Saltmarshes vegetation typology and mapping: the combination of field data with multispectral satellite images.

Saltmarshes are complex ecotones forming an interface between aquatic and terrestrial habitats. These wetland ecosystems of high ecological value provide a wide range of ecosystem services. Fringing fresh and saltwater bodies, they are submitted to diverse gradients of natural and anthropic origin, and are subject of high conservation issues. In a context of global change, vegetation mapping appears as a useful tool to understand and preserve the dynamics of saltmarsh plant communities. As part of the management plan of the French National Natural Reserve (NNR) of Arès and Lège – Cap Ferret saltmarshes, a technique combining a spatially explicit in situ vegetation survey with multispectral satellite images was tested for vegetation mapping.
Between June and September 2011, 676 vegetation records were carried out. They were positioned systematically every 50 m to cover the whole range of the NNR (2 km²). After removing records that were considered as floristically heterogeneous, the combination of correspondence and cluster analyses enabled to identify six groups of records representing characteristic plant communities. In parallel, several spectral bands retrieved from high resolution mulispectral images (Pléiades Satellite 1B) acquired via the RTU program (CNES) in April, August and December 2013, were selected and combined with indices of vegetation and water presence (NDVI, NDWI). An interactive supervised classification using data from vegetation records was applied on this false composite image to create a map of the vegetation of the NNR. This final classified map had an overall accuracy of 63%, with varying success between the different groups of records. Although improvements remain necessary to increase the reliability and accuracy of the map, this study indicates that combining high resolution remotely sensed images with field observations can be used as a basis to aid in saltmarsh vegetation mapping with the objective of monitoring vegetation dynamics.