SONAR MEASUREMENTS IN SHIP WAKES SYNCHRONOUS WITH TERRASAR-X OVERPASSES

Transcription

1 SONAR MEASUREMENTS IN SHIP WAKES SYNCHRONOUS WITH TERRASAR-X OVERPASSES ABSTRACT Alexander Soloviev (1), Mikhail Gilman (1), Kathryn Young (1), Stephan Brusch (2), and Susanne Lehner (2) (1) Nova Southeastern University Oceanographic Center, Dania Beach, Florida 33004, USA, (2) German Aerospace Center, Remote Sensing Technology Institute, D Wessling, Germany, A pilot experiment was conducted by NSUOC and DLR in April-June 2008 in the Straits of Florida near Port Everglades, Florida in order to study the dynamics of far wakes of ships. In this experiment, a small boat with downward-looking sonar made snake-like sections through wakes of ships of opportunity during the TerraSAR-X overpasses. The ship identity and its parameters (length, speed, course, etc.) were identified utilizing AIS (Automated Identification Systems). The sonar responded to the clouds of micro-bubbles generated in the ship wake by the propulsion system and ship-hull turbulence. Different sonar frequencies (83 khz, 200 khz, and 455 khz) corresponded to different bubble sizes and bubble-cloud turbulent inhomogeneities. The ship wake was traced in the sonar signal typically from 10 to 30 min after the ship passage. Preliminary analysis of the measurements suggests that the visibility of the wake in SAR is correlated with the presence of micro-bubbles in the wake. A possible explanation is as follows: the micro-bubbles scavenge surface-active materials in the upper layer of the ocean and bring them to the surface. These materials are then concentrated in narrow zones on the sea surface by convergent motions in the ship wake. The viscoelasticity effect of surface films leads to the suppression of gravity-capillary waves, which contrasts the ship wake on SAR images. Influence of wind-wave field on the ship wake as well as the effect of screening of the wind-wave field by the ship hull add another level of complexity to wake patterns observed in SAR. 1. INTRODUCTION The problem of remote detection of ships and ship wakes holds a great value for such important topics as fishing and pollution monitoring, global security, and navigation safety. Of the available sensing techniques, satellite-based Synthetic Aperture Radar (SAR) proves invaluable due to its relative independence of weather conditions and time of day, large size of covered area, and high spatial resolution. Ships and ship wakes were discovered on the first images of the NASA SeaSat satellite launched in 1978 [1]. With the recent advances of satellite-based SAR technology, interest to remote sensing of ship wakes is increasing. The rapid growth of the number of available SAR platforms could make near-real-time detection of ships from space a reality in a few years. On the SAR image, a ship is usually represented by a bright spot formed by the reflection of the radar signal from the ship construction. Under certain conditions, a moving ship can generate a wake, which may extend for tens of km. For detection purposes, the ship wake creates additional opportunities because of its size and shape and additional parameters that might be extracted from the image, including direction, speed, and probably some other parameters of the ship. The structure of ship wakes and their SAR images is rather complex. Traditional classification specifies ship speckle, near wake, Kelvin wake, centerline wake, narrow-v wake, and other parts of the wake [2, 3]. Often, one or more features of the wake could not be found on satellite SAR images [2 4]. The centerline part of the wake is often the most prominent feature of the wake image in SAR. Usually, the centerline wake appears as a dark scar behind the ship. The dark appearance of the scar is associated with the reduction of surface roughness due to suppression of Bragg-scattering short surface waves by surfactants, wave-current interaction, and turbulence in the wake [5]. The hydrodynamic theory of far wakes of ships, as of today, is comprehended only on a qualitative level. Despite some prominent in-situ experiments targeting remote sensing of ship wakes (see, e.g., [9]), the relation between underwater, surface, and electromagnetic processes of SAR imaging of ship wakes is still not completely clear. Acoustic techniques are an effective tool to study ship wakes, which has however traditionally focused mainly on naval applications (see, e.g., [6 8]).

