Very High Resolution Parametric and Non- Parametric Sartomography Methods for Monitoring Urban Areas Structures

Synthetic Aperture Radar (SAR) is the only way to evaluate deformation of the Earth’s surface from space on the order of centimeters and millimeters due to its coherent nature and short wavelengths. Hence, by this means the long term risk monitoring and security are performed as precisely as possible. Traditional SAR imaging delivers a projection of the 3-D object to the two dimensional (2-D) azimuth-range (x-r) plane. Due to the side looking geometry of SAR sensors problems like foreshortening, layover and shadow should be dealt with. For overcoming the problem of layover in high urban environment different SARtomography methods are developed. The aim of SARtomography methods are not only overcoming layover problem but unambiguous 3-D and 4-D (space-time) dynamic map of the city can be achieved. The assumption, made by the classical interferometric techniques (i.e.: PSI), the present of a single scatterer per pixel, neglects the fact of occurrence of mulitiple scatterers. For instance PSI initial assumption is the presence of single sactterer in each azimuth-cell range. However, this assumption is not plausible in a high rise urban environment where suffers from the presence of multiple scatter. Furthermore, by the advent of Very High Resolution sensors like Cosmos-Skymed (1,2,3) constellation SARtomography is revolutionized. Needless to say that due to the very high resolution of the images (up to 1m resolution), precise shape and deformation of each individual building can be obtained. Nonetheless, it has to be noted that the increasingly impact of layover on new generation of VHR sensors. To this end several practical SARtomography methods such as first order model and Non Linear least Square (NLS) are introduced. This paper has presented the capacity of the new class of VHR spaceborne SAR systems, like COSMO-Skymed, for TomoSAR processing in high urban environment. However, particular problems related to the side looking geometry of SAR has proved to be more obvious comparing to the previous generation of high resolution sensors (10 m resolution): Layover is one of them. The main aim of this paper is at comparing SARtomography methods and their advantage and disadvantage to the older version of SAR methods like PSI for monitoring high urban environment. The project has been implemented on Very High Resolution (VHR) Cosmo-skymed Stripmap mode (up to 3m azimuth-range resolution) images from Astana-Kazakhstan. Like PSI, TomoSAR benefits greatly from the high resolution of Cosmo-skymed data, as the density of coherent pixels and the signal-to-clutter ratio increase significantly with resolution. The results reveal that the number of permanent scatterers found by NLS and first order model are far more than PS method which indicate the superiority of these methods in overcoming the layover problem in high environment urban areas comparing to the PSI; Besides, the mean deformation velocity, height and coherence of every scatterer were obtained by SARtomographic methods are compared with PSI. As it comes to tomographic processing methods different factors must be taken into account such as the accuracy in estimating the heights, computational cost and the model order selection. Results indicate that the first order method has low computational cost but it suffers severely from side-lobes on the other hand the computational cost of NLSM is very high but it is an accurate method as long as the correct model order is selected. SARTomography proved to be an efficient method for detecting multiple scattrers and layover removal in high urban environment.