Observing Inter- and Intra-Annual Glacier Changes and Lake Loading Effects from Synthetic Aperture Radar Remote Sensing

摘要

The new generation of spaceborne Synthetic Aperture Radar (SAR) missions, such as the TerraSAR-X/TanDEM-X or the Sentinel-1 mission, acquire data with unprecedented spatial and temporal resolution. The short revisit times of the SAR satellites and the independence from solar illumination and weather conditions enable regular monitoring of topography changes and displacement variations. This is particularly useful for applications in high-mountain areas where optical spaceborne sensors provide only limited data due to cloud coverage, and where in-situ measurements are difficult to acquire on a regular basis. This thesis focuses on the observation of inter- and intra-annual glacier changes and lake loading-induced deformations in high-mountain areas. An essential element of the work is the exploitation of various SAR processing methods to extract seasonal displacements and topography changes that occur with magnitudes of a 10th of a millimetre over several months up to a couple of metres within a few days. Focus is laid on the reduction of signal disturbance sources that are particularly related to the mountainous setting. The main sources of error in this context are atmospheric effects and radar penetration into the ground. The investigations are carried out in three exemplary case studies in Kyrgyzstan, Central Asia, as introduced below. In the first study, seasonal variations of glacier surface velocities are analysed on the example of the Inylchek Glacier. A detailed analysis of the flow behaviour of the bending area is achieved by applying feature tracking on TerraSAR-X data acquired in 2009 and 2010. Results show that the 3 m resolution StripMap data allows for a distinction between different surface velocities of adjacent longitudinal ice-streams. The processing of SAR data acquired every eleven days also allows the extraction of the surface motion increase during the melting period. In addition, a 100% speed-up of the lower, rather stagnant ablation area of the Southern Inylchek Glacier is captured in summer time, which has been related to the glacier lake outburst flood of Lake Merzbacher. The second analysis deals with glacier elevation changes at the Inylchek Glacier. Three 10 m spatial resolution digital elevation models (DEMs) are generated for February 2012, March 2013 and November 2013 from bistatic TanDKM X data, and are compared with each other as well as to the C-band Shuttle Radar Topography Mission (SRTM) DEM from the year 2000. It is shown that the intra-annual comparison between the high-resolution TanDKM X DEMs is useful to even depict seasonal changes, although uncertainties are high due to approximated radar penetration depths. A new method to assess the glacier elevation changes in void areas based on elevation binning and slope steepness is introduced. Mass balances calculated for the decadal changes agree well with results of previously published long-term studies. The third investigation is related to water-level induced deformations occurring at the Toktogul Reservoir. Significant changes of the water level lead to load changes on the crust, which results in subsidence in the case of water level increase and uplift in case of water level lowering. The deformation analysis is accomplished by applying the Small BAseline Subset (SBAS) method on 2004 - 2009 Envisat and 2014 - 2016 Sentinel-1 data. Whereas both datasets allow to depict an overall deformation trend within the analysed time periods, only Sentinel-1 imagery shows that the observed deformations are closely correlated to seasonal water level changes. Results are verified by a very good agreement to deformation rates estimated from elastic forward modelling with a numerical Earth model. The phase of the SAR data is heavily influenced by atmospheric effects, which are mainly related to seasonally varying stratification of the atmosphere and the daily cycle in evaporation at the open water surface. To determine the best method for the reduction of atmospheric effects, numerical weather model-based approaches are compared with phase-dependent approaches. It is found that phase-based methods are currently superior to weather model-based approaches for the application in high-mountain areas, whereas the power-law methods works best.