Modeling of Subsurface Scattering from Ice Sheets for Pol-InSAR Applications

Abstract

Remote sensing is a fundamental tool to measure the dynamics of ice sheets and provides valuable information for ice sheet projections under a changing climate. There is, however, the potential to further reduce the uncertainties in these projections by developing innovative remote sensing methods. One of these remote sensing techniques, the polarimetric synthetic aperture radar interferometry (Pol-InSAR), is known since decades to have the potential to assess the geophysical properties below the surface of ice sheets, because of the penetration of microwave signals into dry snow, firn, and ice. Despite this, only very few studies have addressed this topic and the development of robust Pol-InSAR applications is at an early stage. Two potential Pol-InSAR applications are identified as the motivation for this thesis. First, the estimation and compensation of the penetration bias in digital elevation models derived with SAR interferometry. This bias can lead to errors of several meters or even tens of meters in surface elevation measurements. Second, the estimation of geophysical properties of the subsurface of glaciers and ice sheets using Pol-InSAR techniques. There is indeed potential to derive information about melt-refreeze processes within the firn, which are related to density and affect the mass balance. Such Pol-InSAR applications can be a valuable information source with the potential for monthly ice sheet wide coverage and high spatial resolution provided by the next generation of SAR satellites. However, the required models to link the Pol-InSAR measurements to the subsurface properties are not yet established. The aim of this thesis is to improve the modeling of the vertical backscattering distribution in the subsurface of ice sheets and its effect on polarimetric interferometric SAR measurements at different frequencies. In order to achieve this, polarimetric interferometric multi-baseline SAR data at different frequencies and from two different test sites on the Greenland ice sheet are investigated. This thesis contributes with three concepts to a better understanding and to a more accurate modeling of the vertical backscattering distribution in the subsurface of ice sheets. First, the integration of scattering from distinct subsurface layers. These are formed by refrozen melt water in the upper percolation zone and cause an interesting coherence undulation pattern, which cannot be explained with previously existing models. This represents a first link between Pol-InSAR data and geophysical subsurface properties. The second step is the improved modeling of the general vertical backscattering distribution of the subsurface volume. The advantages of more flexible volume models are demonstrated, but interestingly, the simple modification of a previously existing model with a vertical shift parameter lead to the best agreement between model and data. The third contribution is the model based compensation of the penetration bias, which is experimentally validated. At the investigated test sites, it becomes evident that the model based estimates of the surface elevations are more accurate than the interferometric phase center locations, which are conventionally used to derive surface elevations of ice sheets. This thesis therefore improves the state of the art of subsurface scattering modeling for Pol-InSAR applications, demonstrates the model-based penetration bias compensation, and makes a further research step towards the retrieval of geophysical subsurface information with Pol-InSAR.