|
EARSeL: 2nd Workshop on Remote Sensing of the Coastal Zone Porto, Portugal, 9-11 June 2005 |
SESSION COASTAL HABITAT |
Eric De Oliveira, Jacques Populus
IFREMER, Plouzane, 29280, France
eric.de.oliveira@ifremer.fr, jacques.populus@ifremer.fr
Seaweeds have limited spatial extent, but they represent an essential factor in terms of biological production and their use in industry is more and more important. As a consequence, spatial distribution and biomass assessments need to be undertaken. New techniques of global observation (satellite and aerial imagery, …) allow detecting the presence of seaweed beds in the inter-tidal domain, but not to distinguish the different species.
The aim of this study is to model seaweed species distribution according to environmental parameters. We focused on fixed species of algae, such as Fucus sp.. First, we identified environmental parameters, such as substratum nature, immersion time, exposure, etc., which determine seaweed distribution. Secondly, we used field sampling to compute the distribution laws of seaweed according to the environmental parameters selected. Thirdly, we used the distribution laws and the environmental parameters to perform predictive mapping of seaweed belts with a fuzzy logic method.
Seaweed presence is directly dependent on substratum nature. In the inter-tidal domain, we used an alternative because seaweed beds can be observed directly. We detected seaweed presence with Spot satellite imagery. The second parameter is immersion time. For each elevation value (surveyed by Lidar), we converted water tidal levels into annual percentages of immersion. Seaweed distribution according to this parameter should be comparable in different areas. The third environmental variable used was exposure to waves. During the fixation phase, seaweeds cannot stand a high level of exposure. We used a model of wave propagation to delineate areas of different exposure level.
The presence of seaweed species for each parameter was estimated from field sampling, with 3D measurements (dGPS). Higher and lower limits of dominant seaweed belts were delineated. With reference to the three environmental variables selected, the distribution laws of each seaweed species were estimated.
Classifications by fuzzy logic have been applied with eCognition software. Two phases have been used in this method: the first was a phase of segmentation to obtain polygons. Each polygon was homogeneous according to the environmental parameters selected: vegetation cover, immersion time and exposure level. During the second phase, we implemented the distribution laws estimated from field sampling. The probability of seaweed presence depends firstly on the seaweed cover, secondly on immersion time and thirdly on exposure level. Finally a presence probability for each targeted species was performed.
Last Update: 2005-03-17