IJCSNS International Journal of Computer Science an Network Security, VOL.9 No.5, May 2009 59 Differential Synthetic Aperture Raar Interferometry (DINSAR) for 3D Coastal Geomorphology Reconstruction Mage Marghany an Mazlan hashim, Natural Tropical Resources Information & Mapping Research Group (NATRIM Research Group) Faculty of Geoinformation Science an Engineering Universiti Teknologi Malaysia 81310 UTM, Skuai, Johore Bahru, Malaysia Summary This paper introuces a new metho for three-imensional (3D) coastal geomorphology reconstruction using ifferential synthetic aperture interferometry (DInSAR). The new metho is base on an integration between fuzzy B-spline algorithm an DInSAR metho. DInSAR algorithm is involve two parts: (i) 3D map simulation which is base on interferogram simulation an (ii) satellite orbit parameters. 3D coastal geomorphology reconstruction is realize by fuzzy B-spline algorithm with the mipoint isplacement metho an the terrain roughness. Consequently, fuzzy B-spline was use to eliminate topographic phase from the interferograms. The stuy shows the DInSAR technique provies information about coastal geomorphology change with accuracy of ± 0.1 m. Keywors: DInSAR, interferogram, Fuzzy B-spline algorithm, 3D reconstruction. 1. Introuction Synthetic Aperture Raar interferometry (InSAR) is a relatively new technique for 3D topography mapping [10]. Scientists an researchers have been efine InSAR as a technique that utilizes interference of waves for precise etermination of istance [5]. In SAR interferometry path length ifferences with millimeter accuracies can be etecte base on the interferometric phase generate by conjugating two SAR images of the same scene at ifferent times with slightly ifferent viewing angles [2]. In this context, it coul be a major tool for 3D coastal geomorphology reconstruction in real time. Consequently, synoptic ata over large areas at comparatively low cost can be prouce by InSAR. The coastal geomorphology features etc., spit, unes an beach profile can be reconstructe by SAR interferometry. Accoring to Zekber et al. [10], topographic information as well as movement information can be acquire from the phases. In fact, phases are corresponing to ifferential range change in the interferogram for two or more SAR images of the same scene. Recently, Luo et al.,[2] have introuce a technique which is base on utilization three pass ifferential interferometry (TPDI) to measure topography isplacement. They reporte that the isplacement will be result in component calle eformation phase in interferomateric phase, if the topography surface eforme at interval of SAR repeat [2]. In this paper, we aress the question of utilization fuzzy- B-spline in 3D topography reconstruction before phase unwrapping. In fact, there are several factors coul be impact the accuracy of DEMs are erive from phase unwrapping. These factors are involve raar shaow, layover, multi-path effects an image misregistration, an finally the signal-to noise ratio (SNR) [9]. This emonstrate with RADARSAT-1 SAR fine moe using integration between DInSAR [2] an fuzzy B-spline algorithm Mage an Mazlan [4]. Three hypotheses examine are: (i) fuzzy B-spline which is base on triangle-base criteria an ege-base criteria can be use as filtering technique to reuce noise before phase unwrapping. (ii) 3D topography reconstruction can be prouce using satisfactory phase unwrapping (iii) high accuracy of eformation rate can be estimate by using the new technique. 2. DInSAR-Fuzzy B-spline Proceures The proceures of involving fuzzy B-spline in DInSAR are shown in Fig. 1. Following, Luo et al.[2] if the surface isplacement is as a result of single or cumulative surface movement occurre between the acquisition times of three RADARSAT-1 SAR images S 1, S 2 an S 3, the component of surface isplacement in the raar-look Manuscript receive May 5, 2009 Manuscript revise May 20, 2009
60 IJCSNS International Journal of Computer Science an Network Security, VOL.9 No.5, May 2009 irection,, contributes to aitional interferometric phase as 4 4 4 (( R R ) ) (( R R ) ) r 1 2 1 3 (1) where R, R 1 2 an R3 are slant range from satellite to target respectively at ifferent time, is the RADARSAT- 1 SAR fine moe wavelength which is about 5.6 cm for C HH - ban. Finally r is the projection of isplacement P P 1 2 on look of sight (LOS) S1 P1. where, s s M, s are the master an slave complex amplitues, respectively. The numerical values 10-5 an 0.5 are threshol values use in this stuy. Accoring to Yang et al. [9], the weighte square error is efine as: I I 2 we ( j y, k x) a x b y c s ( j y, k x) y 1 x 1 I I I I (4) where s I an I=(s,M) are onate both pixels location at j,k in slave an Master images while y,x onate the relative coorinates of ajacent pixel from j,k an an x, y { 1,0,1}. Finally, ai, bi c I are the complex coefficient. Equation 2 can be written base on weight error as. Fig. 1. Fuzzy