Abstract : The intertidal mudflat of the Bourgneuf Bay (SW France) is an important physical and ecological environment. Generally, sediment mobility is an important problem for both coastal dynamics (nourishment areas, erosion areas) and the development of shellfish or biological production. It is controlled by hydrodynamics, eolian processes as well as the cohesive sedimentary behaviour. The cohesive phenomenon is itself controlled by the physical properties of sediments. Hyperspectral imaging is an effective tool for sustainable management of complex coastal systems from regional to global scales. Airborne remote sensing images have the potential to provide synoptic maps of the intertidal sediments' distributions.
The main objective of this thesis is to map parameters relevant to the understanding of the sediments' dynamics from hyperspectral remote sensing data DAIS and ROSIS. These parameters include sediment grain size, moisture content and mineralogy, as well as the presence or absence of biofilm, all having an influence on the sediment cohesiveness/erodability. With this purpose in mind, we performed the Modified Gaussian Model [MGM] to extract bio-geophysical properties from reflectance spectra from each pixel of the image mudflat. The MGM algorithm is a powerful tool to deconvolve spectra into two components, firstly Gaussian curves for the absorptions bands (Chl-a, H2O), and secondly a straight line in wavenumber range for the continuum (scattering, grain size). New methodologies in hyperspectral remote sensing have been developed and implemented during this study, such as the Automated MGM approach [AMGM]. This work also contains the analysis of laboratory spectra (ASD Fieldspec3) and sedimentological analysis (laser grain size analysis, XRD, SEM, calcimetry, etc.) that allowed us to find the relationship between the specific absorptions (e.g. chlorophyll-a; H2O), the components of the continuum (e.g. grain size) and the physical properties. It also contains a detailed description of the sediments spectral behaviour with respect to dehydration. Laboratory analyses enabled us to test whether the MGM continuum is a good proxy for water content and grain size estimation. Finally, we applied the AMGM method to DAIS and ROSIS images. We obtained very precise maps of grain size, water content and biomass parameters.