2 In order to improve our understanding of the visibility of ship wakes on SAR images, a pilot experiment was conducted in April-June 2008 by the Nova Southeastern University Oceanographic Center (NSUOC) and the German Aerospace Center (DLR). The experiment included several locations in the Straits of Florida, Gibraltar Straits, Hawaii, and Australia (Tab. 1). In this paper we focus on the data collected in the Straits of Florida where in situ measurements have been conducted. The data obtained in the Strait of Gibraltar and the Hawaiian and Australian locations are discussed in the accompanying paper [10]. Date 2008 Time of Overpass UTC Table 1. Status of the data collected during the NSUOC-DLR pilot experiment Location Terra SAR-X polariza tion A I S S O N A R W E A T H E R V I D E O Field Notes WAMOS X-band radar ALOHA Station 04/20 23:21 Florida HH /12 23:21 Florida HH /17 11:17 Florida /23 23:22 Florida VV /29 23:13 Florida VV /09 23:13 Florida HH L 06/14 11:17 Florida L 06/14 23:23 Florida HH L 06/20 23:13 Florida L 06/25 11:17 Florida L 06/25 23:21 Florida HH L 06/30 11:25 Florida /14 06:30 Gibraltar VV + L 06/21 18:14 Gibraltar VV + L 06/25 06:30 Gibraltar HH + L 07/02 18:14 Gibraltar HH + L 06/18 16:28 Hawaii HH - L L 06/19 04:31 Hawaii HH - L L 06/23 16:37 Hawaii HH - L L 06/24 04:40 Hawaii HH - L L 06/24 16:19 Hawaii VV - L L 04/19 19:45 Australia HH - L L 05/11 19:45 Australia HH - L L 06/19 09:00 Australia HH + L L 06/24 09:08 Australia HH - L L + available, - - unavailable, L looking for data HF Radar WERA In this paper we present the results of collection of sea-truth data (sonar echo-intensity, weather, and ship operation parameters) simultaneous with SAR imaging. The use of profiling sonar can give information about the turbulence and bubble clouds in the water column. In this work, we have been pursuing a hypothesis that the visibility of the ship wake in SAR is correlated with the presence of micro-bubbles in the wake. It is generally accepted that the bubbles injected into the wake by the ship passage and screws scavenge surfactants dissolved in the sea water and bring them to the surface [3, 9]. Thus, we can study an important part of the mechanism of the creation of a slick that makes the wakes visible on SAR images. Additional factors, which appear on high resolution SAR imagery, are wind-wave effects and screening of the wind-wave field by the ship hull. The materials are organized as follows: Section 2 describes the instrumentation and measurements made during the NSUOC-DLR pilot experiment in the Straits of Florida. Section 3 provides the main results of this experiment. Section 4 summarizes the results of this paper.

3 2. INSTRUMENTATION We have been using sonar producing two downward-looking beams of 83 khz and 200 khz frequency, and two separate side-looking beams, 455 khz each. The sonar has a GPS subsystem and puts position- and time-stamps for each acoustic ping. The interval between pings was no more than 1 sec. for each channel. The sonar was operated from a 20-ft NSUOC boat Lucy Forman. Figure 1. Schematic diagram of the multi-frequency sonar mounted on a small boat, and the geometry of the beams. Images from the site We have been exploring wakes of relatively large (upwards of 100 m length) commercial ships of opportunity in the vicinity of Port Everglades in South Florida. The ship position and speed was retrieved from the Automated Identification Systems (AIS) data using a commercial service. In the area of the experiment, the AIS coverage extends to some 15 miles off the Port entrance. The idea of the experiment was to cross the ship wake several times on a snake-like trajectory at approximately the time the TerraSAR-X overpass. With transducer in the water, the boat moved at 4 5 kt speed, which, with the average wake width m and lifetime min, allowed us to make from 4 to 10 full crossings of the wake before the wake dissipated. In operation, the transducer was placed approximately 1.1 m below the water surface. The boat was run with constant motor power, which however resulted in different true speed when moving in different directions with respect to wind and currents. An example of the sonar data is shown in Fig. 2. While SAR images give a momentary snapshot of the wake, the procedure described above results in several crossing profiles made practically at the same place but for different wake age, or, in other words, at different distances from the wake-producing ship (illustrated by Fig. 4). Recalculating wake age into distance from the ship using the ship speed, and making also a correction for the drift that shifts the wake but does not affect GPS readings, we can attribute each of the wake crossings to a certain part of the wake on the image. The moment of wake creation was defined as the time when the ship s stern was passing near the boat location. Comparison of wake signatures in different sonar frequencies is shown in Fig. 3. According to some authors (see, e.g., [11]), sonar echo-intensity could be used to estimate the dissipation rate of turbulent kinetic energy due to the mechanism of acoustic Bragg scattering in the scale of the wavelength. We have found that of the three channels used

