Development and Applications of an Interferometric Ground-Based SAR System Tadashi Hamasaki (1), Zheng-Shu Zhou (2), Motoyuki Sato (2) (1) Graduate School of Environmental Studies, Tohoku University Aramaki Aoba, Sendai, 980-8579, Japan hamasaki@cneas.tohoku.ac.jp (2) The Center for Northeast Asian Studies, Tohoku University Kawauchi, Sendai, 980-8576, Japan zszhou@cneas.tohoku.ac.jp, sato@cneas.tohoku.ac.jp ABSTRACT An Interferometric Ground-Based Synthetic Aperture Radar (In-GB-SAR) system has been developed. The system employs similar principles as the airborne or space borne SAR. In the developed In-GB-SAR system, the technique of differential interferometry is used to detect small changes on the target at different times and under different conditions. Experimental results demonstrate that the developed system is very efficient and can detect small changes of the scale of 2cm at the range distance of 5m. Moreover, within the framework of the study, a new migration algorithm for signal processing was developed. The GB-SAR has a wide range of applications to environmental monitoring, resources management as well as monitoring of different natural phenomena. Key words: Interferometoric Ground Based SAR (In-GB-SAR), Differential Interferimetory, detect and measure minute changes INTRODUCTION Based on the principles of conventional SAR algorithms, we have developed a ground-based SAR system (GB-SAR). It has fully polarimetric and interferometric functions to observe various natural and artificial targets and can be successfully used for the environmental and other studies. As an active sensor, the GB-SAR radiates microwave and registers the amplitude and phase information of the backscattered waves based on the same principles as the conventional SAR. A SAR can be used at any time of day and night and under any atmospheric conditions (e.g., through clouds, rain, fog, smoke and dusty environment) where optical sensors are incapable. Moreover, utilizing the backscattering information obtained from multifrequency, multiangle radar systems as well as the information contained in the phase difference, it is possible to carry out superior analyses about the objects in relation to both spectral and spatial characteristics. As the developed GB-SAR can obtain data with a very high spatial resolution, it enables to conduct extremely precise analyses of features using radar polarimetry and interferometry. Unlike conventional airborne and space borne SARs that observe large areas, the spatial extent of the observed area by the GB-SAR is small. Therefore, it is used for investigation of local artificial structures, such as skyscrapers, bridges, tunnels, and a nuclear power plant. In addition, it can be used for monitoring of a base rocks crash, a snow slide and landslide movements as well as for the studies related to archaeological investigation and mine detection. INTERFEROMETRIC GROUND-BASED SAR SYSTEM The system diagram and overview of the developed GB-SAR system are shown in Figs.1 and 2 and they illustrate a stepped-frequency radar system that is built VNA (Vector Network Analyzer) to measure the scattering information in the frequency domain from a low frequency (min. 50MHz) to a very high frequency up to 20GH. It is obviously different from conventional SAR system, whose data acquisition is usually carried out in a very narrow frequency band with a fixed center frequency. Another difference is the size of the antenna scanning aperture. For an airborne and apace borne SAR it is possible to acquire data along a very long survey line, but planar 2D scan with many scan lines is very difficult. GB-SAR has the limitation of the scanning aperture, but a planar 2D scan can be realized easily. Antenna
Figure 1. System diagram of the GB-SAR. Figure 2. GB-SAR system. scanning in 1D or 2D on the rail controlled by an antenna positioner and a position controller achieves synthetic aperture radar. A scanning aperture is 20m and 1.5m in the horizontal and vertical directions, respectively. A PC controls data acquisition and antenna position. Polarimetry is realized by using dual polarized horn antennas. Hence, interferometric and polarimetric functions can be realized. EXPERIMENT The purpose of the experiment of this study was to demonstrate the developed In-GB-SAR system, could detect small displacements with a high accuracy. Applying differential interferometry algorithm, small changes of a target surface between different times under different conditions can be detected and measured. The target of the experiment was an exterior wall of a wooden house shown in Fig. 3. In general, measurement of wooden targets is difficult by using radar since a reflectivity of wood is lower than that of other targets, such as, metal or stone. In a differential interferometric SAR measurement, two SAR measurements were carried out under two different conditions which are called a normal case and a special case. In the normal case defined as Case1, the target is in a usual condition as shown in Fig. 3, while in the special case, defined as Case2, some additional objects were attached to the wall and the window was opened to make small displacements as shown in Fig. 4. The additional targets were two wooden plates, a wooden box and an aluminium plate. The specifications of additional targets are shown in Table 1. The experiment configurations are shown in Fig. 5. The distance from the antenna to the target is about 5m. The antenna elevation angle was 30 degrees. The parameters of the experiment are shown in Table 2. Figure 4. GB-SAR target in Case2. Figure 3. GB-SAR target in Case1.
