Ph Defense

Giovanni FRATI

"Hyperspectral and Full-waveform LiDAR synchronized remote sensing applied to Pays-de-la-Loire region’s sandy coasts "

Tuesday, December 9th at 2 p.m. - Faculty of Sciences and technicals - University of Nantes

Abstract : Remote sensing allows an accurate monitoring of coastal sand dune ridges. These latter protect the retrolittoral areas against marine submersions. In the context of the Pays-de-la-Loire region sandy coast monitoring, two main parameters are analyzed: 1) morphology, giving access to sedimentary stocks and movements, and 2) vegetation cover, allowing to assess the dune ridges stability. Dune morphologies and their végétation cover are studied using laser ranging (LiDAR and their chemical and pigment compositions are obtained using hyperspectral imaging (HIS). Full waveform LiDAR and hyperspectral imaging data were both acquired via bi-hatch aircraft. The acrosstrack scanning (whiskbroom) of the 1064 nm laser beam were synchronized with the alongtrack scanning (pushbroom) of the camera. The dune typology were determined by using hyperspectral image hierarchical classification, based on the vegetation cover spectral properties. The mapping of Ammophila arenaria (marram grass), white dune stabilisation proxy, is difficult because of its proximal spectral signature with other plants of the dune and because of its high phenological variability from a year to another. Ground ranging is only possible when the vegetation cover allows it, which is not the case with the marram grass, because of its foliar component density. However, full-waveform analysis shows that this particular leaf structure induces a caracteristical deformation of the signal, which allows to identify the marram grass between other plants. As a result, this deformation can now be used to map the marram grass and to correct dune topographical determination errors.

Thesis Director : Patrick LAUNEAU, Professor Nantes University
Co-supervisor : Marc ROBIN et Cyril MICHON, GEOFIT