Algorithm of Interferometric Coherence Estimation for Synthetic Aperture Radar Image Pair Andrey Sosnovsky and Victor Kobernichenko Ural Federal University, pr. Mira, 19, Yekaterinburg, 620002, Russian Federation sav83@e1.r Abstract. Interferometric coherence is an important indicator of relia- bility for interferograms obtained by interferometric synthetic aperture radar (Interferometric SAR, InSAR). Areas with low coherence values are unsuitable for interferometric data processing. Also, it may be used as a classification parameter for various coverage types. Coherence magni- tude can be calculated as an absolute value of the correlation coefficient between two complex SAR images with averaging in a local window. The problem in coherence estimation is in its dependence on phase slope caused by relief topography (topographic phase). A method for suppres- sion of the topographic phase influence is proposed, based on the spatial phase derivation. Keywords: Synthetic aperture radar images, InSAR systems, Coherence estimation 1 Introduction Interferometric data processing for extraction of information about the Earth terrain and its changes becomes one of the general guidelines in the develop- ment of contemporary space-based radar systems together with the implemen- tation modes of ultra-high spatial resolution (1-3 meters) and full–polarimetric processing. The method of space-based radar interferometry implies a joint pro- cessing of the phase fields obtained by simultaneous scattering of the terrain with two antennas or by non-simultaneous scattering with one antenna moved by two different parallel orbits [1, 4]. This method combines high accuracy of the phase measurements with high resolution of the synthetic aperture radars (SAR) technology. Technology of the differential SAR interferometry (InSAR) makes possible to get maps of the elevation changes between the radar passages. The stages of the interferometric processing are implemented in specialized software systems for the remote sensing data processing such as the SARscape, IMAGINE Radar Mapping, Photomod Radar, RadarTools. So we can talk about Informa- tion technology (IT) of the digital elevation models (DEM) generation by remote sensing data. However, for practical application of these technologies, one has to overcome a number of significant problems. Two general problems of interferometry are 172 the temporal and spatial decorrelation of the received data and the problem of phase unwrapping, i.e. a recovery of the absolute phase information from the relative phase, which is wrapped into [−π, π]-interval. Another important field of investigation in radar interferometry is selection of the most efficient processing algorithms and obtaining the experimental estimates of the generated DEM accuracy. This work is devoted to the first of the mentioned problems, i.e. to the data decorrelation (summary temporal and spatial) and its estimation. 2 Interferometric coherence Interferometric coherence is an important indicator of suitability of the data scene obtained by a radar remote sensing system for the further processing and solving the final problem, i.e. generation of digital elevation model or terrain changes map. The coherence factor is calculated as the absolute value of the correlation coefficient between samples of two complex radar images (single-look data complex, SLC) got in the local windows Σ ż1 (m, n) · z̄2 (m, n) γˆ0 = |ρˆ˙0 | = p , (1) Σ|ż1 (m, n)|2 · |z̄2 (m, n)|2 where z1(2) (m, n) is the SLC samples (z̄1(2) (m, n) are complex-conjugate sam- ples) [2–5], γˆ0 takes values in interval [0, 1], near-zero values correspond to areas of high or full decorrelation, which are not suitable for interferometric data pro- cessing. The values higher than 0.5 mean good data correlation. However, this approach entails some problems because, in fact, a random variable is estimated here, but not a random process. So, any phase gradients caused by both natural topography variability and by point-of-view geometry (remote sensing radar systems have a side-scattering configuration) lead to the degradation of the estimate (1). Its value depends on the slope and tends towards the value |ρ12 (N )|ρ=0 , i.e. a bias of the estimate for independent Gaussian values of the correlation coefficient (N is the number of samples) [3], which in practice takes the value about 0.1–0.3. Thus, coherence loses its properties as measure of the quality of the interferogram, its value becomes dependent on relation between topographic and fluctuation components of the phase. 