REVIEW OF THE STUDY ON THE UNDERWATER TOPOGRAPHY DETECTION WITH SAR IMAGERY IN SINO-EUROPEAN DRAGON COOPERATION PROGRAMME J. Yang (1), J.Zang (1), (2), J. Meng (1) (1) First Institute of Oceanograpy, State Oceanic Administration, Qingdao, Cina 266061; email: yangjg@fio.org.cn,mengjm@fio.rog.cn (2) Key Laboratory of Marine Science and Numerical Modeling, State Oceanic Administration, Qingdao, Cina 266061; email: zangjie@fio.org.cn ABSTRACT Sallow underwater topograpy can be observed by SAR and be detected wit SAR imagery. Wit te support of Sino-European Dragon Programme, te study on te underwater topograpy detection wit SAR imagery was done during 2004 to 2007. In past tree years, te underwater topograpy SAR detection was studied wit ERS SAR and ENVISAT ASAR image. Tese studies include te development of te model and algoritm of te underwater topograpy detection wit SAR image, te application study of underwater topograpy SAR detection and analysis of SAR imaging ability of underwater topograpy in Cina s coastal areas. Tese studies ave developed a new underwater topograpy detection model wit its SAR images acquired by different sensors in different time, and te underwater topograpy of Suangzi reefs in Nansa Islands and in Taiwan Soal were detected by SAR. Furtermore, te areas of Cina s coastal areas were te underwater topograpy can be detected wit SAR image were concluded. 1. INTRODUCTION Syntetic aperture radar (SAR) is one of te important microwave sensors, and it was widely used in te ocean monitoring and detecting. For SAR works at te microwave band and microwave cannot penetrate into sea water. So te underwater topograpy was observed by SAR indirectly. Tere are many complicated pysical processes during te underwater topograpy SAR imaging. It is well known tat te underwater topograpy SAR imaging mecanism includes tree main pysical processes. Firstly, te underwater topograpy modulates te tidal current wic flows over te underwater topograpy. Secondly, te tidal current modulates te sea surface micro-scale waves. Finally, te micro-scale waves in te sea surface modulate te backscattering of te microwave from SAR. From 1980 s, many studies about te underwater topograpy detection by SAR ave been carried out [1,2,3,4]. Tese studies focus on te underwater SAR imaging mecanism, te influences of te radar s parameters, te wind speed and direction, te magnitude and direction of tidal current on te underwater topograpy SAR detection and so on. Studies on te underwater topograpy SAR detection in Cina began in te late 1990 s, and were limited by te lack of te SAR data. Tese studies only covered te underwater topograpy SAR imaging mecanism and te SAR image simulation of te underwater topograpy. Tere is no application example of underwater topograpy SAR detection. Wit te support of Sino-European Dragon Programme, te underwater topograpy detection wit SAR image was studied wit many ERS SAR and ENVISAT ASAR images. Te model and algoritm of te underwater topograpy detection wit SAR image was developed and some application examples were given. Furtermore, te areas in Cina s coastal area were te underwater topograpy can be detected wit SAR image were concluded. Te details of tese studies will be sown in te following. 2. STUDY ON MODEL AND ALGORITHM OF UNDERWATER TOPOGRAPHY DETECTION WITH SAR IMAGE On te study of te model and algoritm of te underwater topograpy SAR detection, a new detection model was developed, wic syntesizes te SAR images of te underwater topograpy acquired by different sensors in different time. Tese SAR images are used simultaneously in te underwater topograpy detection. Furtermore, te damped Newton-row action metod was developed for te underwater topograpy SAR detection. 