Tidal flat inundation frequency and sediment accretion dynamics
Dense Sentinel-1 SAR time-series and Sentinel-2 optical composites reconstruct inundation frequency curves and surface elevation on intertidal flats, revealing sediment accretion or erosion rates relevant to coastal flood defence and blue-carbon accounting.
Sensors
- Sentinel-1 IW GRD (C-band SAR): 20 m ground range resolution, 6-day repeat at mid-latitudes (3-day with both satellites), all-weather day-night acquisition. Backscatter contrast between wet inundated surface and dry exposed sediment is the primary inundation signal. Archive runs from 2014.
- Sentinel-2 MSI: 10 m visible and NIR bands, 20 m SWIR, 5-day revisit with both satellites. Used for optical waterline delineation and NDWI-based inundation mapping on cloud-free acquisitions. Cloud cover is a hard constraint; in persistently overcast coastal climates, usable scenes per month can fall to single figures.
- TanDEM-X CoSSC (X-band interferometric SAR): Global DEM at 12 m posting, vertical accuracy approximately 2 m absolute and better than 1 m relative over flat terrain. Provides a baseline topographic surface against which waterline-derived elevation models are validated or corrected. Not freely available; access requires a DLR science or commercial agreement.
- Landsat 8/9 OLI: 30 m multispectral, 16-day single-satellite revisit. Extends the optical archive back to 1984 via earlier Landsat missions, enabling decade-scale change detection of flat extent and waterline position. Coarser than Sentinel-2 but invaluable for long-baseline trend analysis.
What the waterline method actually measures
An intertidal flat is, by definition, exposed at low water and submerged at high water. If you observe the same flat repeatedly at different tidal stages and record the waterline position each time, you can assign an elevation to every point on the flat: the elevation equals the predicted water level at the moment that point transitions from wet to dry. Repeat this across hundreds of SAR acquisitions and you build a continuous inundation frequency curve. From that curve, a digital elevation model of the flat surface emerges without a single altimeter overflight.
The method was formalised in peer-reviewed literature using ERS and Envisat data and has since been applied at scale with Sentinel-1. Its precision depends almost entirely on the quality of the tidal prediction used to assign water levels to each image. Where a well-validated hydrodynamic tidal model exists, vertical uncertainties of 10 to 20 cm are achievable. Where only a simple harmonic tide model is available, errors can exceed 30 cm, which matters when you are trying to resolve accretion rates of a few millimetres per year.
Where the method holds and where it breaks
The waterline approach has three hard limits worth stating plainly. First, tidal range. Below roughly 0.5 m, the vertical spread of waterline observations is too narrow to resolve meaningful elevation differences; the resulting DEM is effectively flat and uninformative. Many sheltered lagoons and semi-enclosed seas fall into this category. Second, vegetation. Saltmarsh and mangrove canopy obscure the sediment surface from optical sensors and alter SAR backscatter in ways that are not straightforwardly related to inundation state. Vegetated flats require separate treatment; this page addresses bare or sparsely vegetated intertidal sediment.
Third, meteorological water-level anomalies. Storm surges and wind set-up displace the actual water surface from the astronomically predicted tide. If surge events are not filtered or corrected, they introduce systematic elevation biases. In practice, acquisitions taken during known surge events are flagged and excluded, or a barometric correction is applied using reanalysis pressure fields. Even so, residual errors from unmodelled surge remain a recognised source of uncertainty in published studies.
Reading accretion and erosion from multi-year elevation trends
A single waterline-derived DEM is a snapshot. Two DEMs separated by several years become a sediment budget. The difference surface, expressed in centimetres per year, reveals whether the flat is aggrading, stable or eroding. Published studies using Sentinel-1 over European and East Asian tidal systems have detected net accretion signals of 1 to 5 cm per year on accreting flats, and erosion of comparable magnitude on flats subject to sediment starvation or sea-level pressure.
The practical floor on detectable trend depends on DEM uncertainty and the length of the time series. With a 10-year Sentinel-1 archive and careful tidal modelling, trends above roughly 5 mm per year are generally distinguishable from noise. Shorter time series or poorer tidal models push that floor higher. This is not a weakness unique to satellite methods; traditional RTK GPS surveys on mudflats face similar sampling limitations and are far more expensive to repeat systematically.
For blue-carbon accounting, elevation trend data translate into estimates of organic carbon burial rate when combined with field measurements of sediment bulk density and carbon content. The satellite component constrains the spatial pattern of accretion; the carbon conversion requires ground-truth samples. No purely remote-sensing workflow eliminates that field component.
Coastal flood defence: what planners actually need from this data
A tidal flat in front of a sea wall is not passive scenery. It attenuates wave energy, and its elevation relative to mean sea level determines how much attenuation it provides. A flat that has eroded by 20 cm over a decade now allows larger waves to reach the wall toe. Quantifying that change, at spatial resolution fine enough to identify the most vulnerable sections, is exactly what the waterline-derived elevation trend provides.
Coastal engineers typically want the data in two forms: a georeferenced elevation raster for input to wave-propagation models, and a change map showing statistically significant accretion or erosion polygons over a defined baseline period. Both are direct outputs of the waterline workflow. The elevation raster can be differenced against a lidar survey where one exists, giving a cross-validated product with quantified uncertainty bounds, which is what a flood-risk regulator will ask for before the data enters a formal assessment.
