Quality evaluation of DInSAR results from the phase statistical analysis

Resumen

The Earth is continuously monitored by sensors capable of providing diversified information about several phenomena. In particular Synthetic Aperture Radar (SAR) is able to observe the Earth in all-day and all-weather conditions. Differential SAR Interferometry (DInSAR) regards a set of techniques able to measure displacement of the terrain. In the last decades these techniques have become very powerful geodetic tools for land deformation monitoring and have gained a prominent role in several fields, ranging from geophysical to engineering, from commercial to civil protection purposes. Measurements of direction and magnitude of landslides phenomena, monitoring of subsidence movements due to natural events or human activity are some examples of DInSAR applications. These techniques allow not only the analysis of a single deformation episode but also to estimate, with millimetric precision, the temporal evolution of large-scale deformations phenomena occurring on the Earth surface. This research covers transversally several aspects within the DInSAR applications and particularly is focused on analyzing step by step the DInSAR algorithm of the Coherent Pixel Technic (CPT) developed by the Universitat Politècnica de Catalunya (UPC) highlighting the conditions for a correct estimation of the solution. The achieved precision can be really high but anyway it is important to understand all the limitations that can change from one data-set to another. Particular attention has been paid on analyzing the different deteriorating factors that can affect the estimation, such as the quality of the data, the atmospheric artifacts or the spatial and temporal sampling of the phenomenon. In this regard, a theoretical study on the propagation of the interferometric phase noise has been carried out and validated through simulations. For reducing the impact of the interferometric phase noise normally only pixels with a certain quality can be processed. These pixels can be selected through different methods described in detail in the literature. On the one hand, each selection technique is able to detect a specific kind of target at a determined resolution. On the other hand, there is an obvious compromise between the quality and the pixels¿ density over the scene. Indeed, working with high densities could increase the reliability and the accuracy of the final result as long as poor quality pixels are not included in the processing. In this regard, in this Thesis a combination of the different kinds of targets has been studied with the objective of a joint processing at different resolutions. Finally the constant growth of the spaceborne, airborne and ground-based SAR sensors in terms of technology and number of available devices could be the beginning for exploiting at the same time different data-set in a joint processing. The integration of data coming from different sensors with different orbits, different resolution, different carrier frequencies and incidence angles can help in covering in a better way a complex displacement scenario and can provide more complete information.