Linking Changes of TomoSAR 3-D Reflectivity Profiles and Pol-InSAR Measurements in Forest Scenarios

Abstract

Changes of 3D synthetic aperture radar (SAR) reflectivity profiles in forest scenarios are effected by changes of both the 3D distribution of structural elements (induced by growth, management, logging, mortality, disturbance etc.) and their dielectric properties (induced by rainfall, droughts, daily water cycle, seasonality, etc.). Their parameterization at different spatial and temporal scales in terms of changing 3D reflectivity components and their unambiguous separation and interpretation are critical missing aspects. Time series of tomographic (TomoSAR) reconstructions would provide a way to image directly 3D reflectivity changes, but the limited vertical resolution and / or the imaging algorithm can introduce interpretation ambiguities. A different way to proceed is to consider the corresponding TomoSAR coherence matrices. Reflectivity changes can then be represented and separated in terms of subspace directions common to two TomoSAR coherence matrices at different times. Experimental results suggest that the components with maximum reflectivity increase and decrease are enough to characterize the change between the respective profiles. Not only, but each of them can be well modelled by means of a Dirac-delta profile, and their characterization requires only a height information. As a consequence, the reconstruction of the TomoSAR reflectivity change can be performed by means of a reduced number of interferometric acquisitions. Concepts of this kind become attractive for spaceborne implementations, in which suitable baseline distributions to realize significant height resolution and radiometric sensitivity repeatedly in time may not be achievable. In this context, the objective of this work is to understand how far (even single-baseline) polarimetric interferometric (Pol-InSAR) acquisitions at different times can be used to separate and characterize reflectivity change components provided an initial TomoSAR observation. Understanding and establishing this link would allow on the one hand the interpretation of changes in Pol-InSAR measurements, and on the other hand the direct exploitation of their (higher) sensitivity to reflectivity changes. First of all, differences between subspace-based change characterization methodologies of the kind reported in after an appropriate extension to TomoSAR including polarimetric information will be addressed. The application of the developed scheme to a TomoSAR / Pol-InSAR case is then further analysed. The analysis will be aimed at discussing primarily the effect of the vertical wavenumber(s) in the Pol-InSAR acquisition(s), the allowed vertical resolution of the reconstructed reflectivity changes, and the potential of the changing components to describe different types of changes (structural or dielectric). The analysis will be supported by real airborne data acquired by the DLR’s airborne platforms E/F-SAR over the Traunstein forest (South of Germany) in relevant campaigns with temporal separations from hours up to months with underlying daily, weekly and seasonal reflectivity changes.