Consistent assimilation of spaceborne radar interferometry (InSAR) data into integrated terrestrial systems

The goal of this project is to develop procedures that allow an optimal integration of spaceborne radar interferometry (InSAR) data and their variance-covariance information into geo-scientific process models. We recently initiated a pilot study to investigate sub-aerial surface displacements caused by natural and anthropogenic determinants in the Lower-Rhine-Embayment. For this study we used 47 SAR from the ERS satellites with a horizontal resolution of about 10 m over a swath width of typically 100 km and determined the surface displacement by the differential InSAR (D-InSAR) permanent scatter technique. The result is compared with leveling measurements given by GeoBasis NRW.

The limiting factors concerning the accuracy of this technique are the satellite orbit errors (baseline geometry), the tropospheric delay (hydrostatic and "wet" components), the accuracy of the reference elevation model and errors during the interferometric processing of the data. There are approches, which consider these factors, but due to the large amount of data and parameters - in our test area up to 100 differential interferometric products can be stacked - they all performed stepwise.

We propose an one step approach, where all measurements are integrated - also the terrestrial leveling data and other available informations (e.g. rain radar, hydrological and weather models) - in a consistent model. The application of tailored numerical strategies (e.g. sparse techniques, array-algebra, finite covariance functions, Toeplitz solvers, ...) is necessary to achieve a rigorous and consistent modeling. But due to the large number of observations and parameters this approach is computational extreme extensive and therefore benefits from the SimLab TerrSys and the interdisciplinary contacts within Geoverbund ABC/J.