MULTIPATH EFFECT ON DPCA MICRONAVIGATION OF A SYNTHETIC APERTURE SONAR

Transcription

1 MULTIPATH EFFECT ON DPCA MICRONAVIGATION OF A SYNTHETIC APERTURE SONAR L. WANG, G. DAVIES, A. BELLETTINI AND M. PINTO SACLANT Undersea Research Centre, Viale San Bartolomeo 400, La Spezia, Italy wang@saclantc.nato.int The Displaced Phase Centre Antenna (DPCA) technique makes use of the correlation of sea bottom back scattering to determine the trajectory of a Synthetic Aperture Sonar (SAS). Multipath has been identified as the main environmental factor degrading the accuracy of DPCA in shallow water environments. An experiment has been carried out with Saclantcen s 100 khz multi-aspect SAS demonstrator deployed from R/V Alliance. The 256 channel receive array was placed, both horizontally and vertically, at various depths in shallow water channels, in order to quantify the relative levels of the multipath signals as a function of depth, range, sea state and bottom type. The ping to ping signal correlation was measured and compared with the results from a sonar performance model. 1 Introduction The performance of DPCA micronavigation for SAS applications has been investigated extensively both in theory and experiment in recent years [1 3]. The technique makes use of the correlation of the sea bottom back scattering to estimate the displacement of the sonar between successive pings. The main environmental factor determining the navigation accuracy achievable for a given sonar system is ultimately the signal to noise ratio. The signal is the direct sea bottom back scattering, while the noise consists of background noise of the sea, system noise, surface and volume reverberation and multipath interference such as surface reflected bottom scattering etc. When SAS is used in shallow and very shallow waters, the multipath interference may be a dominant source of noise. Thus the accuracy of the DPCA micronavigation is degraded, possibly limiting the achievable SAS length. In addition, the multipath will also degrade the quality of the SAS imagery, leading to ghost targets and loss in image contrast (e.g. filling in of shadows). Thus multipath could be the most important environmental factor limiting the range achievable by an SAS system in shallow water. An experiment has been carried out to investigate the effects of multipath on SAS in shallow water channels with different water depths and bottom characteristics. A 100 khz sonar with a total receiver aperture of 1.92 m was deployed from R/V Alliance in horizontal and vertical configurations. The data from vertical transmissions provide very revealing information about the shallow water channel because of the high angular resolution of the sonar. The signal level and ping to ping correlation were predicted with a new sonar performance model, ESPRESSO, and compared with the experimental results. 465 N.G. Pace and F.B. Jensen (eds.), Impact of Littoral Environmental Variability on Acoustic Predictions and Sonar Performance, Kluwer Academic Publishers. Printed in the Netherlands.

2 466 L.WANG ET AL. 2 Experiment The experiment was carried out in shallow water near La Spezia (Italy) in two areas, off Tino island (referred to as area I in the following) and Monesteroli (area II), from the middle of October to early November, It was one part of a larger experiment designed to further the understanding of SAS micronavigation, combining data-driven techniques such as DPCA and Aided Inertial Navigation Systems. The main piece of equipment used is the MASAI (Multi Aspect Synthetic Aperture sonar Imaging) towbody, which consists of a 100 khz multi-element sonar array and a high grade inertial navigation system (INS). Figure 1 shows the deployment of the MASAI towbody from R/V Alliance. 2.1 MASAI System The MASAI array consists of 256 receiver elements at a 7.5 mm spacing to form an aperture of 1.92 m. The signal used for MASAI 01 was a 10 ms chirp with 100 khz centre frequency and 10 khz bandwidth. The centre 64 channels were used for transmission and defocused to give a beam width of 40 o. The vertical (resp. horizontal) beamwidth of the elements was about 40 o (resp. 100 o ). The source level was 206 db re 1µPa at one meter. The depression angle of the sonar was 22 o in the horizontal configuration. Figure 1. Deployment of the MASAI system from the Alliance.

3 MULTIPATH EFFECT ON SAS NAVIGATION 467 The INS used in the MASAI system is a strapdown inertial navigation system with ring laser gyros. The INS was aided by a 1.2 MHz DVL (Workhorse Navigator by RDI), a pressure sensor and a GPS. The aiding was peformed with NAVLAB, a software developed by the Norwegian Defense Research Establishment. 2.2 Environmental Data The shallow water channels in the location of the trial were flat with a water depth about 26 m in area I and 32 m in area II. samples were taken in the areas in order to determine the bottom characteristics. Two methods were used to obtain the sound speed and density in the sediments. One was to derive the parameters from grain size of the sediments and, the other one was to measure directly from the core samples. Table 1 shows the results of grain size analysis for the cores of the top 6 cm in the two areas. It was observed that the sediment in area I was harder than in area II. Sound speed and density in the sediments measured directly from the core samples are given in Table 2. It can be seen that the acoustic impedance of the sediment in area I was much higher than that in area II. Wind speed was measured by a meteobuoy during the trial. The measured wind speeds at the time of sonar transmissions (most of the time corresponding to sea state in the range 2 3) were used in the sonar model for surface scattering strength and surface reflection loss. A wave rider was also deployed to measure the wave height. A CTD was used throughout the trial to obtain the sound speed profile in the water column in both area I and area II. An isovelocity profile of about 1526 m/s was found for the whole period. Table 1. analysis Tino (area I) Monasteroli (area II) Depth (cm) Sand/silt/clay φ Sand/silt/clay φ /46.7/ /40.2/ /33.3/ /42.8/ /49.5/ /39.5/ Table 2. Measured sound speed and density Tino Monasteroli Depth (cm) Sound speed (m/s) Density (g/cm 3 ) Sound speed (m/s) Density (g/cm 3 )