4 (83 khz, 200 khz, and 455 khz), the signature in 200 khz is the most prominent and long-living; this channel is used through the rest of the paper. Absolute calibration of the sonar used, as well the analysis of sonar returns from ship wake in multiple frequencies, are still pending. Figure 2. A 3D illustration of a part of the boat track and sonar profile (beam 200kHz) of the wake of Kopalnia Borynia, May 29, 2009, time about 23:30 UTC (minutes and seconds in different points of the track are shown near the markers). For the figures showing sonar contour plots below, the acoustic cross-section is given in decibels. The signal from the distances less than 0.8 meters from transducer has been ignored, which means that the sonar data start from a 2 m depth. 3. MEASUREMENTS During the experiments, the main challenge was to find the ship of opportunity that during TerraSAR-X acquisition would be on the image footprint, within the AIS area, and within the reach of the sonar-equipped boat. In some cases, adverse weather conditions or boat operation logistics prevented us from making sonar measurements during the TerraSAR-X satellite overpasses. In a few cases, the TerraSAR-X images could not be taken during sonar measurements because of scheduling conflicts. In three cases (from the total of 12 attempts during this pilot experiment see Tab. 1), the in situ measurements in the wake of a ship of opportunity and the TerraSAR-X images overlapped in time and space. 1) May 23: decelerating and turning ship; TSX polarization is VV. 2) June 9: steadily moving ship; TSX polarization HH. 3) June 25: drifting ship; TSX polarization HH. In this section, we analyze these three cases. Pertinent information to these cases (as well as to the May 29 example shown in Figs. 2 and 3) is given in Tab. 2. Wind information is obtained by averaging data from the NDBC/NOAA Fowley Rocks and Lake Worth stations.

5 Figure 3. Sonar return in different frequencies for the second of the wake crossings shown in Fig 2. Table 2. Pertinent information to the NSUOC-DLR experiment in the Straits of Florida Date (year 2008) May 23 May 23 May 29 June 09 June 25 June 25 TSX image time 23:22:02 23:22:02 23:13:33 23:22:03 23:22:03 TSX mode Stripmap Stripmap Stripmap Stripmap Stripmap TSX polarization VV VV HH HH HH Ship name Alianca Inca Xin Su Kopalnia Eurus Paris MSC Emma Caribe Legend Zhou Borynia Ship length, m Ship speed, kt Decelerating to Ship heading, deg. Turning to ~ Wind speed, m/s Wind Direction, deg Steadily Moving Ship (TSX Polarization HH) Figs. 4 and 5 (left panel) show the wake of Eurus Paris on the TerraSAR-X image taken on June 9, Four wake crossings with sonar were made before the wake dissipated (Fig. 4). Fig. 6 shows the sonar contour plots in more detail. The wake is clearly seen on the SAR image. On the sonar images, we can see that as the wake ages, its acoustic signature widens and splits into several clusters. The residual bursts of the sonar return on the last crossing may be associated with the faint railroad track feature [9] on the SAR image (Fig. 5, left panel). Similar patterns could also be seen in the far wakes of other ship on the same SAR image (Fig. 5, right panel). These elusive features are usually found on SAR images under low wind speed conditions. A prominent feature of both ship wakes shown in Fig. 5 is the wake asymmetry with respect to the wind direction: a bright line could be seen on the upwind side of the centerline wake (also discussed in Section 3.2). No distinct wind asymmetry has been found in sonar imagery though. Apparently, the processes that are ultimately responsible for the local variations of surface roughness, such as wave breaking on the upwind side of the wake and dampening due to