Table 1. Sizes of additional targets Table 2. Parameters of the experiment. If a high accuracy of displacement measurement is expected, 2-D data acquisition should be carried out. However, it takes very long time, thus is inefficient. In order to realize short measurement time and efficiency, we performed 1-D data acquisition and applied the differential interferometric technique. SIGNAL PROCESSING Imaging Reconstruction Firstly, the frequency domain data acquired by VNA was transformed to the time domain by IFFT before SAR processing was carried out. In the SAR processing, a modified technique of the diffraction stacking algorithm was introduced. The conventional diffraction stacking algorithm can be written as shown in Equation (1). Px ( ', z') = f( τ, x, z) dx (1) r r r where 2 2 2 2 t + t + r + r ( x x') z ( x x') z τ = (2) ν ( x ', z ) is the position of the reconstruction, ( t x, z ) is the transmitter position, ( t x r, z ) denotes the receiver position, r ν is the propagation speed of the electromagnetic wave in the medium, and τ is the travel time of electromagnetic wave from the transmitting antenna to the target and back to the receiving antenna via the point of the image reconstruction. Figure 5(a). Front view of the experiment configuration Figure 5(b). Side view of the experiment configuration.
Figure 8. Windowing g (θ r ). Figure 7. Principle of the modified algorithm. The modified algorithm is shown as in Equation (3). P ( x ', z ') = f (τ, xr, z r ) g (θ r ) dxr (3) where g (θ r ) is a windowing function in a spatial domain. The windowing function is according to angle from the antenna radiation center to the position of the reflector. In the case of using horn antenna, reflection signals are stronger from antenna radiation center than one from side positions. The principal of the modified algorithm is shown in Fig. 7. A Gaussian windowing function used in this processing is shown in Fig. 8. Phase Extraction Phase information is necessary in the interferometry, but it is not possible to extract the phase directly from the imaging data. In this study, the Short Time Fourier Transform (STFT) method has been used to extract phase information. The STFT can be described as shown in Equation (4). S ( r, ω ) = s ( r ) h* ( r r )e jω r dr (4) where s (r ) is a profile in a spatial domain along each one trace of reconstructed image, h( r ) is a windowing function in the spatial domain. RESULT AND DISCUSSION Figure 9(a). Reconstructed image using the modified technique. Figure 9(b). Reconstructed image using conventional technique.
Figure 10(a). Phase distribution in Case1. Figure 10(b). Phase distribution in Case2. Reconstructed images using the modified and the conventional SAR processing techniques respectively are shown in Fig. 9. By using the modified technique, noise and artifacts are reduced and the boundaries of the house wall can be observed clearly. And we can confirm that the modified technique demonstrates better results. In the analysis, we found that the metal door caused strong reflection at the center of the image in Fig. 9. For both Case1 and Case2, the phase information is extracted from each reconstructed image by using the modified technique. Phase information images of both cases at spatial frequency of 25m-1 are shown in Fig. 10. Actually, if we extract phase information from Fig. 9(b), we cannot obtain clear phase due to the existing artifacts. The phase difference on the plane positioned on the wall can be calculated from Fig. 10 (a) and (b), as shown in Fig. 11. The projection image in Fig. 11 corresponds to the area indicated in Fig.4. From Fig. 11, it is possible to verify changes of small displacements by the opened window, the aluminium plate and wooden plates on the wall. However, unexpected noise caused by the system appears above at the height 4. And it can be confirmed by the raw time domain data. And interference caused by the metal door is happened in lower part. One trace of phase difference at azimuth distance of 3.5m is extracted from Fig. 11 and shown in Fg.12. The thickness of the wooden plate 1 can be calculated from the phase difference in Fig. 11. The calculated thickness is 1.12cm. Compared to its actual thickness of 2cm, it does not agree completely. However, we can check and detect the changes of small displacement. CONCLUSION Figure 11. Phase difference on the wall. Figure 12. Change of the phase difference at Azimuth Distance of 3.5m in Fig. 11.
A new In-GB-SAR system has been developed. Applying differential interferometry, detection of small displacements that exceeds the range resolution defined by the used frequency band was realized with phase information. Further improvement on the measurement precision is necessary. And a modified image reconstruction algorithm was introduced and its efficiency was demonstrated by the reconstructed images. In future an amplifier to the radar system may be installed and it might work well for distant targets. REFERENCES [1] D. Leva, G. Nico, D. Tarchi, J.F. Guasch and A.J. Sieber, Temporal Analysis of a Landslide by Means of a Ground-Based SAR Interferometer, IEEE Trans. Geoscience and Remote sensing, vol. 41, No. 4, APRIL 2003. [2] Z.-S. Zhou, Application of a Ground-Based Polarimetric SAR System for Environmental Study, Doctoral Dissertation of Tohoku University, Japan, August 2003. [3] Z.-S. Zhou, W.-M. Boerner, and M. Sato, "Development of a Ground-Based Polarimetric Broadband SAR System for Non-Invasive Ground-truth Validation in Vegetation Monitoring, IEEE Transactions on Geoscience and Remote Sensing, in press [4] Z.-S. Zhou, and M. Sato, Ground-Based Polarimetric SAR Systems for Environment Study, IEEE AP-S 2003 Digest, vol. 1, pp.202-205, Columbus, June 2003. [5] D. Tarchi, H. Rudolf, M. Pieraccini and C. Atzeni, Remote monitoring of buildings using a ground-based SAR: application to cultural heritage survey, International Journal of Remote Sensing, 2000, vol.21, No.18, 3545-3551.