Single beam sounders

Brief Description

The echosounder provides water depth by measuring the two-way travel time of a high frequency pulse emitted by a transducer. The system must be calibrated to allow for errors introduced by temperature and salinity and other factors that affect sound velocity. The choice of echosounder depends on many factors including accuracy requirements, depth of water and resolution. Typical frequencies range from 10kHz to 200kHz. Echosounders operate vertically below the survey vessel to gather a single line of sounding. Extensive technical guidance exists documenting the procedures that should be followed for the various portions of a bathymetric survey, and internationally accepted standards have been developed to define different classes of accuracy requirements for the different elements (e.g., horizontal and vertical control) that comprise a bathymetric survey. For instance, the highest established IHO accuracy standards (Class 1) are normally followed for all inshore nautical charting and dredging-related bathymetric survey applications, primarily because of their importance to navigation safety.

See also NOAA's summary table Summary view of single beam sounding technique (212 KB PDF).
Echosounder depth profile image: courtesy Sonardata.

NOTE: The content below is derived from Chapters 3.1 - 3.2 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.

Back to top of page


Contents

Available systems | Principles of operation for single beam acoustic bottom classification systems

Available systems

Echosounder, single beam acoustic techniques for seabed classification using measures associated with substrate type have been widely applied, especially on continental shelves. Three commercial systems are noted below which use an extension of the technique employed by some Western Australian bottom fishing operators, probably amongst others, in earlier times. The approach is essentially to use some measure of acoustic backscatter signal magnitude to provide an estimate of the "hardness" of the seabed, and some measure of the length of the echo return to provide an estimate of seabed roughness. In the experience of the authors, one method, used to infer the presence of reef structures, was to run parallel to the depth contours with the sounder set over what was assessed to be a sandy bottom, at a gain setting which was just less than sufficient to provide a second bottom return. When second returns were observed, with or without sudden changes in depth, it was inferred that the substrate contained hard, reef dominated structures. Given the limited dynamic range of the sounder displays in use at the time, it was not clear how well this interpretation system functioned. The process did, however, contain the essence of several systems later developed commercially, in using the magnitude of the second echo as a measure of bottom hardness. Here and in the sections concerning the other broad-scale acoustic sensors "hardness" is used as a descriptor of the acoustic impedance of the substrate type and hence of the impedance contrast offered to an acoustic wave by the water-seabed interface.

The three existing commercial systems, and a number of variants not marketed commercially apply signal processing technology to nominal normal incidence single beam echosounders. Such systems will be referred to here as single beam systems to distinguish them from the multibeam systems to be discussed the multibeam section. The Tasmanian company SonarData hosts a valuable web site which covers the echosounder based systems discussed here and some related variants from other companies and institutions. These additional entries include the CSIRO developed ECHO software which the Division of Marine Research use on their Simrad scientific echosounder systems installed on RV Southern Surveyor. The SonarData web site has the address www.sonardata.com. This describes the Echoview software produced by that company and, through the sequence "Support and Download" and "Useful Links" connects to a variety of useful sites and publications, including material on benthic classification.

Single beam echosounders may be used to obtain a variety of information about the reflective characteristics of the seabed. They send a pulse of sound at a particular frequency (usually between 30kHz and 200kHz) that reflects from the seabed and the echo is picked up by the transducer. Three commercial bottom classifiers available in the market are the RoxAnn system, the QTC View system and the more recent ECHOplus system. The ECHOplus system is marketed by the United Kingdom company SEA (Advanced Products) Ltd. which appears to use similar techniques to RoxAnn. ECHOplus and RoxAnn are presented as similar in concept and function, although ECHOplus is advertised as being suited to dual-frequency sounder systems. In addition, ECHOplus is a digital system whereas RoxAnn is an analogue system.

Back to top of page


Principles of operation for single beam acoustic bottom classification systems

The general empirical basis for acoustic bottom seabed classification is well established, although a full theoretical basis to describe interaction of the incident ping with the bottom is not. Acoustic bottom classification systems use wide beam echosounders (beamwidth typically 12-55 ° ) to obtain information on seabed acoustic "hardness" (acoustic reflection coefficient) and acoustic "roughness" (as a backscatter coefficient). Pace et al. (1998) discuss inversion approaches which could enable seabed geoacoustic parameters to be estimated from normal incidence data. SACLANTCEN have developed the BORIS model to return the time series response of the seafloor similar to the signal received by echosounders. However, it is doubtful whether inversions will allow reliable estimates of bottom type for the complicated and variegated seabed types experienced in the real world. Shell components in particular can cause unpredictable returns, and particular echo shapes need not have a unique cause.

Wavefront curvature and echo shape

Interaction of an echosounder ping with the seabed

Figure 1. Interaction of an echosounder ping with the seabed (figure supplied by Andrew Balkin). The left hand side of the figure depicts the energy of the ping as it reflects from a horizontal seabed, and the right hand side shows the cross-section of the ping that is in contact with the seabed at the particular instant. In the centre frames, the back edge of the ping has not reached the seafloor, and a circle is ensonified. In the bottom frames, the back edge of the ping has already reached the seafloor, and an annulus is ensonified.

