Skip to Main content Skip to Navigation
Theses

Inversion des formes d'ondes LiDAR pour l'estimation des caractéristiques des cultures et des forêts par des techniques probabilistes et variationnelles

Abstract : The use of LiDAR in remote sensing allows a precise description of the vegetation cover architecture. The aim of this thesis is the development of LiDAR data inversion approaches using physical and statistical signal modeling in order to estimate the biophysical properties of dominant crops (wheat, maize) of the South-West of France and a forest cover in China. The work firstly focused on estimating LAI and crop height by small footprint LiDAR waveforms inversion. A realistic crop waveform simulations database is performed using the Radiative Transfer Model (MTR) DART. The inversion consists in using the Look up Table technique which consists of looking for the closest simulation to the actual observation. The second inversion approach focused on LAI profile estimation of the forest trees. A variational approach to estimate LAI by waveform inversion is proposed. It relies on a simplified MTR and LAI profile smoothing technique based on Markov chains. The Bayesian formulation of the problem leads us to a non-linear cost function. It is minimized using a new multi-scale gradient technique. The developed approaches show clearly their performance by applying them to real crop data (corn and wheat) and forest.
Document type :
Theses
Complete list of metadatas

Cited literature [137 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02070697
Contributor : Abes Star :  Contact
Submitted on : Wednesday, December 4, 2019 - 2:38:09 PM
Last modification on : Thursday, October 15, 2020 - 3:09:10 AM
Long-term archiving on: : Thursday, March 5, 2020 - 9:40:24 PM

File

2018TOU30303a.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02070697, version 2

Collections

Citation

Sahar Ben Hmida. Inversion des formes d'ondes LiDAR pour l'estimation des caractéristiques des cultures et des forêts par des techniques probabilistes et variationnelles. Sciences de la Terre. Université Paul Sabatier - Toulouse III, 2018. Français. ⟨NNT : 2018TOU30303⟩. ⟨tel-02070697v2⟩

Share

Metrics

Record views

128

Files downloads

887