Growing stock volume estimation in temperate forsted areas using a fusion approach with SAR Satellites Imagery
Forest monitoring plays a central role in the context of global warming mitigation and in the assessment of forest resources. To meet these challenges, significant efforts have been made by scientists to develop new feasible remote sensing techniques for the retrieval of forest parameters. However, much work remains to be done in this area, in particular in establishing global assessments of forest biomass. In this context, this Ph.D. Thesis presents a complete methodology for estimating Growing Stock Volume (GSV) in temperate forested areas using a fusion approach based on Synthetic-Aperture Radar (SAR) satellite imagery. The investigations which were performed focused on the Thuringian Forest, which is located in Central Germany. The satellite data used are composed of an extensive set of L-band (ALOS PALSAR) and X-band (TerraSAR-X, TanDEM-X, Cosmo-SkyMed) images, which were acquired in various sensor configurations (acquisition modes, polarisations, incidence angles). The available ground data consists of a forest inventory delivered by the local forest offices. Weather measurements and a LiDAR DEM complete the datasets. The research showed that together with the topography, the forest structure and weather conditions generally limited the sensitivity of the SAR signal to GSV. The best correlations were obtained with ALOS PALSAR (R2 = 0.61) and TanDEM-X (R2 = 0.72) interferometric coherences. These datasets were chosen for the retrieval of GSV in the Thuringian Forest and led with regressions to an root-mean-square error (RMSE) in the range of 100─200 m3ha-1. As a final achievement of this thesis, a methodology for combining the SAR information was developed. Assuming that there are sufficient and adequate remote sensing data, the proposed fusion approach may increase the biomass maps accuracy, their spatial extension and their updated frequency. These characteristics are essential for the future derivation of accurate, global and robust forest biomass maps.