Side-scan sonars provide an acoustic oblique image analogous to an aerial photograph of the seafloor. By ensonifying a swath of seabed and measuring the amplitude of the back-scattered return signals, an image is built up of objects on the seabed and some information on the morphology and substrate content comprising the seabed. High frequency sonar (e.g. 500kHz) provides high-resolution images, but with a small width (50 - 100m) of the seabed. Lower frequency systems (e.g. 100kHz) provide larger width coverage (e.g. 500m) of the seabed but with lower resolution. A side-scan sonar
See NOAA's summary table summary view of side-scan-sonar technique (239 KB PDF)
NOTE: The content below is derived from Chapter 4 of Acoustic Techniques for Seabed Classification (2005) (11 MB PDF) by J D Penrose, P J W Siwabessy, A Gavrilov, I Parnum, L J Hamilton, A Bickers, B Brooke, D A Ryan and P Kennedy.
Introduction | Basic side scan sonar operation | Considerations in side scan sonar operations | Examples of current side scan sonar systems | Processing and classification of side scan sonar data | Seabed surveys and classifications using side scan sonar | Use of interferometric side scan
When considering the range of acoustic techniques available for seabed mapping and classification, there is a significant step between technologies that provide information about discrete points along a line below the vessel such as single beam sounders and sub bottom profilers and swath type technologies such as sidescan sonar and multibeam systems.
In the operation of a single beam sounder, the time to first return is used to provide the depth below the vessel and from analysis of the subsequent backscatter, parameters relating to the nature of the seafloor can be derived. No matter how large the footprint of the acoustic beam is however (which is a function of depth and transducer angle), only a single set of parameters for each acoustic transmission can be obtained, as discussed above. In the case of backscatter analysis, often the return from a number of ‘pings’ are averaged to derive a value for the ‘roughness’ and ‘hardness’ of the seabed. The values obtained are taken as being for a single position below the vessel and surveys therefore provide a line of discrete points along the vessels track for which a number of parameters such as depth, hardness and roughness have been derived for each point. The spacing of these points is a function of the depth, vessel speed and number of pings used to calculate the required parameters. The situation is similar for sub bottom profilers that use a combination of power and low frequency to penetrate the seabed to provide information about the structure and nature of the substrate below the surface. The returns from each ping from a sub bottom profiler are also used to generate information about a single point below the vessel and are built up into a continuous track line. Swath technologies differ fundamentally from single beam sounders and sub bottom profilers in footprint shape and in that they provide spatially referenced information about the variability of recorded parameters from within the footprint (Figure 1).
Figure 1. Comparison of coverage between acoustic mapping technologies.
The two fundamental products of bathymetry and backscatter produced by single beam systems are also however the basic products of swath technologies. Swath technologies are typically divided into two main types of equipment; sidescan sonars and multibeam sonars. Although this chapter is dedicated to a review of activities relating to sidescan sonars some comparisons and discussion of their relative merits are necessary. Sidescan sonars and multibeams are often used in unison to gain complementary data sets, but recently the uses and types of information provided by multibeam and newer interferometric sidescan equipment have converged. To briefly introduce a comparison of swath systems it can be said that traditionally side scan sonars provide only backscatter data and little information about depth forming a wide, often almost photo realistic image of the seabed (Figure 2).
Figure 2. Example raw ‘waterfall’ sidescan record.
The primary product of multibeam sonars however is bathymetry and although backscatter is also recorded, the imagery acquired is generally accepted to be of a lower quality than that recorded from sidescan sonars. Recently however, sidescan sonars known collectively as ‘interferometric sonars’ that can provide depth information are becoming more popular. The crux of the comparison between multibeam systems and sidescan sonars was recently eloquently stated by Lloyd Huff at a 2004 University of New Hampshire workshop attended by Australian researchers, "Multibeam manufacturers are trying to provide better quality backscatter and sidescan manufacturers are trying to provide bathymetry" .
As the subject of this review document is ‘seabed classification’, the means of processing and classification of the information acquired is also addressed, although it should be noted that sidescan sonars are rarely used as a single technique for this purpose. At minimum the acoustic records of sidescan sonars are validated using direct sampling techniques such as video or still photography or direct geophysical and biological sampling by grab, but are often used in conjunction with multibeam or single beam sonars and sub bottom profilers.
