Back to:

Wave Hazard Assessment for Selected Sites on the West Coast of Dominica, West Indies

  Back to:

Caribbean Disaster Mitigation Project


5. Bathymetry: Detailed Information for Detailed Results

Three primary sources were integrated to create the composite bathymetric map: nautical charts, a Landsat MSS image, and a SPOT image.

The nautical charts were raster-scanned digital copies of paper charts. They included Prince Rupert Bay at a scale of 1:22,000, Roseau Roads and Woodbridge Bay at a scale of 1:12,000, and a chart of Dominica at a scale of 1:75,000. The paper charts were produced by the Defense Mapping Agency of the USA, but the bathymetric information on them was taken from work done by the British Admiralty over a century ago. Although the individual soundings are still accurate, for the most part, they are sparse. There are distances of hundreds of meters between soundings, and the printed numbers themselves are scaled at tens of meters wide. It was necessary to upgrade this data with information from remote sensing.

The MSS image was acquired in 1986. The image consists of four bands with a nominal resolution of 80 meters. There are heavy clouds over the east and center of the island, with a few scattered clouds over Pointe Michel and Coulibistri.

The SPOT image was acquired on December 20, 1994. This image has three bands, green, red, and near-infrared, with a nominal resolution of 20 meters. The 10-meter panchromatic band for this image was not usable. This image has clouds as well, but only about half as much as on the MSS. The west coast is mostly clear, with some haze.

There were differences on the order of 100 to 300 meters in the registration between the sources of data used in this project. In the absence of definitive information, all of the data was registered to the SPOT satellite image. This was accomplished by visually matching features between data sets and either moving or ‘rubber sheeting’ to register to the SPOT image. Some misregistrations remain, especially in the delineation of the coastline. It should be remembered that the coastline from the nautical charts is a smoothed, conservative registration, whereas the coastline from the remote sensing shows individual rocks and sometimes includes clear shallows.

The remote sensing contained bathymetric information because the amount of sunlight reaching the bottom, reflecting, and exiting the water depends on the depth and the color of the bottom. Light from the sun is reflected, scattered, or absorbed by water, depending on the wavelength. By measuring the difference in the reflected light at various wavelengths, especially in the blue, green and red wavelengths, the depth underwater may be estimated. (Lyzenga, 1978) When using several bands of the electromagnetic spectrum, the depth estimate variable is effectively continuous, with no lowest resolution.

During the analysis of the SPOT image it was determined that a boundary layer, possibly a thermocline or halocline, existed at approximately 10 to 12 meters. To account for this, two different regression equations were derived: one for 0 to -11 meters, and one for -11 meters down. No useful information was obtained deeper than -20 meters, however, given the wave heights impacting the leeward shore, this is not thought to impact the study.

Overall, the impact of the improved bathymetry on the wave hazard study appears to be minor. However, tests reveal that there are a few specific sites where there may be as much as a 10 % increase or 30% decrease in wave impact. This is due to the tendency for the techniques used to derive bathymetry for the original study to overestimate depth in some areas. The additional detail available with the new bathymetric data set will allow much greater detail in the spatial variation of wave energy along the coastline.

Previous Page | Next Page