Remote sensing techniques adapted to high resolution mapping of tropical coastal marine ecosystems (coral reefs, seagrass beds and mangrove)

Abstract

The need for high resolution maps in the management of tropical environments is increasing and emphasized by the rapid anthropogenic development often occurring in coastal zones. In areas subject to humid tropical climate, such as the West Indies, cloud coverage often disturbs image acquisition by orbital imagery.Moreover, as thesetropical coastal ecosystems, i.e.,coral reefs,mangroves and seagrass beds, are intricate and geographically complex, high resolution data must be used to accurately restore these features. Digitized aerial photographs meet these requirements by providing higher-resolution images than orbital remote sensing devices. This paper reports on the method applied to map coral reefs, mangroves and seagrass beds in the Bay of Robert in Martinique Island (French West Indies) by digitizing true-colour aerial photographs. But, as these photographs present two drawbacks, i.e., they are not geographically oriented and their margins are distorted, a reference map was used to design a rectification model. Aerial photographs were first subjected to true-colour digitization. After separation of the red, green and blue bands, an unsupervised classification was achieved, then various masks each corresponding to a given range of depths were built and applied to the initial image. The new images obtained were analysed. Great carewas taken in the selection oftraining sites to gather the pixels characterized by a high spectral similarity which corresponded to precise themes. After ground identification, these training sites enabled a supervized classification to be established, then a confusion matrix was built. The final map consisted of 31 themes. Each theme was related to a particular feature of the benthic communities. The error matrix and subsequent field samplings confirmed the reliability of the final map. In addition to their use in ecosystem studies, e.g., calculation of surface areas; biocoenosis distribution, etc., these thematic maps are of overriding importance for the management of coastal areas.