The following sections describe the two types of sidescan systems available with reference to selected manufacturers, specific models and published literature. Software used for processing, classification and visualisation is discussed. The choice of examples used and any omissions does not necessarily reflect any bias in the choice of equipment or preferences of the author. It is outside the scope of this document to compare all the equipment available and to provide an exhaustive list of all the work and research that has been carried out and published. The discussion is also generally limited to that relevant to surveys of continental shelf type depths, with a bias towards ‘habitat mapping’ type work carried out in Australian coastal waters.
Basic side scan principles are covered well in Blondel and Murton (1997), and Fish and Carr (1991), although there have been many recent advances in personal computer based digital acquisition and processing packages since their publication. Sidescan sonars typically consists of two transducers mounted either side of a vessel, ROV, AUV or more commonly, a towed body or ‘fish’ (Figure 3).
Figure 3. Typical sidescan sonar fish.
Each transducer produces a thin fan shaped beam that is concentrated on a thin line that runs from below the fish perpendicular to the direction of travel out to a maximum range that is restricted by frequency, power and transducer design (Figure 1). As the pulse of sound emitted by the transducers interacts with the seafloor at angles off normal, most of the energy is reflected away from the transducer. The acoustic backscatter that is reflected back to the transducer from the seabed is recorded for an extended period of time for each ping forming a time series of amplitudes. Using the vessels position, speed of sound in water, and the height off the bottom, the position on the seabed can be predicted for any point on this time series and a line of instantaneous backscatter amplitudes can be created that is referenced to positions along the beam footprint on the seabed. This short statement however belies the range and complexity of corrections that are made to the image at this stage to compensate for various geometric and radiometric distortion.
As the transducers move through the water, the lines of data recorded from each subsequent ping are formed into an acoustic image of the area. A example of a typical raw ‘waterfall sidescan image is shown in Figure 2.
This image is a record of the instantaneous intensity of the backscatter and is affected by the following factors, in decreasing order of importance (after Kvitek et al., 1999).
The vessel can then be manoeuvred to obtain the desired coverage of the seafloor from successive tracks. Although historically sidescan sonar records have been recorded on thermal paper printouts (Figure 4), most manufacturers (and a number of aftermarket equipment suppliers) offer sophisticated digital acquisition systems with advanced features.
Figure 4. Sidescan sonar thermal paper recorder.
Figure 5. Display from sidescan digital acquisition system.
Whilst surveying, these systems offer a number of ways to view the acoustic imagery live whilst surveying. The raw data obtained is displayed as a ping by ping record known as a waterfall. Using an input from a GPS, this raw data is positioned correctly in geographic space and live ‘mosaics’ can also be produced building a composite image of the seabed superimposed on charts as the survey is underway (Figure 5). The current generation of digital acquisition software therefore function as navigation aids for the survey assisting the vessel in following defined survey lines, ensuring the required coverage is obtained and avoiding obstacles.
Sidescan systems are available from a number of manufacturers. These units vary in terms of frequency or combination of frequency and configuration (towed or vessel mounted, digital or analogue). High end systems also provide special features such as focused beams, high tow speeds, chirp technology (wide frequency range), synthetic apertures or in the case of inferometric systems, the ability to obtain bathymetry. The latter are considered here as a special case and are described in a later section.
Before discussing available equipment, it would be prudent to introduce some concepts, compromises and drawbacks relating to the use of sidescan sonar. A good introduction to the practicalities of using sidescan sonar can be found in Bennell (2001) and Kenny et al. (2003).
Typical frequencies of sidescan sonars used in nearshore mapping range between 100 and 500kHz. Although higher frequencies can provide maximum resolution of approaching a few cm, their ranges are significantly more limited than that at lower frequencies. At 100kHz maximum ranges are typically around 200 to 300m per transducer, forming total swaths widths of up to 600m, with a typical maximum resolution of 0.15m. At 500kHz ranges would be reduced to 75m per side, but with increased resolution. The choice of frequency has obvious ramifications for survey planning.