2.1 Model From te SAR imaging mecanism of te underwater topograpy, it is known tat tidal current is very important in te underwater topograpy SAR imaging. Tidal current is periodic and te magnitude and direction of tidal current in different time are different. So SAR images of te same underwater topograpy acquired in different time are different. Tis produces te detection result of te underwater topograpy wit one scene of SAR image is partial and cannot reflect te true case. Te existed model of te underwater topograpy detection wit SAR image is based on te single SAR image. Te detection results of te underwater topograpy wit SAR images acquired in different time are different, for te tidal current is different in te Proc. Dragon 1 Programme Final Results 2004 2007, Beijing, P.R. Cina 21 25 April 2008 (ESA SP-655, April 2008)
different SAR imaging time. Tis conclusion is also obtained in te correlation analysis of te underwater bottom topograpy and its SAR images [5]. Upon te foundation of te fact mentioned above, a new detection model of te underwater topograpy detection wit SAR images acquired by different sensors in different time was developed in te study of Dragon Programme [6]. Te following are te equations of tis model. k i= 1 ζ ( u) ( v) + + = 0 (1) u Cbu + u + v fv = g (2) v Cbv + u + v + fu = g (3) 2 ai ( αicos θi + αi cosθi sinθi 2 + βi cosθi sinθi + βi sin θi Gi ) = 0 (4) were, a i is te weigt of te i-t scene of SAR image. (u,v) is te eastern and nortern component of tidal current. ζ is tide eigt. is water dept. f is te Corolis parameter. C b is te bottom friction. (α i,β i ) is te transform coefficient of tidal current in te i-t scene of SAR imaging time relative to te reference tidal current. θ i is te wind direction in te i-t scene of SAR imaging time. G i is te backscattering coefficient of te i-t scene of SAR image. Te initial condition is: t = t = v, 0 t = t 0 = V0 ζ t=t = 0, 0 0 u U, 0 = H (5) were, (U 0, V 0 ) is te reference tidal current. H 0 is te initial water dept. Te open boundary condition is to input tide as: N ζ = H cos( w t g ) (6) i= 1 i were, H i, g i and ω i are te amplitude, pase lag and frequency of te constituents respectively. Te wall boundary condition is: = 0, = 0, ζ = 0 和 = 0 were, n is te normal direction of te wall boundary. i i (7) 2.2 Algoritm After te detection model of te underwater topograpy wit SAR image is establised, te problem of te underwater topograpy detection wit SAR image becomes an optimization problem. Tat is to solute te detection model under some initial and boundary conditions by some optimization metod. Some metods, suc as steepest descent metod [7] and conjugate gradient metod, ave been used in te solution of tis problem. But te convergence speed of tese metods is slow. A metod named damped Newton-row action metod [8] was used in te underwater topograpy detection wit SAR image [9]. Te damped Newton-row action metod as good convergence speed and numerical stability. 3. APPLICATION STUDY ON UNDERWATER TOPOGRAPHY DETECTION WITH SAR IMAGE Te underwater topograpy of Suangzi Reefs in Nansa Islands and tat of Taiwan Soal were detected wit ERS SAR and ENVISAT ASAR images. 3.1 Underwater topograpy detection of Suangzi Reefs in Nansa Islands Tree scenes of ENVISAT ASAR images are used in te underwater topograpy detection of Suangzi Reefs. Tey are respectively acquired in August 21 st 2004 (titled te first SAR image), in August 19 t 2004 (titled te second SAR image) and in August 18 t 2004 (titled te tird SAR image). Te study area of te underwater topograpy detection is 114 16 ~114 26 E, 11 21 ~11 30 N. Te detection model of te underwater topograpy wit te single SAR image is sown in te following. ζ ( u) ( v) + + = 0 (8) u Cbu + u + v fv = g (9) v Cbv + u + v + fu = g (10) 2 2 G( u, v, θ) = cos θ + cosθ sinθ + + sin θ (11) were, (u,v) is te eastern and nortern component of tidal current. ζ is tide eigt. is water dept. f is te Corolis parameter. C b is te bottom friction. θ is te wind direction in SAR imaging time. G is te backscattering coefficient of SAR image. Te initial and boundary condition are te same as te detection model wit SAR images acquired by different