4 468 L.WANG ET AL. 3 Sonar model The modelled data presented in this paper were obtained using a tool being developed at SACLANTCEN called ESPRESSO. The tool includes a reverberation model based on the technique of "Geometric beam tracing" [4,5], originally developed to model propagation loss. The adaptation of this technique to reverberation modelling is described in [6]. The reverberation model approximates the seabed, sea surface, and water column as lines of scatterers; the reverberation contribution from each scatterer is allocated to a time bin to create a reverberation time series. The acoustic models used to obtain scattering strengths and reflection losses are described in [7] (the bistatic version of the seabed scattering strength model was used). The absorption coefficient in water was obtained using the algorithm described in [8]. 4 Experimental and modelling results Although a large amount of data was collected with the sonar in a horizontal configuration at fixed positions in both areas during the trial, the vertical configuration was used only in area II. The results presented here are all from the data obtained in area II since the data from the vertical configuration provided more detailed information about the channel. Figure 2. Beamformed sonar data with the array in vertical configuration, showing the arrival angles of the different paths measured with respect to the vertical.

5 MULTIPATH EFFECT ON SAS NAVIGATION Sound Field in the Vertical Plane In order to study the details of the sound field in the channels, the MASAI sonar was deployed vertically in area II. The various arrivals can be examined by beamforming the signals received by the 256 channels, with an angular resolution of 0.45 o. One of the measured sound fields as a function of arrival angle and range is shown in Fig. 2. The sonar was lowered with ropes from the ship up to a depth of 9.7 m. The transmission angle was depressed by 6.8 degrees from the horizontal direction, as it was measured by the INS. This unwanted misalignment was compensated at first order 1 by shifting the y- axis origin of Fig. 2. The lines in the figure indicate where the different arrivals are expected for the given geometry. The continuous lines are for the direct bottom and direct surface scattering paths, while the dashed lines give the paths of the specular surface reflection path of the bottom scattering and its reciprocal path. The signal levels from higher order mutipaths are too low to be observed. The measured field reveals that, for a fixed return time, there is a diffuse surface bounce that spreads much more than what can be explained by the depression angle from the vertical. This spread is likely to be due to the roughness of the sea surface which produced non-specular surface reflections from bottom scatterers at closer range than that corresponding to the specular path. These non-specular returns came from angles of arrival both larger and smaller than that of the specular path. 10 Surface scattering strength 10 Bottom scattering strength 0 0 Scattering strength (db) Measured (max) Measured (mean) Measured (min) 6 knots 11 knots Scattering strength (db) Measured (max) Measured (mean) Measured (min) Grazing angle (deg) (a) Figure 3. Back scattering strength Grazing angle (deg) (b) 4.2 Measurement of Back Scattering Strength from Sea Surface and Bottom Direct measurement of the back scattering strength from sea surface was made possible by the vertical configuration. Figure 3(a) shows the measured surface back scattering strength as a function of grazing angle using the vertical configuration of the sonar vs the 1 The depression angle causes also a spread in grazing angle (of the same order of magnitude as the depression angle) for fixed range due to the large horizontal beamwidth.

6 470 L.WANG ET AL. theoretical model [7]. The measured surface back scattering is plotted in three curves with minimum, mean and maximum values to indicate the spread of the data. The measured average wind speed was 11 knots over the period of sonar transmission. Two curves from the model are given in the figure for wind speed 6 and 11 knots for comparison. The surface scattering strength was almost constant for grazing angles less than 30 o. The increase of surface scattering strength measured at large grazing angles is due to the artifact of assuming a sinc beam pattern for the transmitter in the calculation of the scattering strength. The near constant surface back scattering suggests the main mechanism of the scattering was due to air bubbles below the surface induced by wind. The measured bottom back scattering strength is shown in Fig. 3(b) with two different model predictions. In the first (green line) sound speed and density derived from logarithmic grain size φ in Table 1 were used (i.e., c = m/s, ρ = g/cm 3 ). In the second (red line) sound speed and density measured from the core sample in table 2 used (i.e., c = m/s, ρ = g/cm 3 ). The sediment volume scattering parameter and loss tangent were adjusted to fit the data. The bottom scattering strength predicted with the measured sound speed and density from the core is higher than that using derived results from the grain size analysis. It is likely that the sediment in the core was compressed during the core sampling process, giving an increase of sound speed and density in the core samples. It is expected that the true value may be some where in between the values obtained by the two methods Signal level (sonar depth 14.8 m) Measured Signal level (sonar depth 8.7 m) Measured Signal level (db) Signal level (db) Range (m) (a) Range (m) (b) Figure 4. Signal level at various sonar depths. 4.3 Signal Level at Various Sonar Depths Figure 4 shows the received signal from one receiver channel as a function of range at two different sonar depths, one at about middle depth, the other one at shallower depth. The signal level predicted by ESPRESSO is in overall agreement with the measured results, although the model takes only specular reflection into account. Applying the bottom parameters obtained from grain size analysis results in a better prediction of the signal level, while using the direct measured parameters from the core sample results in a prediction higher than the measured data at longer ranges.