The great utility of sidescan sonar is the wide swath and surveys may attempt to exploit this to provide the fastest coverage of an area. There are a however a number of considerations that must be made that depend on a trade off between the financial resources available and the quality and coverage of data that is required. The resolution vs range compromise is discussed above, but as the beam angles of the transducers are typically fixed there are also a range of geometric considerations affecting coverage that must also be considered. The first of these is that the transducers are angled so that the fish is effectively blind to an area each side of the ‘nadir, the centre line of the direction of travel (Figure 6).
Theoretically this means that if full coverage of the seabed is required swaths must be overlapped by at least 50% to achieve 100% coverage. Often however full coverage is not necessary to gain the required information relating to boundaries on the seabed and swaths are typically overlapped by 20 to 30%. The second is that to achieve the maximum coverage possible, the transducers must be a suitable height above the seabed (between 10 to 20% of the swath width). Should this height off the bottom not be able to be achieved, for instance where the water is too shallow, the swath width will be reduced from maximum and will be typically a multiple of water depth. For most 100kHz systems this multiple is about 10 per side with swaths widths maximized at 200m per side with the fish traveling 20m from the seabed. There are therefore water depths at which the use of higher frequencies will not compromise range.
Figure 6. Beam pattern of sidescan sonar.
Unsurveyed blind spots are also caused by acoustic shadows behind high relief terrain (Figure 4.6). These shadows become larger at larger ranges. To obtain full coverage the seabed must be ensonified in two directions and overlaps between subsequent swaths must be 100%. A further consideration when planning survey tracks must be made which relates to the spread of the beam with range (Figure 6). This has the effect that imagery of similar features obtained at far ranges appears quite different than that obtained at near ranges. Because of this effect, the requirement for high quality imagery often limits the range that can be used in a survey. This effect is exacerbated when the fish is subject to movement. A bad example of this effect is shown in Figure 4.7 where seagrass hummocks appear quite different in the near range than the far range of a raw sidescan record.
Figure 7. Sidescan sonar image of seagrass hummocks exhibiting beam spread in far ranges.
The survey speed is also a consideration when planning surveys. The ping rates of equipment are selected by the operator and are generally limited by the range in use. For longer ranges, the backscatter signals will take longer to be received and the delays between pings will be greater. This also means that along track resolution of data will be less than for data obtained at similar speeds with lower ranges. A compromise must be reached where the survey speed is maximized but the required data quality is achieved. Sidescan sonars are normally towed at 4 to 6 knots, but more modern high end equipment is now available that can be towed at up to 14 knots. As an example, on a 200m range setting, a sidescan sonar will typically ping no more than three times in a second. At 6 knots this gives an along track resolution of approximately 1m.
Although sidescan sonar is normally thought of as being deployed as a towed body, it can also be mounted on the hull of a vessel, or on the body of an ROV or AUV. The great advantage of towing is that the fish and therefore the transducers can be maintained at an optimum depth above the bottom in deeper and varying depths of water, maintaining footprint, resolution and geometry. Towed bodies are also less susceptible to movement due to sea state, although for most vessel mounted systems this issue has been overcome with the use of accurate motion reference units and heading gyrocompasses. Sophisticated software can then use the records of pitch, roll, yaw, heave and heading to compensate for the motion of the vessel due to sea and weather conditions. It should also be noted however that a number of more expensive towed sidescan sonars also have these features helping in monitoring the attitude of the towed body.
Towing has a further disadvantage that it is difficult to predict the horizontal position of the fish with reference to the GPS antenna on the surface. This really relates to attempts to predict the horizontal distance the fish is behind the vessel known as the ‘layback’. Accurate acoustic positioning systems for towed bodies are available at varying cost which can help position the fish (see technical section on positioning). Estimates of layback can also be made by performing a patch test where overlapping tracks in opposite directions are performed over a distinct feature on the seabed and the layback is adjusted during processing until the features on both tracks line up.
Sea condition also affects the operation of sidescan sonar. Even vessel mounted systems with sophisticated positioning have a limit of conditions that they can operate in and the records from and ability to position towed systems quickly deteriorate in poor conditions. Often the survey set up will have to be made with reference to the weather conditions so the vessel is running into and away from the prevailing sea or wind. The limits that can be worked are specific to the vessel, conditions and area, but typically larger vessels can operate in poorer conditions.
