Hyperspectral remote sensing of wild oyster reefs

The invasion of the wild oyster Crassostrea gigas along the western European Atlantic coast has generated changes in the structure and functioning of intertidal ecosystems. Considered as an invasive species and a trophic competitor of the cultivated conspecific oyster, it is now seen as a resource by oyster farmers following recurrent mass summer mortalities of oyster spat since 2008. Spatial distribution maps of wild oyster reefs are required by local authorities to help define management strategies. In this work, visible near infrared (VNIR) hyperspectral and multispectral remote sensing was investigated to map two contrasted intertidal reef structures: clusters of vertical oysters building three-dimensional dense reefs in muddy areas and oysters growing horizontally creating large fiat reefs in rocky areas. A spectral library, collected in situ for various conditions with an ASD spectroradiometer, was used to run Spectral Angle Mapper classifications on airborne data obtained with an HySpex sensor (160 spectral bands) and SPOT satellite HRG multispectral data (3 spectral bands). With HySpex spectral/spatial resolution, horizontal oysters in the rocky area were correctly classified but the detection was less efficient for vertical oysters in muddy areas. Poor results were obtained with the multispectral image and from spatially or spectrally degraded HySpex data, it was clear that the spectral resolution was more important than the spatial resolution. In fact, there was a systematic mud deposition on shells of vertical oyster reefs explaining the misclassification of 30% of pixels recognized as mud or microphytobenthos. Spatial distribution maps of oyster reefs were coupled with in situ biomass measurements to illustrate the interest of a remote sensing product to provide stock estimations of wild oyster reefs to be exploited by oyster producers. This work highlights the interest of developing remote sensing techniques for aquaculture applications in coastal areas. (C) 2016 Elsevier Ltd. All rights reserved.