B-spline block iagram for 3D Coastal geomorphology reconstruction by DInSAR The phase ifference, isplacement as R 4 R. (2), only from the surface There are various ecorrelation factors can be effecte the phase unwrapping such as geometrical, thermal, temporal, an Doppler Centroi. These factors are contribute to reuce the signal-to-ratio (SNR). In fact, the phase unwrapping coul be ue to low SNR [10]. In this context, noise filtering is essential stage prior to phase unwrapping. In such tropical zone as Malaysia which is ominate by heavy vegetation covers which are the main source for eccorelation problem uring InSAR or DInSAR proceures. This ecorrelation coul be effecte of amplitues of the complex master an slave images. Furthermore, Unreliability of the wrappe phases coul be raise up ue to ecorrelation. Following, Yang et al., [9], the egree of coherence can be efine base on the basic rule of fuzzy theory as ' s s ' 2 ( ( j y, k x) ( j y, k x) v( j, k )) 2 ys xs (1 (2s 1) 1) var( v) (5) where 2s+1 is winow size which is taken here as 3 x 3, v is the aitive noise, is sum of ifference phase D an v. Then, fuzzy B-spline 3D surface topography reconstruction was introuce by Mage an Mazlan [4], an moifie to involve phase ifference an correlation coefficient of master an slave complex amplitue patches is given by M O ' C i,4 ( p) j,4 ( q) k ) i0 j0 M O ' S ( p, q) C S ( p, q) M O i0 j0 m,4 ( p) l,4 ( q) k ) m0 l0 ( i, 4 p) an j, 4( q) an { CS} ( D ) (6) are two basis B-spline functions, are the biirectional control net. The curve points S(p,q) are affecte by { w e } in case of p [ r i, r i P1] an q [ r j, r j ] P ' 1, where P an P are the egree of the two B-spline basis functions constitute the B-spline surface. Two sets of knot vectors are knot p=[0,0,0,0,1,2,3,,o,o,o,o], an knot
IJCSNS International Journal of Computer Science an Network Security, VOL.9 No.5, May 2009 61 q=[0,0,0,0,1,2,3,.,m,m,m,m]. Fourth orer B-spline basis are use j,4(.) to ensure continuity of the tangents an curvatures on the whole surface topology incluing at the patches bounaries. Accoring to Tsay an Chen [8], the quality of etermine DEM is function of the accuracy of GCPs which collecte using GPS uring the RADARSAT-1 SAR pass over on 1999 an 2004 along the coastline of Kuala Terengganu. Finally, height map is create an statistically compare with groun fiel ata to acquire precisely coastal geomorphology s DEM. notice that the new metho preserves etaile eges with iscernible fringes. Inee, Fig 3. Shows smooth interferogram, in terms of spatial resolution maintenance, an noise reuction, as compare to traitional conventional methos [2,5,8,10]. 3. Result an Discussion The coherence image of topographic pair along the Kuala Terengganu mouth river is shown in Fig. 2. Clearly, the coherence values are range between 0.0 an 1.0 where 0.0 value is represente incoherence while 1 is represente perfect coherence. Fig. 2, however, shows the high coherence value of 0.8 which is correspone to urban an sany areas while low coherence value is correspons to vegetation zone ue to the impact of ecorrelation in tropical zone such as Malaysia. Inee, baseline ecorrelation is major contribution of noise as well as the changes in atmospheric conitions in which is causing ifficulties in phase reconstruction [10]. Therefore, ecorrelation coul attribute for low accuracy of igital elevation moel [9]. Fig. 3. Interferogram of eformation pairs December 1999 an March 2004. The 3D fringes are inicating that the actual pattern of eformation along the coastline specially in the spit area (Fig. 4). It is interesting to fin that the coastal geomorphology patterns are expose to tremenous changes since 1999 to 2004. The rate change of spit is 2.4 m/yr with maximum elevation height of 2.4 m (Fig. 5). Fig. 2. Coherence image Fig. 3 shows interferograms of two pairs. It is obvious that there is a great eformations which are occurre in pairs of 1999 an 2004 ata (Fig. 3). The topographic phase of Fig.3 is moulate into eformation of the pair interferomateric phase. This is clearly obvious along the coastline (Fig. 3). This can be use to explain the changes have been occurre along the coastal geomorphology which can clearly notice in the spit area. Further, it can be Fig. 4. 3D Fringes of eformation phase Prouce from fuzzy B-splines. Clearly, Fuzzy B-splines metho has maintaine the fringe information an enoise the inteferogram. Further, the new technique provies fringe pattern with variety properties such as ense an non-ense fringes