Older style analogue systems such as the Edgetech 272TD have been used in the offshore industry since the late 1970’s, but are still in production (Figure 4.3). A towed fish providing 105kHz or 390kHz operation, it has a reputation for producing high quality imagery with ease of support in the field. Ranges from 105kHz operation are up to 200m per side, reduced to 75m per side at 390kHz. The standard 260 surface unit and paper chart reader (Figure 4.4) can be replaced with aftermarket systems from CodaOctopus or Edgetech and a range of configurations of personal computer based digital acquisition software from manufacturers such as Chesapeake Technologies are available. Whilst still being robust and supportable units, analogue sidescans such as the 272 have the disadvantage of having an expensive multicored cable. The move to digital sidescan systems has reduced the number of cores to two assisting in improvements in weight, drag, cost and times to repair. Lying between analogue and digital systems, the GeoAcoustics 159D (114/410kHz) is a common workhorse unit that uses two cores to transmit analogue data to the SS981 surface unit.
More modern dual frequency digital systems are available in the form of the Edgetech DF1000 (105/390kHz), C-MAX CM2 (100/325 or 325/780kHz), CodaOctopus 460PX (100/325kHz or 325/760kHz) or Klein 3000 (130/455kHz). (see examples of manufactures imagery )The Klein 3000 has the ability to acquire simultaneously at both frequencies and claims ranges of twice that of other systems of similar frequencies. All these systems are supplied with digital acquisition systems which can also be interfaced with live mosaic and navigation aftermarket software. At the high end of the market are more sophisticated pieces of equipment that have additional features. The Edgetech MPX has the ability to be towed at 14 knots, over twice as fast as conventional equipment. The Benthos SIS-1624 uses chirp technologies to gain high quality imagery at a range of frequencies simultaneously and the Klein 5500 obtains very high quality imagery by using focused beams. Geoacoustics have also pioneered a synthetic aperture sidescan sonar for very high resolution surveys (Hiller, 2005). A comprehensive and extensive round up of sidescan sonar systems is listed in the product review of the April 2004 issue of Hydro International (Hydro, 2004).
Although composite mosaics can be produced live whilst surveying, final imagery and classification require post processing. A number of standard formats such as XTF, QMIPS and SEGY have become popular over the years for the recording of hydroacoustic data and most acquisition systems have the ability to produce one or all of these formats. Most post processing software similarly has the ability to read files in at least one of these formats. All the formats have the ability to store the ping by ping amplitudes of the backscatter referenced with information recorded from a number of other sensors and equipment including GPS, depth sounder, cable out counter, acoustic positioning system, motion reference unit and gyrocompass. A file produced for each track therefore contains all the information necessary to process that track into a fully georeferenced image.
The process of amalgamating all the files into a composite mosaic of the image produced is known as ‘mosaicing’ and is performed by software made available by the manufacturer or any one of a number of third party manufacturers of hydroacoustic postprocessing software. An example of a sidescan mosaic is shown in Figure 4.8a. Well known examples of these are Triton Elics, Caris, QPS, QunietiQ and Chesapeake Technologies. In the production of a mosaic, the software performs geometric and radiometric corrections to compensate for the distortions caused by the vastly differing geometry between the returns from close to the fish and from those from the maximum extent of the range. To perform these corrections it is necessary to know the altitude of the sidescan sonar off the seabed. This is often accomplished whilst surveying by a bottom tracking routine that detects the first return for each ping and records the height off the bottom for each ping in the file. Problems with the detection of the bottom in survey can be corrected during the post processing stage. A suitable pixel size that the mosaic will be produced at is also chosen taking into consideration the resolution of the original acquisition frequency, the detail required and size of the file that will be produced.
Mosiacs produced can be considered as images of the seabed and resemble in many ways a monotone aerial photograph. They are typically produced in georeferenced raster image formats such as GeoTIFF or GeoJPEG suitable for use in GIS systems.
