Ice Sheet Subsurface Density from Polarimetric and Interferometric SAR

Information about ice sheet subsurface properties is crucial for understanding and reducing related uncertainties in mass balance estimations. Key parameters like the firn density, stratigraphy, and the amount of refrozen melt water are conventionally derived from in situ measurements or airborne radar sounders. Both types of measurements provide a great amount of detail, but are very limited in their spatial and temporal coverage and resolution. Synthetic Aperture Radars (SAR) can overcome these limitations due to their capability to provide day-and-night, all-weather acquisitions with resolutions on the order of meters and swath widths of hundreds of kilometers. Long-wavelength SAR systems (e.g. at L- and P-band) are promising tools to investigate the subsurface properties of glaciers and ice sheets due to the signal penetration of up to several tens of meters into dry snow, firn, and ice. Understanding the relationship between geophysical subsurface properties and the backscattered signals measured by a SAR is ongoing research. Two different lines of research were addressed in recent years. The first is based on Polarimetric SAR (PolSAR), which provides not only information about the scattering mechanisms, but also has the uniqueness of being sensitive to anisotropic signal propagation in snow and firn. The second is related to the use of interferometric SAR (InSAR) to retrieve the 3D location of scatterers within the subsurface. Particularly multi-baseline InSAR allows for tomographic imaging (TomoSAR) of the 3D subsurface scattering structure. So far, the potential of the different SAR techniques was only assessed separately. In the field of PolSAR, modeling efforts have been dedicated to establish a link between co-polarization (HH-VV) phase differences (CPDs) and the structural properties of firn [1]. CPDs have then been interpreted as the result of birefringence due to the dielectric anisotropy of firn originating from temperature gradient metamorphism. Moreover, the relation between the anisotropic signal propagation and measured CPDs depends on the vertical distribution of backscattering in the subsurface, e.g. generated by ice layers and lenses, which defines how the CPD contributions are integrated along depth. Up to now, assumptions of density, firn anisotropy, and the vertical backscattering distribution were necessary to invert the model, e.g. for the estimation of firn thickness [2]. However, the need for such assumptions can be overcome by integrating InSAR/TomoSAR techniques. In the fields of InSAR and TomoSAR for the investigation of the ice sheet subsurface, recent studies are mainly concerned with the estimation of the vertical backscatter distribution, either model-based or through tomographic imaging techniques. InSAR models exploit the dependence of the interferometric volume decorrelation on the vertical distribution of backscattering. By modeling the subsurface as a homogeneous, lossy, and infinitely deep scattering volume, a relation between InSAR coherence and the constant extinction coefficient of the microwave signals in the subsurface of ice sheets was established in [3]. This approach approximates the vertical backscattering distribution as an exponential function and allows the estimation of the signal extinction, which is a first, yet simplified, indicator of subsurface properties. Recent improvements in subsurface scattering modeling [4], [5] showed the potential to account for refrozen melt layers and variable extinctions, which could provide information about melt-refreeze processes and subsurface density. With TomoSAR, the imaging of subsurface features in glaciers [6], and ice sheets [5][7][8] was demonstrated. Depending on the study, the effect of subsurface layers, different ice types, firn bodies, crevasses, and the bed rock (of alpine glaciers) was recognized in the tomograms. This verified that the subsurface structure of glaciers and ice sheets can result in more complex backscattering structures than what is accounted for in current InSAR models. SAR tomography does not rely on model assumptions and can, therefore, provide more realistic estimates of subsurface scattering distributions. This study will address a promising line for future research, which is the combination of PolSAR and InSAR/TomoSAR approaches to fully exploit their complementarity and mitigate their weaknesses. As described above, on the one hand, PolSAR is sensitive to the anisotropic signal propagation in snow and firn, even in the absence of scattering, but provides no vertical information. On the other hand, InSAR (models) and TomoSAR allow assessing the 3-D distribution of scatterers in the subsurface, but provide no information on the propagation through the non-scattering parts of firn. In a first step, an estimation of firn density was achieved by integrating TomoSAR vertical scattering profiles into the depth-integral of the PolSAR CPD model [9]. This approach is in an early experimental stage with certain limitations. The density inversion can only provide a bulk value for the depth range of the signal penetration and measurements at several incidence angles are required to achieve a non-ambiguous solution. Furthermore, multi-baseline SAR data for TomoSAR are currently only available from a few experimental airborne campaigns. Finally, the density estimates have to be interpreted carefully, since the underlying models are (strong) approximations of the real firn structure. This could be addressed in the future by an integration with firn densification models. Nevertheless, this combination of polarimetric and interferometric SAR techniques provides a direct link to ice sheet subsurface density, without parameter assumptions or a priori knowledge, and the first density inversion results show a promising agreement with ice core data [9]. This contribution will present first results of the density inversion, discuss its limitations and will show investigations towards a more robust and wider applicability. One aspect will be the use of InSAR model-based vertical scattering profiles instead of TomoSAR profiles, which reduces the requirements on the observation space and increases the (theoretical) feasibility with upcoming spaceborne SAR missions. [1] G. Parrella, I. Hajnsek and K. P. Papathanassiou, "On the Interpretation of Polarimetric Phase Differences in SAR Data Over Land Ice," in IEEE Geoscience and Remote Sensing Letters, vol. 13, no. 2, pp. 192-196, 2016. [2] G. Parrella, I. Hajnsek, and K. P. Papathanassiou, “Retrieval of Firn Thickness by Means of Polarisation Phase Differences in L-Band SAR Data,” Remote Sensing, vol. 13, no. 21, p. 4448, Nov. 2021, doi: 10.3390/rs13214448. [3] E. W. Hoen and H. Zebker, “Penetration depths inferred from interferometric volume decorrelation observed over the Greenland ice sheet,” IEEE Transactions on Geoscience and Remote Sensing, vol. 38, no. 6, pp. 2571–2583, 2000. [4] G. Fischer, K. P. Papathanassiou and I. Hajnsek, "Modeling Multifrequency Pol-InSAR Data from the Percolation Zone of the Greenland Ice Sheet," IEEE Transactions on Geoscience and Remote Sensing, vol. 57, no. 4, pp. 1963-1976, 2019. [5] G. Fischer, M. Jäger, K. P. Papathanassiou and I. Hajnsek, "Modeling the Vertical Backscattering Distribution in the Percolation Zone of the Greenland Ice Sheet with SAR Tomography," IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 12, no. 11, pp. 4389-4405, 2019. [6] S. Tebaldini, T. Nagler, H. Rott, and A. Heilig, “Imaging the Internal Structure of an Alpine Glacier via L-Band Airborne SAR Tomography,” IEEE Transactions on Geoscience and Remote Sensing, vol. 54, no. 12, pp. 7197–7209, 2016. [7] F. Banda, J. Dall, and S. Tebaldini, “Single and Multipolarimetric P-Band SAR Tomography of Subsurface Ice Structure,” IEEE Transactions on Geoscience and Remote Sensing, vol. 54, no. 5, pp. 2832–2845, 2016. [8] M. Pardini, G. Parrella, G. Fischer, and K. Papathanassiou, “A Multi-Frequency SAR Tomographic Characterization of Sub-Surface Ice Volumes,” in Proceedings of EUSAR, Hamburg, Germany, 2016. [9] G. Fischer, K. Papathanassiou, I. Hajnsek, and G. Parrella, “Combining PolSAR, Pol-InSAR and TomoSAR for Snow and Ice Subsurface Characterization,” presented at the ESA POLinSAR Workshop, Online, Apr. 2021