Context and objectives
Coastal defence and nature conservation authorities need detailed vegetation maps of the Belgian coast for policy planning and evaluation. For several decades, vegetation and topographical maps of the mobile dunes, mud flats and salt marshes have been produced for the Belgian coast by means of visual interpretation of aerial photographs. This tech-nique however, does not provide enough detail about the vegetation classes in compari-son to the effort needed, and the technique is not as precise as it should be. By using sensors with a higher information content (hyperspectral sensors or digital camera’s) a better ratio between cost and effectiveness can be achieved.
The present project aims for the development of a more objective, a more detailed and a more cost effective method based on the application of airborne hyperspectral data. Hy-perspectral remote sensing is a passive technique that records the reflected sunlight in several fine spectral bands. This offers an almost continual reflectance spectrum for each pixel. This should enable the very detailed differentiation of plant species, associations and structures.
Both existing and newly acquired hyperspectral data are used in the project. All data has been radiometrically calibrated and geometrically and atmospherically corrected. After these pre-processing steps a supervised approach is adopted, i.e. extensive botanic field surveys are used to make a spectral library of the plant species, plant associations and canopy structures to be distinguished.