Further information on the processing of sidescan imagery can be found in Bennell (2001).
Classification of sidescan imagery refers to the action of aggregating areas of similar acoustic signature, and then attributing them with information relating to their biological or physical characteristics. Areas can then be described using a suite of standard descriptors known as a classification scheme. This is very rarely accomplished using only a single technique and although the sidescan imagery obtained can be of a high resolution, it generally requires ‘ground truthing’ or validation by fine scale techniques such as video and still photography or direct sampling. The data sets required to produce a classified map of the seabed are illustrated in Figure 8.
The process of classification of the seabed in this case can then be considered as involving two discrete tasks, although in practice there is some overlap between them. Initially, imagery of the seabed is segmented into discrete areas exhibiting a particular acoustic signature. This signature is thought of as being characterized by the amplitudes of the backscatter in the imagery and the interpixel relationships or texture within regions of the image. The segmentation can be carried out either manually by visual analysis, or automatically by specialised image classification software. It can also be carried out on either the raw waterfall images of the individual tracks or on the mosaic of all the tracks. There are a number of advantages to carrying out the analysis on the raw images as they are typically of a higher resolution than the mosaic imagery and contain information relating to the original time series recorded. This means that analysis can be carried out with reference to the original geometry with which they were acquired. Mosaiced images do not contain information on survey direction and pixels in the image cannot be referenced to their across track positions in the swath, or the beam angle for which they were acquired. It is essential therefore that all possible corrections should be applied to the sidescan data before mosaicing. The specialised nature of sonar formats means that if the raw waterfall images are to be analysed, dedicated software and some means of mosaicing the classified imagery is required.
This functionality is provided by a number of products from specialised hydroacoustic software manufacturers described in the following section. As sidescan mosaics are simply standard TIFF or JPEG images, they can be analysed by a number of image processing and GIS products such as ESRI ArcGIS, ERDAS Imagine and ER Mapper.
Figure 8. Processing of sidescan sonar and video data into a classified map of the seabed a) sidescan sonar mosaic, b) raw waterfall image c) classified video track d) timestamped and georeferenced frame of video e) classified map .
Visual segmentation of the sidescan imagery is usually accomplished in GIS software by digitizing polygons around areas of similar texture or intensity on the mosaic. This is performed with reference to all other data that is available for the area. Maintaining a continuous fixed viewing scale whilst digitizing helps maintain some uniformity in the mapping resolution.
Automated classification of sidescan imagery is generally carried out through analysis of the texture or amplitude within a window of a defined size . This window is then stepped across and down the image until each pixel, or group of pixels is assigned a value for each type of analysis performed. Pixels or groups of pixels with similar values are then amalgamated with reference to the required classification scheme.
As sidescan sonar obtained over large areas and varying conditions or depths typically varies in intensity, textural indices, more robust to variations in gain are often used to classify the imagery
The types of textural analysis vary, but are well documented in published literature. The most common type of textural analysis of sonar images uses Gray Level Co-occurrence Matrices (GLCM) to provide a range of second order statistics related to the texture of an area of image. The GLCM features published by Haralick et al. (1973) have been incorporated into a home grown package known as TexAn (Blondel and Murton, 1997; Blondel, 1998)) which has been used in a number of surveys (Blondel, 1998, Huvenne et al., 2002). GLCMs were also used by Cochrane and Lafferty (2001) and have been incorporated into a number of commercial sonar processing packages such as QTC Sideview (QTC, 2002) and Triton Elics Delphmap. Brown et al (2002) used Delphmap to segment sidescan images in a comprehensive investigation of texture, sediment and biotope.
There are a number of other statistical methods of analysis that have also been investigated. Griffiths, et al. (1997) analyse real data using statistical scattering models and Reut (2000) and Finndin (1995) use power spectrum analysis of the backscatter envelope to classify images. Co-variance models are widely used in image analysis (Jain, 1989) but have been neglected in sonar classification (Finndin, 1995). Carmichael (1998) and Finndin (1995) have provided some treatment of this subject. Jiang, et al. (1993) and Mignotte et al. (2000) have employed Markov random fields to characterise the seabed texture and then employed a neural network for the classification. Neural networks have also been applied by Stewart, et al. (1994) in an investigation of textural features based on spectral estimates, grey-level run length, spatial gray level dependence matrices and gray level differences. Attention has been paid to derivation of features from fractal and multifractal measures by Linnett (1991) and Carmichael et al. (1996) and Pace and Gao (1988). Investigation into feature extraction using a spatial point model has been undertaken by Linnett, et al. (1995).
