertical velocity precision (stable terrain, multi-year stack) | ~1-2 mm/yr; degrades to 2-5 mm/yr in high-relief vegetated terrain |
| TanDEM-X DEM relative vertical accuracy | <2 m (90th percentile) over moderate terrain; worse in steep shadowed gorges |
| SRTM DEM absolute vertical accuracy | ~16 m at 90th percentile globally; relative accuracy considerably better over short baselines |
| DEM-differencing minimum detectable incision rate | ~0.2-0.5 m/yr over a 12-year SRTM-to-TanDEM-X epoch in rugged terrain |
| GNSS vertical velocity precision (continuous, multi-year) | 0.5-1 mm/yr after multi-year averaging at well-sited IGS stations |
| Sentinel-1 archive depth | From 2014 (Sentinel-1A launch); usable interferometric archive from late 2014 onwards |
| SRTM baseline epoch | February 2000 (11-day Shuttle mission) |
| Deliverable formats | GeoTIFF velocity and uncertainty rasters, shapefiles, CSV velocity tables, PDF technical report |
Analytics Satellize can run
| InSAR LOS velocity map | SBAS or persistent scatterer time-series analysis of Sentinel-1 SLC stack, with GACOS or ERA5 atmospheric correction | GeoTIFF LOS velocity raster with coherence and uncertainty layers |
| Decomposed vertical velocity field | Ascending/descending LOS combination constrained by IGS GNSS tie-points; north-south component from GNSS interpolation | Georeferenced vertical and east-west velocity grids with propagated uncertainty |
| DEM-difference incision map | Co-registered SRTM-to-TanDEM-X differencing with outlier masking, void filling and stable-terrain normalisation | GeoTIFF elevation-change raster with per-pixel noise estimate; gorge incision summary statistics by drainage segment |
| Uplift-incision balance assessment | Pixel-wise comparison of InSAR/GNSS vertical velocity against DEM-derived incision rate, with uncertainty propagation | Classified GIS layer (uplift-dominated, near-balance, incision-dominated zones) plus PDF interpretive report |
| Atmospheric correction quality report | Comparison of ERA5 and GACOS corrections against phase-elevation empirical fits; residual variance analysis | Per-interferogram correction quality metrics and flagged high-uncertainty zones |
| GNSS velocity field integration | Download and processing of IGS and regional CORS station time-series; trend estimation with seasonal signal removal | CSV and shapefile of station vertical and horizontal velocities with formal uncertainties |
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.