Mutual Calibration of TanDEM-X DEMs on Natural Targets Derived from Sentinel-1 Time Series

Abstract

The TanDEM-X mission, has been very successful in generating a global digital elevation model (DEM) at 12m posting, that is widely used for a variety of scientific applications. High-quality DEMs are derived by exploiting the phase difference between two Synthetic Aperture Radar (SAR) images of TanDEM-X bistatic acquisitions. The two acquisitions are simultaneous and therefore unaffected by temporal changes. In this sense, newly acquired InSAR DEMs are often used to monitor topographic changes, by performing DEM differencing. Small residual phase and baseline uncertainties in the order of a few millimeters may cause residual offsets and tilts in the order of a few meters in single-scene DEMs from the TanDEM-X mission. As a consequence, the mutual calibration of two DEMs becomes crucial for monitoring changes in the same order of the residual offsets and tilts. Traditionally, single InSAR-derived DEM are calibrated considering a set of tie-points, which are a priory selected points with known location and height. They need to be timely consistent with the acquired DEM and are mainly GPS or LiDAR measurements. The selection of such measurements is therefore often performed manually, which is a very time-consuming practice. Moreover, the availability of such external measurement in the considered area endangers the success of monitoring topographic changes on most regions that are difficult to access. To overcome these limitations, we elaborate on the selection of natural tie-points derived from the pre-selection of persistent scatterer candidates from Sentinel-1 time-series. Since Sentinel-1 coverage is global, with a revisit time of 12 days, it is possible to have a time-consistent selection of natural tie-points. Natural targets from Sentinel-1 under certain conditions, such as a sufficient interferometric coherence and signal-to-noise ratio to assure a high quality of the selected tie-points, are assumed to be appropriate for mutual calibration of two single-pass TanDEM-X DEMs and therefore for the derivation of accurate DEM changes. This method is envisioned to be self-sufficient, without the need for GPS or LiDAR measurements, and to be fully automatic without any manual intervention. In this work, we will detail the proposed approach, together with practical examples of mutual TanDEM-X DEM calibration and validation.