sensors in different time. Fig.1 is te SAR images of Suangzi Reefs and te detection results of tese tree SAR images respectively. Te detection results of tree SAR image were compared wit te real water dept from te sea cart. Te comparison sowed tat te underwater topograpy of reefs in islands area can be detected by SAR. (a) SAR image of Suangzi Reefs 11.5 o 11.5 o 11.5 o 11.48 o 11.46 o 11.44 o 11.42 o 11.4 o 11.38 o 400 cm 350 300 250 200 150 100 50 11.48 o 11.46 o 11.44 o 11.42 o 11.4 o 11.38 o 400cm 350 300 250 200 150 100 50 11.48 o 11.46 o 11.44 o 11.42 o 11.4 o 11.38 o 400 cm 350 300 250 200 150 100 50 11.36 o 11.36 o 11.36 o 114.28 o 114.32 o 114.36 o 114.4 o 114.28 o 114.32 o 114.36 o 114.4 o 114.28 o 114.32 o 114.36 o 114.4 o (b) Detection Results Figure 1.ENVISAT ASAR images and teir detection results of underwater topograpy of Suangzi Reefs (Te left is te first SAR image. Te middle is te second. Te rigt is te tird. ) 3.2 Underwater topograpy detection of Taiwan Soal Taiwan Soal is a mesa wic is surrounded by steep slopes, and tere are undreds of sandbanks. Water dept of Taiwan Soal is about several decades meters. Ten scenes of ERS SAR and ENVISAT ASAR images of Taiwan Soal were collected to te detection of te underwater topograpy wit SAR image. Te acquiring time of tese SAR images is sown in Tab.1. Table 1. Acquiring time of SAR images of Taiwan Soal No. Sensor Data & time (UTC) 1 ERS-2 SAR 97-07-13 02:37 2 ERS-2 SAR 98-05-24 02:37 3 ERS-2 SAR 99-10-31 02:37 4 ERS-2 SAR 00-01-09 02:37 5 ERS-2 SAR 04-06-06 02:36 6 ENVISAT ASAR 06-09-24 02:07 7 ENVISAT ASAR 06-10-05 14:08 8 ENVISAT ASAR 06-10-13 02:10 9 ENVISAT ASAR 06-11-09 14:08 10 ENVISAT ASAR 06-11-17 02:10 Fig.2 is te scematic figure of te study area and te coverage of ten scenes of SAR images. In Fig.2, te sloped rectangles are te coverage of SAR images and te sadowed rectangle is te study area. Te study area locates in 118.55 ~118.75 E,23.0 ~23.3 N. 26 o 25.5 o 25 o 24.5 o 24 o 23.5 o 23 o 22.5 o 22 o 8,10 7,9 1-6 21.5 o 117 o 118 o 119 o 120 o 121 o 122 o Figure 2 te scematic figure of te study area and te SAR images coverage in Taiwan Soal In addition, water dept in tree profiles paralleling to longitude direction was in-situ measured in September 11 t 2005. Te location of tree profiles is sown in te Tab.2. Table 2. Te location of te tree profiles No. Profile 1 Profile 2 Profile 3 te location 118.68 E, 22.99 ~23.28 N 118.63 E, 22.99 ~23.21 N 118.60 E, 22.99 ~23.21 N
According to te analyses of tese SAR images, tere are seven scenes of SAR images can be used in te underwater topograpy detection of Taiwan Soal, and te oters cannot be used because tere is no information of te underwater topograpy in SAR images. It is caused by te differences of te tidal current in SAR imaging time. Te underwater topograpy of te study area in Taiwan Soal was detected in two cases. One is tat only te single SAR image was used in te underwater topograpy detection every time (te equations of te model is Eqs.8-11), and te oter is tat many SAR images acquired by different sensors in different time were used simultaneously in te underwater topograpy detection every time (te equations of te model is Eqs.1-4). m Figure 3. Te fourt SAR image (left) and its detection result (rigt) of study area in Taiwan Soal (a) Profile 