6 films and turbulence, are concentrated in the top layer of water and could not be studied directly in this pilot experiment. Sonar observations of ship wakes made in [6, 12] demonstrate that at some distance astern the ship, the acoustic wake splits into two parts. In our experiments however, we cannot isolate two distinct parts clearly. The difference may be due to many factors, including difference in weather conditions and the type and size of ships in this and cited papers. Figure 4. (left) SAR image of the wake of a moving ship, Eurus Paris ( June 9, 2008). The purple-colored rectangles show the wake areas with wake age correspond to the four crossings of the wake, which are shown on the right side. The red spot near the ship speckle is the ship position based on the AIS data. (right) Plots of sonar echo-intensity (for details see Fig. 6); the panels for East-to-West crossings (second and fourth from the top) are flipped horizontally to achieve the same geographical orientation of all four panels.

7 Figure 5. Asymmetry of centerline ship wake: clips (no georeferencing) from the TerraSAR-X image on June 9, 2008: (left) a close-up of the wake of Eurus Paris; (right) wake of another ship in the same area (ship identity and parameters are unknown). The problem of visibility of different arms of the Kelvin wakes is discussed in detail in [4].

8 Figure 6. Four crossings of the wake of a moving ship on June 9, An irregularity around second 130 is due to the sharp turn of the boat. The bursts at the edges of the wake signatures at the last crossing may be associated with the railroad track feature of the SAR image (see Fig. 5, left panel). 3.2 Decelerating and Turning Ship (TSX Polarization VV) Just before the imaging time on May 23, 2008, the ship Alianca Inca was moving in SSW direction with reducing speed, then it made sharp U-turn to the left and finally stopped. At the time of the TSX image (Figs. 7 and 8), the ship has almost completed the turn and heads NE.

9 Figure 7. (left) The whole SAR image (no georeferencing) of Miami area of 23 May (right) a clip with the decelerating ship, Alianca Inca (the right of the two bright speckles, see also the top of the image on the left). Though the wake of this ship is very faint on the TerraSAR-X image on Fig. 8 (left panel), the acoustic feature of the wake is prominent (Figs. 9 and 10). The main acoustic features of this wake are essentially the same as for the case of a steadily moving ship analyzed in Section 3.1. The initial tilt of the wake signature on the sonar image is supposedly due to ship turning; it can barely be seen after the first three crossings of the wake (Fig. 9). Figure 8. (left) A close-up (no georeferencing) with the wake of a decelerating ship (see Fig. 4 (right), northern part). A colored rectangle show approximate area of sonar cross-sections shown in Figs. 9 and 10. (right) A close-up (no georeferencing) showing the wake of Xin Su Zhou from the southern part of the same image

10 The low contrast of the wake on the SAR image (Fig. 8, left panel) can in part be explained by the VV polarization mode. There are different opinions about the polarization most favorable for detection of ship wakes (see, e.g., [12 15]), which were mostly made for previous generation of SAR satellites with m resolution and need yet to be reviewed for higher-resolution imagery. Our experience thus far (which we cannot confirm quantitatively yet) is that HH polarization is preferable. One hint for the possible explanation could be found in [16] where wave breaking is considered to be an important mechanism responsible for radar contrast of hydrodynamic features. If, according to that theory, the input from wave-breaking is similar for both polarizations while the background scattering is much lower for HH-polarization, then the contrast due to wave-breaking zones for HH polarization should be higher. Figure 9. First crossings of the wake on 23 May The boundaries of the wake during the first intersection are tilted, supposedly due to the ship turning.

