Synthetic Aperture Radar Tomography (TomoSAR) provides 3D information from the illuminated scene, giving thus the opportunity to infer 3D structure information from forests. In this paper, the temporal variations of forest structure are analyzed in a TomoSAR dataset covering a period of five years over the Traunstein forest, in the south of Germany. First, a TomoSAR inversion algorithm based on Compressive Sensing is applied to retrieve the 3D backscatter and then forest structure is estimated from it. Finally, forest structure dynamics are analyzed confronting the maps of forest structure estimated from TomoSAR to reference data from other sources such as high resolution Lidar data, high resolution optical images and terrestrial measurements.