1 (b) Profile 2 (c) Profile 3 Figure 4. Te comparison between te detection result wit te fourt SAR image and te in-situ measured water dept in tree profiles On te first case, te underwater topograpy of te study area was detected by te detection model wit te single SAR image. Under te given initial and boundary conditions, all te seven SAR images were used in te underwater topograpy detection respectively. It ave been proved te detection results of tese SAR image matc te in-situ measured water dept well by te comparison between tem in tree profiles. Tis sowed tat te underwater topograpy of Taiwan Soal can be detected by SAR. Te left in Fig.3 is te fourt SAR image of te study area, and te rigt is te gray map of its detection result. It is sown tat te sand waves are
very distinct in te rigt figure. Te comparisons between te detection result of te fourt SAR image and te in-situ measured water dept in tree profiles are sown in Fig.4. It can be seen tat te detection result fit te true topograpy well in Fig.4. Among te detection results of tese SAR images of Taiwan Soal, te average absolute error between te detection result and te true water dept is best to 2.23 m, and te average relative error is best to 7.50%. Oterwise, te dependence of underwater topograpy detection wit SAR image on te initial water dept was studied by virtue of te underwater topograpy detection of Taiwan Soal wit te single SAR image [10]. Te result sowed tat te detection result of te underwater topograpy is dependent on te initial water dept. On te second case, te underwater topograpy of te study area was detected by te detection model in different combinations of te tird, te fourt, te fift and te eigt SAR image. Under te given initial and boundary conditions, te underwater topograpy of study area was detected in five cases of te combinations of SAR images [7]. Te detection result of te combination of te fourt, te fift and te eigt was sown in Fig.5. (a) Profile 1 (b) Profile 2 Figure 5. Te detection result wit te combination of te fourt, te fift and te eigt SAR image. Te detection result wit te combination of te No.4, 5&8 SAR image was compared wit te in-situ measured water dept in tree profiles (sown in Fig.6). Te average absolute error is 2.09m and te average relative error is 7.07%, wic are smaller tan te errors of te detection wit te single SAR image. It is sown tat te underwater topograpy detection wit many SAR images acquired by different sensors in different time can improve te detection result. m (c) Profile 3 Figure 6. Te comparison between te detection result wit te combination of te No.4, 5& 8 SAR image and te in-situ measured water dept in tree profiles 4. IMAGING CAPABILITY STUDY OF UNDERWATER TOPOGRAPHY BY SAR IN CHINA S COASTAL AREA In offsore Cina, Boai Sea, Yellow Sea, East Cina Sea and Sout Cina Sea are distributed from te nort to te sout and te total area of sea is about 4,700,000 square kilometers. Over 5400 islands distribute in Cinese sea area. Te lengt of main land coastline is about 18,000 kilometers. Tere are many coastal sea areas or areas of islands were water dept canges from several meters to several decade meters. Some underwater bottom topograpy of tese areas is uncangeable, but oters cange wit te action of strong ocean current, typoon and storm tide. In Cina s coastal area, tere are many areas were te underwater topograpy can be detected by SAR. So it is essential to analyze te SAR imaging capability of te underwater topograpy in Cina s coastal area for te detection in te future. Te areas in Cina s coastal area were te underwater topograpy can be detected by SAR was concluded by analyzing te caracters of te tidal current field, te underwater topograpy and SAR images in Cina s coastal area. Tere are four areas