11 Figure 10. Last two crossings of the wake on May 23, 2008 However, the wake of a steadily moving ship in the lower part of the big image in Fig. 7 can be seen better (Fig. 8, right panel). We may argue that with a relatively low wind speed, proximity to the shore, and wind direction from the shore, the area around the decelerating ship is very calm, so the contrast of the slick due to ship passage will be low. The distance from the shore to the wake in the right panel of Fig. 8 is almost twice that of the left panel, so the background there is not so dark, the slick is more pronounced, and the wave-breaking responsible for the bright upwind side of the wake is more intense. 3.3 Drifting Ship (TSX Polarization HH) A wake of a drifting ship is seen on the TerraSAR-X image of June 25, 2008 (Fig. 11). The source of the wake, MSC Emma, is a large container ship with almost 300 m length, so the westward direction of the wake signature on the SAR image could be explained by the shadowing of wind and waves by the ship hull, which is consistent with easterly wind and southward ship heading during this scene. A wake of a much smaller Caribe Legend (Fig. 11, lower right panel) moving north has apparently both hydrodynamic and shadow features. The boat with sonar crossed the wind-wave shadow-type wake west of the ship (Fig. 11, left panel) and passed near the ship stern while the ship was slowly drifting south. There is practically no acoustic signature of the wake in either case (Fig. 12).

12 Figure 11. (top) The SAR image of a drifting ship, MSC Emma; (bottom left) a close-up (no georeferencing) showing a wake of MSC Emma due to wind and wave screening; (bottom right) a close-up (no georeferencing) of Caribe Legend located approximately 15 miles south from MSC Emma: the ship is heading North, both its centerline wake (behind the ship) and shadow wake (skewed astern due to the ship motion) can be seen.

13 Figure 12. (top panel) A sonar profile downwind of a drifting ship MSC Emma (June 25, 2008); (bottom panel) sonar profile near the stern of MSC Emma. 4. CONCLUSIONS The factors that control visibility of ship wakes in SAR and sonar images are related to hydrodynamics, air-bubble dynamics, surface wave dynamics, scattering of electromagnetic waves, and others. The availability of high-resolution SAR imagery from TerraSAR-X and a novel approach to in situ measurements in ship wakes using profiling sonar have provided new insight into the problem of remote sensing and hydrodynamics of ship wakes. The NSUOC-DLR pilot experiment in the Straits of Florida in Summer 2008 has resulted in several interesting observations, which may contribute to the development of a comprehensive model of far wakes of ships. We present below a few generalizations that could be made from the analysis of the above cases, as well as other cases where we had separate SAR and acoustic images of ship wakes. 1) The length of centerline wake behind the ship on SAR images appears to correlate with the presence of its hydro-acoustic signature. 2) The horizontal patterns of the hydro-acoustic and SAR signatures of ship wakes, however, may not exactly coincide. In particular, the asymmetry of ship wakes with respect to wind direction previously observed on photo images [17] is also noticeable on high resolution SAR imagery, but not in acoustic profiles of the centerline wake area for steadily moving ships. 3) Even in the extreme case of drifting ship, we may see a wind-wave shadow type wake, which is probably a result of screening of the wind-wave field by the ship hull. Although this wake has no underwater part, it appears that in some cases, such wake may be fairly pronounced on SAR images. Collection of larger statistics from satellite observations and in situ measurements is required for better understanding of ship wake hydrodynamics and further improvement of radar imaging algorithms. These steps are necessary for the development of a new generation of remote sensing algorithms for ship and ship wake detection. 5. ACKNOWLEDGEMENTS We gratefully acknowledge the support of our field work by Drs. Edward Keith, Greg Foster, and Kristi Foster, Capt. Lance Robinson, Capt. Brian Ettinger, Elizabeth Goergen, Jenna Lueg, Allison Brownlee (all NSUOC). Sonar measurements have been done as a part of the NSUOC project, Hydrodynamics and remote sensing of far wakes of ships. The high-resolution TerraSAR-X images have been provided under the OCE0143 EXTROP-X proposal (Investigator Dr. Susanne Lehner, DLR). We would like to thank Tom Vickers and Dave Betts (Humminbird Marine Electronics), and Robert Gecy for their help in processing sonar data. We thank Dr. Karen Fisher (LANL) for a fruitful discussion of the results.

