Signal decomposition of hypertemporal Sentinel-1 time series to optimize the information gain for land applications

点击翻译按钮获取中文摘要

With the launch of the European Copernicus Program and the related launch of the satellites Sentinel-1A & B for the first time dense C-band time series are available reliably and free of charge. By combining the data of both satellites repetition rates of six days are feasible, which introduces a new temporal dimension to the SAR data acquisitions. This new dimensions allows and requires innovative data analysis strategies that optimize the SAR data processing as well as the gain of information.Overall aim of this project is the development and the validation of a suited Sentinel-1 data analysis strategy. The proposed approach aims at the temporal decomposition of these dense SAR time series. The gained spatiotemporal information will be analyzed in terms of its suitability as database for specific applications. At the same time a novel speckle filter will be developed based on the temporal decomposition approach. This novel filter has the characteristic to filter only in the time domain. Thus, the original geometric resolution of the original data will be maintained. Additionally, an extension of this filter which operates in the spatiotemporal domain will be developed as well.The temporal signal will be decomposed into components of different frequencies. High frequency components represent random backscattering variations caused by speckle or precipitation effects. Medium and low frequency components are related to biophysical processes such as soil moisture changes or plant growth. The investigation of which component is related to which biophysical process is scope of this project. Furthermore, the minimum requirements with regard to the temporal sampling rate to allow for the detection of the impact of biophysical processes will be investigated. Additionally the potential of the C-band phase related parameters InSAR coherence and phase closure error as soil moisture indicators will be analyzed.According to the results the usability of the temporal components will be investigated for several applications. Major focus is put on landcover mapping including REDD+ requests, the detection of landcover change, the analysis of wetland dynamics and the delineation of soil moisture indicators.