Characterization of forest using tomographic SAR imaging techniques at P-band : preparation of the BIOMASS mission
Résumé
The characterization of forests is crucial for understanding the carbon cycle. P-band Synthetic Aperture Radar (SAR) is a particularly effective means of identifying key descriptors of tropical or temperate forests, thanks to its ability to penetrate dense volumetric environments and measure their biomass. In this context, ESA is launching the BIOMASS mission, which features the first P-band spaceborne SAR sensor dedicated to the study of forests. SAR tomography is a powerful tool for determining the 3D structure of forests based on their electromagnetic response. Existing techniques often separate canopy and ground reflections using SAR tomography and polarimetric diversity as well as advanced processing methods. This thesis demonstrates that comparable performance, with improved stability, can be achieved using single-polarization data and parametric tomographic focusing. This thesis proposes a simple and sufficient modeling of forests using a low-dimensional representation. The vertical reflectivity density of tropical and temperate forests is modeled at P-band, with a ground component represented by a Dirac function and volume associated with a narrow peak. The performance of this low-dimensional parametric tomographic approach is assessed using P-band data collected during the TropiSAR campaign over a tropical forest, and using the TomoSense dataset acquired over the temperate forest. Results show that this approach can accurately and reliably estimate key structural parameters of the observed tropical and temperate forests. A low resolution dataset is simulated from the TropiSAR data in order to provide a representative sample of the future BIOMASS data, yielding better performances than those from methods that will be used in its upcoming acquisition phase. The proposed approach has been extended to the dual-baseline interferometric configuration, which is the longest phase of the mission, providing adequate and well-designed estimates of the structure of the forest. The estimate accuracy from this minimal tomographic configuration has been improved using the synergy between the two modes of the BIOMASS mission through a regularization method.
Auteurs, date et publication :
Auteurs Pierre-Antoine Bou
Date : 2025