Sidescan sonar texture classification has also been the subject of a substantial number of Ph.D. thesis from Universities in the U.K., the research from which has inspired some of the papers cited above. From Heriot-Watt University, these include Linnett (1991), Tress (1996), Shippey (1991), Clarke (1992) and Shang (1995) and from University College, London, Dunlop (1999).
Despite the wealth of research describing textural analysis of sidescan imagery, there are only a limited number of commercially available packages dedicated to such analysis. The Seaclass extension for Delphmap from Triton Elics uses GLCMs to analyse mosaiced imagery but does not operate on the raw waterfall data. It employs a neural network to group the pixels into classifications based on the statistics. QinetiQ use fractal analysis in their mosaicing, classifying and GIS package Classiphi. Textural analysis in Classiphi can be carried out at the waterfall or mosaic level and it provides scope for training the system in the seabed types that are to be classified for supervised classification.
In their comprehensive range of sonar classification software, QTC provide for classification of sidescan imagery in their Sideview product. This takes a slightly different approach to classification of the imagery in that it uses a range of textural analysis techniques to segment the seabed. As well as using basic statistics and GLCMs, it also uses fast Fourier transforms, power spectra and fractal dimension. Using a unique clustering method it allows for both supervised and unsupervised classification based on the best features to capture seabed diversity. Sideview also offers a feature that allows sophisticated compensation of distortions and artifacts in the raw sidescan data.
GeoAcoustics ( UK) offer a textural analysis package that can operate on imagery of any type, although when used with sidescan imagery it also functions as a mosaicing tool. Providing supervised classifications, GeoTexture is trained in the seabed types required in the segmentation. GeoTexture also provides sophisticated means of removing distortions and artefacts in the imagery before processing. Ocean Imaging Consultants also produce classification software for their OIC Toolkit.
The range of available publications of surveys processed is small but growing and manufacturers web sites should be monitored for recent work. Hydro (2004) highlights some surveys using automated classification software.
Although commercial hydrographic operations have been using sidescan sonar for object detection and monitoring and seabed classification for over 30 years in oil and gas and dredging work, the majority of surveys are only published in reports, much of which will not be available to the public. This review will therefore be focused towards habitat mapping projects in continental areas and will be biased towards information that is freely available relating to relevant surveys and projects.
Sidescan records from at least as early as the 1970’s appear to show evidence of seagrass beds and trawl tracks. Although many surveys have undoubtedly been carried out since this time, no published literature appears to be available relating to habitat mapping in shallow water using sidescan sonar until the late 1990s. Work carried out between these periods appears to be mainly related to deeper water and the identification of large scale geological features. Although these studies are not unrelated to seabed classification, they are beyond the remit of this review.
Barnhardt and Kelley (1998) used sidescan sonar to map and classify an area of complex seafloor in the Gulf of Maine. This survey uses visual classification of the sonar image validated with sediment samples. McRea et al. (1999) used sidescan sonar to characterise rockfish habitats near Kruzof Island, Alaska using sediment and video samples for validation. It is interesting to note that the sidescan used in this survey was interferometric and capable of providing bathymetry, although visually classified backscatter was predominantly used to delineate habitat type. In a comprehensive and quantitative survey of an area in the English channel, Brown et al. (2002) compared sediment type, and biological sampling with sidescan sonar classifications from DelphMap software to classify benthic biotopes. GLCMs were used by Cochrane and Lafferty (2001) to distinguish areas of different texture in the Channel Islands, California.
Aside from the above review of recently published data, a number of large habitat mapping programs are underway around the world that use sidescan sonar in conjunction with other techniques. The Geological Survey of Ireland in their Irish National Seabed Survey are in the process of mapping an area of ocean 10 times the size of their land mass using a range of techniques including sidescan sonar. They are currently in the final phase of this project, mapping the inshore shallow areas. In Canada and the US the Gulf of Maine Mapping Initiative (GOMMI) group are in the process of mapping a 165,000 square km area. This is to name but a few of the projects that are ongoing or have been completed.