14 6. REFERENCES 1. Fu, L. L., Holt, B. (1982). Seasat views oceans and sea ice with synthetic-aperture radar. JPL Publication NASA Jet Propulsion Laboratory, Pasadena, California. 2. Eldhuset, K. (1996): An automatic ship and ship wake detection system for spaceborne SAR images in coastal regions, IEEE Transactions on Geoscience and Remote Sensing. 34(4), Reed, A. M., Milgram and J. H. (2002): Ship wakes and their radar images. Annual Review of Fluid Mechanics. 34, Hennings, I., Romeiser, R., Alpers, W., Viola, A. (1999): Radar imaging of Kelvin arms of ship wakes. International Journal of Remote Sensing. 20(13), Milgram, J. H., Skop, R. A., Peltzer, R. D., Griffin, O. M. (1993): Modeling short sea wave energy distributions in the far wakes of ships. Journal of Geophysical Research. 98(C4), Trevorrow, M.V. (2005): High-Frequency Acoustic Scattering and Absorption Effects with Ship Wakes. Defence R&D Canada - Electronic Document (PDF) Stanic, S., Kennedy, E., Brown, B., Medley, D., Goodman, R., Caruthers, J. (2007): Broadband acoustic transmission measurements in surface ship wakes. Oceans. Sept. 29, 2007 Oct. 4, 2007, Bergmann, P.L., Spitzer, L. (1946). Physics of Sound in the Sea. Peninsula Publishing, Los Altos, US. 9. Peltzer, R. D., Griffin, O. M., Barger, W. R., Kaiser, J. A. C. (1992): High-resolution measurement of surface-active film redistribution in ship wakes. Journal of Geophysical Research. 97(C4), Lehner, S. (2008): Fine features of ship wakes in the problem of ship detection. TerraSAR-X Science Team Meeting, 25/26 November DLR Oberpfaffenhofen, Germany this issue. 11. Shen, C., Lemmin, U. (1998): A new method for probing the turbulence scalar spectrum by ultrasonic sonar scanning. Journal Experiments in Fluids. 24(1), Weber, T. C., Lyons, A. P., Bradley, D. L. (2005): An estimate of the gas transfer rate from oceanic bubbles derived from multibeam sonar observations of a ship wake. Journal of Geophysical Research. 110(C04005), doi: /2004jc Benilov, A., Bang, G., Safray, A., Tkachenko, I. (2001): Ship wake detectability in the ocean turbulent environment, Twenty-Third Symposium on Naval Hydrodynamics. France. 13. Arnesen, T.N., Olsen, R.B., Vachon, P.W. (2004): Polarisation-dependent signatures of ships in ENVISAT AP mode data. Envisat and ERS Symposium. Salzburg, Austria, 6 10 September Yang, J., Huang, W., Xiao, Q., Fu, B., Yao, L. (2004): Optimal polarization for the observation of ocean features with SAR. Geoscience and Remote Sensing Symposium. Anchorage, Alaska. 15. Greidanus, H., Kourti, N. (2006), Findings of the DECLIMS project Detection and classification of marine traffic from space. SEASAR 2006: Advances in SAR Oceanography from ENVISAT and ERS Missions. European Space Agency, Roma, Italy. 16. Kudryavtsev, V., Akimov, D., Johannessen, J., Chapron, B. (2005): On radar imaging of current features: 1. Model and comparison with observations. Journal of Geophysical Research. 110(C7), C Soloviev, A., Gilman, M., Moore, K., Young, K., Graber, H. (2008): Remote Sensing of Ship Wakes. SEASAR 2008: Advances in SAR Oceanography from ENVISAT and ERS missions. European Space Agency, Roma, Italy.