Satellite-Derived Bathymetry and Coastal Zone Monitoring

The Physical Basis of Satellite-Derived Bathymetry

SDB is based on the way sunlight interacts with the water column. When solar radiation reaches the sea surface, some of it is reflected, some is absorbed and the rest passes into the water. The light that gets through interacts with suspended particles, dissolved organic matter, and the seabed. A portion of the light that reaches the seabed is reflected back through the water column and detected by the satellite sensor.

The amount of light reflected from the seabed depends on the depth of the water, the clarity of the water, the type of seabed, and the wavelength of the light. In clear water, shorter wavelengths such as blue penetrate further than longer wavelengths such as red or near-infrared. As the depth increases, the reflected signal from the seabed becomes weaker. This relationship between the reflected radiance and water depth is what enables bathymetric estimation.

The attenuation of light in water follows the Beer-Lambert law: I = I₀ × e^(−kz), where I is the light intensity at depth z, I₀ is the intensity at the surface, and k is the diffuse attenuation coefficient. The value of k varies with wavelength and water clarity, determining the maximum depth that can be sensed optically.

Since light diminishes quickly in water, satellite-derived bathymetry is usually limited to depths from zero up to about 20 to 30 metres in clear water. In turbid water, the effective limit may be only a few metres or even less. The Gulf region ranges from very clear offshore water to turbid nearshore areas affected by dredging, river discharge, and algal blooms. SDB projects therefore require careful choice of the imagery according to the water conditions.

Multispectral Sensors and Spectral Bands

SDB makes use of multispectral sensors that record the reflected radiance in distinct spectral bands. Common satellite platforms include Sentinel-2, Landsat 8 and 9, WorldView, and Pleiades. The spatial resolution differs. Sentinel-2 has a 10-metre resolution in the visible and near-infrared bands, while commercial high-resolution satellites can provide imagery at sub-metre resolution.

The blue band is the most effective for penetrating water. The green and red bands give additional information about bottom reflectance and the properties of the water column. Near-infrared is quickly absorbed by water and is generally used to detect the boundaries between land and water and to identify surface phenomena.

In optical bathymetry, the ratio method uses the ratio of reflectances in two spectral bands to reduce the impact of the type of seabed. The logarithm of the reflectance ratio is linearly related to water depth under certain assumptions about the optical properties of the water.

The ratio transform is widely employed because it partly compensates for variations in the seabed albedo. Sandy bottoms, seagrass, dark rock, and coral have different reflectance characteristics. A single-band approach alone would mistake a dark bottom in shallow water for a bright bottom in deeper water. The ratio method lessens this confusion.

From Imagery to Bathymetric Surface

The general process for producing satellite-derived bathymetry involves several stages.

1. Image selection and preprocessing: Choose images that are free of clouds, have low sun glint, low turbidity, and minimal surface waves. Make corrections for atmospheric effects to obtain surface reflectance values.
2. Sun glint correction: Eliminate or reduce the specular reflection from the sea surface. Sun glint can saturate pixels and destroy depth information.
3. Land and water separation: Classify the pixels as land, water, or mixed shoreline. The shoreline itself serves as an important zero-depth reference point.
4. Water column correction: Take into account the variation in water clarity and the depth-dependent attenuation.
5. Ground truth calibration: Use acoustic survey soundings, GNSS-surveyed bottom points, or known depths to establish an empirical relationship between spectral values and water depth.
6. Depth inversion and surface generation: Apply the calibrated algorithm to the entire image and generate a bathymetric grid.

Geomak regards SDB as a data layer that must be validated. Without ground-truth soundings, SDB provides only a relative depth model. Calibration using even a limited number of acoustic or direct depth measurements transforms the model into an absolute bathymetric surface with quantifiable uncertainty.

Ground Truth and Calibration

Ground-truth data is essential. Acoustic single-beam or multibeam echosounder surveys provide actual depth measurements along selected transects. These soundings are used to train the empirical algorithm and to assess the residual error. The accuracy of SDB depends on the quality, spatial distribution, and number of ground-truth points.

• Good practice: Collect ground-truth soundings over the full range of depths represented in the imagery, from the shoreline to the optical depth limit.
• Good practice: Include a variety of bottom types in the calibration dataset so that the algorithm is not biased towards one type of seabed.
• Good practice: Match the timing of the ground-truth survey as closely as possible to the time when the satellite image was acquired to avoid errors due to changes in the seabed.
• Poor practice: Calibrating only in a small area and then applying the algorithm to a much larger area that has different optical characteristics.
• Poor practice: Using ground-truth data collected weeks or months after the satellite image, particularly in dynamic coastal areas.

After calibration, Geomak checks the resulting surface using independent soundings that were not used in training the model. The comparison gives an estimate of the root mean square error and a spatial map of discrepancies. This map is then used to define confidence zones within the bathymetric surface.

Strengths and Limitations

SDB has clear advantages in certain situations. It offers extensive spatial coverage at a low additional cost when compared to vessel-based surveys. It can reach areas that are hard to access by boat, such as very shallow water, coral flats, and restricted zones. It is also helpful for temporal monitoring, since satellite archives allow historical comparisons of depth over years or decades.

The limitations are equally important. The accuracy of SDB depth