3 Differential phase coherence estimate To eliminate the effect described above, the following modification of the coher- ence estimate can be offered taking into account influence of the topographic component. Modified coherence evaluates not the samples z1 (m, n) and z2 (m, n) of the SLC–image pair, but the following values: ẇ1 (m, n) = ż1 (m, n) · z̄1 (m + 1, n), ẇ2 (m, n) = ż2 (m, n) · z̄2 (m + 1, n), (2) where the new phase values of the w1 (m, n) and w1 (m, n) will characterize the slope of the topographic phase in the direction of growth of the mth picture co- ordinate. So, this operation performs the phase derivation along one coordinate. 173 Similarly, one can use the gradients along the nth coordinate. Expression for the coherence estimation using values (2) looks as follows: s Σ ẇ1 (m, n) · w̄2 (m, n) γˆ1 = p . (3) Σ|ẇ1 (m, n)|2 · |w̄2 (m, n)|2 Although this estimate is non-Gaussian, it is wealthy, and it can be shown that it is insensitive to the linear phase trend. 4 Experimental results Test now the work of the coherence estimation algorithms for RADARSAT–1 data (wavelength 56 mm). Figure 1 presents a fragment of the radar image of an area with surfaces of different reflectance including surfaces with volume scatter- ing and the water surface, which has generally low coherence. Signal differential phase for the given fragment has a slope in the horizontal direction of the order of 0.3 radians per sample (due scattering geometry). The coherence maps of the fragment were constructed using conventional estimate γˆ0 and modified estimate γˆ1 with the sampling size 11 × 11 are presented in Figs. 2a, 2b. Fig. 1. A RADARSAT-1 radar image scene One can see that the first map is degraded since different surfaces give the same low coherence values regardless on the surface type. The map obtained 174 a) Traditional estimate γˆ0 b) Modified estimate γˆ1 Fig. 2. Coherence map calculated using modified estimate (Fig. 2b) has a good sensitivity to the surface type. However, the estimate has a larger bias at low values than the γˆ0 , and, so, it requires increasing the sample value towards to γˆ0 (Fig. 3a, 3b). A quantitative accuracy assessment for the scene is not available because of poor reference DEM for this territory. a) Traditional estimate γˆ0 b) Modified estimate γˆ1 Fig. 3. Coherence estimates biases for different sample sizes (N =2, 5, 10, 25) 5 Conclusion A modified method for estimation of spatial coherence of the interferometric radar images pairs is developed. The method consists in calculation of the cor- 175 relation between pairs of complex multiplications of neighbour elements. The research is implemented on radar images RADARSAT–1. The result shows that the modification allows one to solve the problem of estimate degradation under the differential phase trend, which always takes place in side-scattering radar systems. Acknowledgment The work was supported by the RFBR grants Nos. 13-07- 12168, 13-07-00785. References 1. Elizavetin,I. V., Ksenofontov, E. A.: Resultaty experimentalnogo issledovaniya voz- mozhnosti pretsizionnogo izmereniya reliefa Zemli interferentsionnym metodom po dannym kosmicheskogo RSA [The results of experimental research of precious Earth relief measurement by interferometric method with space-based SAR]. Issledovaniya Zemli iz kosmosa, 1, 75–90 (1996) (in Russian) 2. Touzi, R., Lopes, A., Bruniquel,J., Vachon, P.W.: Coherence Estimation for SAR Imagery. IEEE Trans. Geosci. Remote Sensing, 37, 135–149 (1999) 3. Lopez-Martinez, C., Pottier, E.: Coherence Estimation in Synthetic Aperture Radar Data based on Speckle Noise Modeling. Applied Optics, 46(4), 544–558 (2007) 4. Kobernichenko, V. G., Sosnovsky, A. V.: Particular Qualities of Digital Elevation Maps Generation in Interferometric SAR Technology. SPIIRAS Proceedings, 5(28), 194–208 (2013) 5. Kobernichenko, V. G., Sosnovsky,A. V.: Coherence Estimation in Interferometric SAR Data Processing. 2014 24th Int. Crimean Conference Microwave and Telecom- munication Technology (CriMiCo–2014). 7–13 September, Sevastopol, Crimea, Rus- sia, 24, 1153-1154 (2014) 176