Recent habitat mapping using sidescan sonar in Australia has been carried out by the Coastal CRC in Western Australia in the Recherche Archipelago (Baxter and Bickers, 2004) and Cockburn Sound and in NSW in the Cape Byron Marine Park between 2002 and 2003 (CRC, 2004). This work was carried out using the University of Western Australia’s Edgetech 272TD sidescan sonar and Chesapeake Technologies Acquisition Software and was validated by video and grab sampling. A Geoacoustics SS981/159D sidescan sonar system was used in the mapping of Pt Addis marine park in 2005 in a partnership project involving Fugro, Parks Victoria and the CRC as discussed the case studies.
The Defence Science and Technology Organisation (DSTO) have used their Klein 5500 sidescan in seabed classification of Sydney harbour as part of the data set comparing survey techniques for the Shallow Survey Conference 2003. The Klein 5500 was also used in the joint comparative study performed by Fugro and the CRC in the Marmion Marine Park off the Western Australian coast. An Edgetech sidescan system was also trialled along with a CMax sidescan and a number of multibeam systems and a GeoAcoustics GeoSwath Interferometric sidescan. This work is outlined in the case studies.
‘Interferometric’ type sidescan sonars have recently become popular for seabed mapping. These sonars employ multiple staves in each transducer to gain depths from interpretation of the phase angle between returning acoustic signals. Kenny et al. (2003) consider their genre as one of four of the main types of acoustic systems amongst single beam, sidescan and multibeam sonars. They compete with multibeam systems in the quest to provide simultaneous high quality bathymetry and backscatter from a single ping. This technique has the advantage that backscatter and bathymetry are perfectly coregistered as they are obtained simultaneously from the same acoustic transmission. Swath widths for interferometric sidescan sonars are generally quoted as being between that of conventional sidescans and multibeams. The research group active in the Gulf of Maine habitat mapping program consider that interferometric sidescans may be more efficient for mapping in waters of less than 30m depth. This issue is addressed further in the case studies.
The range of interferometric sidescan equipment that is currently offered is limited in comparison with the range of standard sidescan systems. Systems are offered as being predominantly vessel mounted although towed types are available. Similar systems are offered by SEA (previously Submetrix) and GeoAcoustics in the UK. GeoSwath systems available from GeoAcoustics are vessel mounted systems available in 125, 250 or 500kHz configurations. Like multibeam systems, these systems incorporate both sophisticated motion and heading sensors for accurate positioning of the bathymetry and backscatter obtained. The range of towed systems is more limited. The Benthos C3D towed system has recently become available in 200 or 100 kHz frequencies and Klein have also completed trials with an interferometric version of their 5000 system known as a 5004. The advantage of towed interferometric systems is that using sophisticated sensors mounted inside the towed body allows surveys to be carried out at a greater range of depths with a single frequency. Towing also allows swath widths and the resolution of the data obtained to be to be kept constant. Positioning of the fish itself is however more difficult to estimate. Again published marine habitat mapping surveys using interferometric sidescan sonars are few, but at GeoHab 2004 a mapping project was presented using this technology (Thorsnes et al., 2004). McRea et al. (1999) used an early interferometric sidescan to map rockfish habitats in Alaska and more recently Ojeda et al. (2004) used both a conventional sidescan and an SEA interferometric system on the same survey. A 250kHz GeoSwath was also used in Sydney harbour to gain information for the Shallow Survey 2003 data set and in 2004/2005 a 250kHz system was tested by Fugro and the Coastal CRC as part of the Marmion Marine Park data set discussed in the case studies.
In Australia, as of early 2005, the availability of interferometric systems is currently limited to two GeoSwath systems. One is a 250kHz unit operated by 3D Mapping in Adelaide, and the other a 125kHz system which was recently purchased by the NSW Department of Environment and Conservation. The NSW based system will soon be used to create habitat maps of all the marine parks in NSW.