Combining SAR and optical for a more complete picture
SAR and optical sensors see the waterline differently. Sentinel-1 C-band backscatter drops sharply when the surface is flooded, giving a binary wet-dry signal that is reliable in most conditions but can be confused by surface roughness from wind on exposed wet mud. Sentinel-2 NDWI is more interpretable spectrally but is blocked by cloud. In practice, the two archives are used together: SAR provides the high-frequency temporal sampling needed to build a dense inundation frequency curve, while optical composites validate the waterline position on clear-sky days and help characterise sediment surface type.
TanDEM-X provides an independent elevation reference. Because it was acquired in a narrow time window (2010 to 2015 for the global DEM), it captures a specific epoch of flat morphology. Comparing a Sentinel-1-derived DEM from a later period against TanDEM-X gives an inter-sensor accretion estimate, though the different radar frequencies (X-band versus C-band) interact differently with wet sediment surfaces, introducing a systematic offset that must be characterised before the comparison is taken at face value.
Satellize processes Sentinel-1 and Sentinel-2 archives over client-defined coastal extents, applying validated tidal models and delivering elevation trend products as GeoTIFF rasters with accompanying uncertainty layers. The workflow draws on the same open-constellation infrastructure used in the Tonga crop-estimation programme, adapted for coastal hydrology.
Archive depth and what a decade of data is worth
Sentinel-1A launched in April 2014. That gives a continuous C-band SAR archive of more than ten years over most coastal regions, long enough to detect statistically meaningful elevation trends on moderately dynamic flats. Landsat extends the optical record back to 1984, though at 30 m resolution and 16-day revisit the waterline method is less precise; Landsat-derived flat DEMs carry vertical uncertainties closer to 30 to 50 cm, useful for detecting gross morphological change but not for millimetre-scale accretion accounting.
The archive depth also matters for inundation frequency curves used in ecological applications. Intertidal habitat classification, particularly distinguishing high flat from low flat zones, requires enough observations across the full tidal range to populate the frequency histogram reliably. Studies suggest a minimum of around 100 cloud-free or SAR acquisitions spread across tidal stages is needed for a stable curve; with Sentinel-1's 6-day repeat, that threshold is typically reached within two to three years over most sites.
Typical figures
| Spatial resolution (SAR waterline DEM) | 20 m native pixel; DEM posting typically 10–20 m after multi-look processing |
| Vertical accuracy (waterline-derived DEM) | 10–20 cm with well-validated tidal model; 30+ cm where only harmonic tide model available |
| Minimum detectable elevation trend | ~5 mm/yr with 10-year Sentinel-1 archive and good tidal correction; higher with shorter series |
| Minimum tidal range for method validity | ~0.5 m; method degrades below this threshold |
| Sentinel-1 revisit (both satellites) | 6 days at mid-latitudes; 3 days where orbital overlap occurs |
| SAR archive depth | Sentinel-1: April 2014 to present; ERS/Envisat usable for earlier epochs at coarser precision |
| Optical archive depth | Sentinel-2: 2015 to present (10 m); Landsat: 1984 to present (30 m) |
| Minimum inundation observations for stable frequency curve | ~100 acquisitions spread across tidal stages; typically 2–3 years of Sentinel-1 |
| TanDEM-X DEM vertical accuracy (flat terrain) | ~2 m absolute; <1 m relative; acquisition epoch 2010–2015 |
| Delivery formats | GeoTIFF elevation and trend rasters, uncertainty layers, GeoPackage change polygons, PDF technical summary |
Analytics Satellize can run
| Waterline-derived intertidal DEM | Tidal-stage assignment of SAR and optical waterline observations; published waterline method (e.g. Mason et al. framework) | GeoTIFF elevation raster with per-pixel uncertainty estimate |
| Inundation frequency curve | Cumulative distribution of wet-dry state observations binned by predicted tidal elevation | Gridded inundation frequency raster (% time inundated) and zonal statistics report |
| Multi-year elevation trend map | Pixel-wise linear regression on DEM time-series; Mann-Kendall significance testing | GeoTIFF trend raster (mm/yr) with significance mask; GeoPackage polygons of statistically significant accretion and erosion zones |
| Sediment budget summary | Volumetric integration of elevation change surface over defined flat extent | Tabular sediment volume change (m³/yr) by sub-unit with confidence intervals; PDF report |
| Blue-carbon accretion proxy layer | Elevation trend combined with published organic carbon burial rate relationships; requires client-supplied bulk density and carbon content field data | Spatially distributed carbon burial potential map; flagged as proxy requiring field validation |
| Flat morphology change detection (long baseline) | Landsat NDWI composites differenced across multi-decade epochs; change vector analysis | Decadal change map at 30 m resolution; area statistics of flat gain and loss |
| Wave-attenuation input layer for coastal engineering | Waterline DEM clipped and formatted to client model grid specification | ASCII grid or NetCDF bathymetric/topographic surface ready for SWAN or XBeach model ingestion |
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.