62 IJCSNS International Journal of Computer Science an Network Security, VOL.9 No.5, May 2009 (Figs.4 an 5). In fact, fuzzy B-splines smooth the fringe pattern ue to the reuction of temporal ecorrelation which is cause by ynamical coastal seimentation an atmospheric conitions. Fig. 6. Coastal geomorphology reconstruction from fuzzy B-spline DInSAR Fig. 5: 3D spit rate change by using fuzzy B-spline algorithm. Table 1 shows a goo agreement between DInSAR s DEM an groun ata with r 2 of 0.86, p of 0.002 an rate of RMSE is ± 0.1 m. It is clear that rate of slope change is 1.5 m which consiers as steep slope. It might be san mining activities have inuce steep slope of spit (Fig. 6). Table 1: Significant Relationship between Groun Data an Fuzzy B-spline Interferomatery Statistical Parameters r 2 P RMSE Values 0.86 0.002 ± 0.1 m In aition, the increasing growth of spit across the estuary thus coul be ue to impact of littoral seimentation rift. Accoring to Mage [3,] the net littoral rift along Kuala Terengganu coastal water is towars the southwar which coul inuce growth of spit length. The high accuracy DInSAR s DEM coul be ue to fee of fuzzy B-spline into unwrappe phase. In fact, integration between Fuzzy B-spline an DInSAR metho has completely maintaine the graients on spit eges [1]. Furthermore, fuzzy B- spline increase the rate of unwrappe phase accuracy. Inee, fuzzy B-spline algorithm is able to keep track of uncertainty an provie tool for representing spatially clustere phase points [6]. 4 Conclusions This work has emonstrate a new technique for 3D reconstruction by implementing fuzzy B-spline within DInSAR technique. In oing so, historical pairs of RADARSAT-1 SAR fine moe ata were use. The results shows that fuzzy B-spline preserves etaile eges with iscernible fringes. Further, the new approach can prouce accurate 3D reconstruction from satellite raar ata such as RADARSAT-1 SAR fine moe. It can be conclue that that the integration between fuzzy B-spline an unwrappe phase inteferogram can prouce highly accurate 3D reconstruction of coastal geomorphology features within accuracy rate of ± 0.1 m References [1] Fuchs, H., Z.M., Keem, an S.P., Uselton, (1977). Optimal Surface Reconstruction from Planar Contours. In: Communications of the ACM, 20(10):693-702. [2] Luo, X., F.,Huang, an G., Liu, (2006). Extraction co-seismic Deformation of Bam earthquake with Differential SAR Interferometry. Journal of New Zealan Institute of Surveyors, 296:20-23. [3] Mage M., (2000). Wave spectra an shoreline change by remote sensing ata. Ph.D. Thesis, Universiti Putra Malaysia, Serang, Kuala Lumpur, Malaysia. [4] Mage, M., an H., Mazlan ( 2006). Three Dimensional Reconstruction of bathymetry Using C-Ban TOPSAR. Data. Photogrammetri-Fernerkunung Geoinformation. 6/2006, S. 469-480. [5] Massonet, D., an T. Rabaute, (1993). "Raar Interferometry: Limits an potential", IEEE Trans. Geosci. Remote Sensing, 3,455-464.
IJCSNS International Journal of Computer Science an Network Security, VOL.9 No.5, May 2009 63 [6] Russo, F., (1998). Recent avances in fuzzy techniques for image enhancement. IEEE Transactions on Instrumentation an Measurement. 47, pp. 1428-1434. [7] Stanely Consultants Inc., (1985). Malaysian national coastal ersoion stuy, Volume II. UPEN, Kuala Lumpur, Malaysia. [8] Tsay, J.R. an H.H. Chen, (2003) InSAR for DEM Determination in Taiwan by Using ERS Tanem Moe Data. Asian J. of Geoinformatics, 15:69-77. [9] Yang, J., T.,Xiong, an Y., Peng (2007). A fuzzy Approach to Filtering Interferometric SAR Data. Int. J. of Remote Sensing, 28:1375-1382. [10] Zebker, H.A., C.L.,Werner, P.A. Rosen, an S. Hensley, (1994) "Accuracy of Topographic Maps Derive from ERS-1 Interferometric Raar", IEEE Trans. Geosci. Remote Sensing, 2.823-836, Dr. Mage Marghany is a senior lecturer at the Faculty of Geoinformation Science & Engineering,, Natural Tropical Resources Information & Mapping Research Group, Universiti Teknologi Malaysia (UTM). He hols B.C. Physical oceanography egree from Alexanria University, M.Sc egree from University Putra Malaysia, Malaysia an a PhD in Environmental remote sensing from University Putra Malaysia, Malaysia. He aware ESA post-octoral fellowship at ITC, The Netherlans. His research interests lie in the areas of raar satellite applications to coastal stuies He has authore over 100 articles in both international an national referee journals, conference proceeings, an workshop in fiel of microwave application to coastal stuies such as ocean wave spectra, moeling shoreline change an oil spill trajectory movements. Dr Mazlan Hashim is currently a Professor of Remote Sensing at the Faculty of Geoinformation Science & Engineering, Universiti Teknologi Malaysia (UTM). He hols B.Surveying egree from UTM, Master Engineering egree from University of New Brunswick, Canaa an a PhD in environmental remote sensing from University of Stirling, Unite Kingom. His research interests lie in the areas of satellite remote sensing mission analysis, satellite igital image processing inclusive of calibrating an valiating satellite image proucts. He has authore/coauthore over 200 articles in both international an national referee journals, conference proceeings, workshop an monographs in fiel of remote sensing, igital image processing an relate technologies.