were te underwater topograpy can be detected by SAR (in Fig.7). Tey are Boai Sea, te area around Subei Soal, te area around Taiwan Soal and te area around Nansa Islands [11]. 50 o 40 o 30 o 20 o 10 o A: Boai Sea B: Te area around Subei Soal C: Te area around Taiwan Soal D: Te area around Nansa Islands 80 o 90 o 100 o 110 o 120 o 130 o Figure 7. Te scematic figure of te areas in Cina s coastal area were te underwater topograpy can be detected by SAR 5. CONCLUSION From te study on te underwater topograpy detection wit SAR image in Sino-European Dragon Programme, te following is concluded: (1) SAR images of te same underwater topograpy in different imaging time are different as te tidal current is different in different SAR imaging time. (2) A new detection model of te underwater topograpy wit SAR images acquired by different sensors in different time was developed, and te algoritm was given. (3) Te underwater topograpy detection wit SAR image is testified by te underwater topograpy detection of Suangzi Reefs and Taiwan Soal wit ERS SAR and ENVISAT ASAR images. (4) Te areas in Cina s coastal area were te underwater topograpy can be detected by SAR were concluded. ACKNOWLEDGEMENT Tis study was supported by te Dragon Project (ID 2566), te National Natural Science Foundation of Cina (Grant No.60672159). Te autor is grateful to European Space Agency for providing SAR data. REFERENCE 1. Alpers W. & Hennings I. (1984). A Teory of te Imaging Mecanism of Underwater Bottom Topograpy by Real and Syntetic Aperture Radar. J. Geopys. Res. 89, 10,529-10,546. 2. Cooper, A. L., Cubb, S. R., Askari, F. & Valenzuela, G. R. (1994). Radar Surface Signatures for te D C A B Two-Dimensional Tidal Circulation Over Pelps Bank, Nantucket Soals: A Comparison between Teory and Experiment. J. Geopys. Res. 99, 7,865-7,883. 3. Holliday D., St-Cyr G. & Woods N. E. (1986). A Radar Ocean Imaging Model for Small to Moderate Incidence Angles. International Journal of Remote Sensing. 7, 1,809-1,834. 4. Vogelzang J. (1997). Mapping Submarine Sand Waves wit Multiband Imaging Radar. 2. Experimental Results and Model Comparison. J. Geopys. Res. 102(1), 1,183-1,192. 5. Yang, J.G., Zang, J. & Meng, J.M. (2006a). Correlation Analysis of Underwater Bottom Topograpy and Its SAR Images. Proceedings of SEASAR 2006: Advances in SAR Oceanograpy from Envisat and ERS Missions. ESA SP-613. 6. Yang, J.G. (2007). Assimilation Model of Underwater Topograpy Inversion wit SAR Data Acquired by Different SAR Sensors in Different Time (P.D dissertation). Institute of Oceanology of Cinese academy of sciences. Qingdao, Cina. 7. Jin, M.B. & Yuan, Y.L. (1997). Formulation and Solution of te Matematical and Pysical Inverse Problem of Detection Sea Topograpy by SAR Image (in Cinese). Oceanologia et limnologia Sinica. 28(Sup.), 27-31. 8. Cui, J.H., Cen, G.Q. & Zang, J. (2003). Damped Newton-row Action Metod for SAR image simulation and underwater topograpy inversion (in Cinese). Advances in Marine Science. 21(1), 68-77. 9. Yang, J.G., Zang, J. & Meng, J.M. (2006b). Underwater Bottom Topograpy Detection of Suangzi Reefs wit ENVISAT ASAR Images Acquired In Different Time. Proceedings of te 2005 Dragon Symposium-Dragon Programme Mid-Term Results, ESA SP-611, 133-139. 10. Yang, J.G., Zang, J. & Meng, J.M. (2006c). Dependence of Underwater Bottom Topograpy Detection wit SAR Image on Initial Water Dept. Te 3rd International Symposium on Future Intelligent Eart Observing Satellites. Beijing, Cina. 11. Huang, J.B., Zang, J. & Yang, J.G. et al. (2006). Imaging Ability Researc of Sallow Underwater Bottom Topograpy by SAR in Cinese Coastal Sea. Proceedings of te 2005 Dragon Symposium-Dragon Programme Mid-Term Results, ESA SP-611, 123-131.