CoastObs maps changes of the land-water boundary over time based on satellite imagery

Why is it important?

Coastal and especially estuarine environments are characterized by high sediment dynamics. The distribution of sediments over an area will change depending on discrete events (high river run-off, spring-tides) or gradual events (such as sea level rise) resulting in changes in coastline. Erosion will occur in high flow conditions and accretion in the opposite situation. A typical example of a dynamical coastal system with continuous coastline changes is a bird foot delta. Monitoring changes is important for local risk assessment and shoreline management.

What does CoastObs offer?

CoastObs can map the changes in land-water boundary over time. By using archived imagery, it is also possible to go back in time with a high resolution. Mapping of the land-water boundary is based on the Normalized Difference Water Index (NDWI; Mc.Feeters, 1996).

How was the data validated?

Satellite-retrieved Chl-a concentrations were validated against ground data collected close in time to the satellite overpass. The Chl-a samples collected by UVIGO, CNR, USTIR and HZ were analysed at University of Stirling using High Performance Liquid Chromatography (HPLC).

Case study example: Wadden Sea, the Netherlands

Figure 1. Morphology change over the Ems estuary 1987-2016 showing the contour lines in a sub-region (Lin, 2016).

For the highly dynamic Ems estuary in the Wadden Sea, images from different optical satellites (Landsat 5 and 8 and Sentinel-2) were used to visualise long-term changes. Water and land surfaces are classified using image-specific thresholds to extract the water surface area for each year from de NDWI images. Then, contour line of the water surface was generated for further investigation of changes.


  • The output is the instantaneous shoreline without considering tides. Tidal level and/or beach profile are needed to compare changes over time.
  • Spatial resolution and accuracy depend on the satellite data used in the analysis (30 m Landsat; 10 m Sentinel-2; less than 2 m with very high-resolution images).