7 MULTIPATH EFFECT ON SAS NAVIGATION Ping to Ping Correlation at Various Sonar Depths DPCA makes use of the correlation of direct sea bottom back scattering between successive pings to estimate the displacement of the sonar. The received signal by the MASAI sonar at ping k can be expressed as X k (t)=s k (t)+n k (t) (1) where s k (t) is the direct bottom scattering and n k (t) is the interference including background noise and multipath signals. The predicted values of the correlation coefficient µ, derived from the modelled signal to noise ratio, are plotted in Fig. 5 together with the corresponding experimental results. It is seen that there are large variations between the two sets of modelled data and that the agreement between the modelled and measured data is not very good. For both models, the correlation peaks at close range, before the onset of multipathing and then decreases as the specular multipath enters into the sonar through the sidelobes of the vertical beam pattern. The correlation then increases again, as the specular multipath falls into the null of the vertical beampattern, to again decrease with range as this nulling effect diminishes. In addition both sets of modelled data predict the onset of higher order multipath at long ranges, which is more important for the model with higher impedance and sound speed in the sediment (in particular giving a higher critical angle) Ping to ping correlation (sonar depth 14.8 m) Measured Ping to ping correlation (sonar depth: 8.7 m) Measured Correlation coef Correlation coef Range (m) (a) Range (m) Figure 5. Signal correlation at different sonar depths. (b) In the experimental data, the correlation peaks at close range. The low values obtained at very short range are probably due to baseline decorrelation, resulting from residual motion of the sonar between pings and the high grazing angles of operation. The peak correlation coefficient increases with the sonar depth, as a result of the reduced sea surface interactions. The correlation then seems to decrease monotonically with range. The increase predicted by the models, due to the nulling of the specular multipath by the vertical beampattern, is not observed. This is indicative of the diffuse nature of the

8 472 L.WANG ET AL. multipathing. The decrease in the correlation with range is faster for the deeper sonar position [Fig. 5(a)] than for the shallower one [Fig. 5(b)], predominantly as a result of the reduction in the grazing angle for given slant range. 5 Summary The effects of multipath on DPCA micronavigation were studied experimentally. Comparisons between the experimental results and sonar model predictions seem to indicate that the diffuse surface reflection significantly affected the signal correlation. More analysis is required to confirm this assumption. It may be necessary to extend sonar models beyond the specular reflection scenario to take this effect into account. Acknowledgements We thank all the people involved in the MASAI 01 experiment. We also thank Eric Pouliquen for his help. References 1. Bellettini, A. and Pinto, M., Theoretical accuracy of synthetic aperture sonar micronavigation using a displaced phase centre antenna, IEEE J. Oceanic Eng., (submitted 2001). 2. Bellettini, A. and Pinto, M., Experimental results of a 100 khz multi-aspect synthetic aperture sonar. In Proc. 5emes Journees d Etudes Acoustique Sous-Marine, Brest, France (Dec. 2000). 3. Wang, L., Bellettini, A., Hollett, R., Tesei, A. and Pinto, M., Interferometric SAS and INS aided SAS imaging. In Proc. Oceans 01, Hawaii (2001). 4. Porter M.B. and Bucker H.P., Gaussian beam tracing for computing ocean acoustic fields, J. Acoust. Soc. Am. 82(4), (1987). 5. Porter, M.B. and Liu, Y.C., Finite element ray tracing. In Theoretical and Computational Acoustics, - Vol. 2, edited by D. Lee and M. H. Schultz (World Scientific Publishing, 1994.) pp Meyer, M. and Davies, G.L., Beam tracing techniques for high-frequency reverberation modelling. In Proc. 6th European Conference on Underwater Acoustics, Gdansk, Poland (2002). 7. APL-UW High-frequency ocean environmental acoustic models handbook. Technical Report 9407, Applied Physics Laboratory, University of Washington, October Fisher, F.H. and Simmons, V.P., Sound absorption in sea water, J. Acoust. Soc. Am. 62(3), (1977).