Because of wavefront curvature a ping from an echosounder with a wide angle beam ensonifies first a circle on the seabed, then progressively ensonifies annuli of increasing radii and lower grazing angles (Figure 1). If an amplitude envelope detector is used, then the signal recorded over a sampling interval is the total specular and backscatter return from some particular annulus. Echo shapes and energies depend on bottom acoustic hardness and roughness. The first part of the resulting echo shape (Figure 2) is a peak dominantly from specular return, and the second part is a decaying tail principally from incoherent backscatter contributions. A smooth flat bottom returns the incident ping with its shape largely unchanged, but greater penetration into softer sediments attenuates the signal strength more than acoustically harder sediments. Rougher sediment surfaces provide more backscattered energy from the outer parts of the beam than smoother surfaces (which simply reflect the energy away from the direction of the transducer), so that a rougher surface is expected to have a lower peak and a longer tail than a smoother surface of the same composition. The length and energy of the tail provide a direct measure of acoustic roughness of the sediment surface. The echo shape is also a function of echosounder characteristics such as frequency, ping length, ping shape, and beam width. Acoustic penetration into the bottom and presence of subsurface reflectors can also affect echo shape through volume reverberation. Acquisition and classification of echo envelopes allows the bottom type to be inferred from the energy and/or shape characteristics of the echoes.

The parts of the first and second bottom returns used by the RoxAnn system.

Figure 2. The parts of the first and second bottom returns used by the RoxAnn system. Energy of the shaded regions is integrated to form two indices - E1 (for the tail of the first echo – summation begins one pulse length from the echo start) and E2 (for all the second echo). From Hamilton (2001).

In reality the situation is more complicated as harder surfaces such as rock tend to have greater roughness and more random orientation of seabed facets than other sediments, resulting in widely varying return shapes and energies which can have an average signal strength resembling that of mud, if suitable averaging techniques are not used (Hamilton et al., 1999). This phenomenon was noted many years ago in deep sea work, and has been "rediscovered" for acoustic bottom applications. "Regarding the reflection of sound by the ocean bottom, experimental studies … have shown that sound reflection is determined by the parameters of the sediment only at comparatively low frequencies. At frequencies above a few kilohertz, bottom relief plays a dominating role. Reflection from a very rough rocky bottom may appear to be less than that from a muddy sediment" (Brekhovskikh and Lysanov 1982; section 1.9). Similarly, losses due to roughness effects can cause sand with ripples, sandwaves, holes, and scours to appear to some acoustic measures to have the same properties as mud. Suitable averaging of echoes can overcome much of this variability, however acoustic bottom classification results are sometimes ambiguous, a point which must always be remembered.

Back to top of page


The need for a reference depth

Effect of depth on echo shape for a very short ping. From Clarke and Hamilton (1999).

Figure 3. Effect of depth on echo shape for a very short ping. From Clarke and Hamilton (1999).

The shape and power of the returned signal can change significantly with depth, even if the bottom type remains the same. Examples are given in Caughey et al. (1994), Caughey and Kirlin (1996), and Figure 3. The returns for a particular bottom type are expanded (dilated) along the time axis for a deeper bottom, and compressed in time for a shallower bottom, so that returns from the same bottom sediment type lying at different depths do not have the same shape. This occurs because signals are sampled or digitised at equal time intervals rather than at equal angles (Caughey et al., 1994). More samples are obtained from one particular angle to another for a deeper bottom compared to a shallower bottom. Before the echoes can be processed they must be transformed to a reference depth e.g. average survey depth. Normalising echosounder waveforms to a reference depth allows signal sampling to correspond to a standard set of incidence angles, as opposed to a set of linearly spaced times (Caughey et al., 1994). For a particular echosounder this conveniently removes the need to allow for beam patterns, and for the backscatter function changing with angle of incidence. Spherical spreading corrections are also applied. Absorption can usually be neglected for short ranges for lower frequencies e.g. 50 kHz, but becomes increasingly significant at higher frequencies. Since the signal to noise ratio decreases with increasing depth, large depth variations over an area could influence these corrections adversely.

To transform a returned signal to a reference depth, time and power corrections need to be made. The time correction is first made to adjust the length of the returned ping. The power correction then corrects the effect of spherical spreading. These corrections are required because signals are sampled or digitised at equal time intervals rather than at equal angles (Caughey et al., 1994).

Figure 3 shows the effect of depth changes on a short rectangular ping. Normalising echosounder waveforms to a reference depth followed by resampling allows signal sampling to correspond to a standard set of incidence angles, as opposed to a set of linearly spaced times (Caughey et al., 1994).

The time correction employed enables returns from the actual depth d and the reference depth d 0 to maintain the same time/angle relationship (Caughey et al., 1994). Sampling at the same angles for different depths removes the need to allow for beam patterns, and the need to allow for the bottom backscatter function changing with incidence angle.

The time correction is (Caughey et al., 1994)

where d = the actual depth; d o = the reference depth.

Therefore

where t' = the corrected time; t = the time from the uncorrected signal.

Interpolation is then performed at times corresponding to reference depth sample times.

Back to top of page


Averaging of returns

Return echo shapes can vary markedly over a small time interval, even for the same bottom type. As a result of ship and sensor movements and natural variability the returns from any particular angle are of a random nature, sometimes adding and sometimes subtracting as bottom facets lying at slightly different angles and depths are encountered. Echoes are also subject to noise, natural variability, and echosounder instability. To obtain acoustic signal stability ten pings are usually averaged. Over rougher terrain simple averaging may not help ping stability, and can act to reduce overall ping levels from their 'true' value, causing rocky surfaces to be classed as muds, a drawback of some commercial systems (Hamilton et al., 1999). In this circumstance a smaller number of pings could be averaged or a different averaging method used e.g. Hamilton et al. (1999) suggested using the average of the one-third highest values in a ping set, under the assumption that higher energy returns are least affected by roughness effects. A system developed by BioSonics allows selection of the highest value in a ping set, or averaging of values over a selected threshold (Burczynski 1999).

Back to top of page


Allowance for slope effects

Particularly for multiecho methods, it is necessary to scan for successive points, or for sets of points, with ‘large’ depth changes. Very large apparent depth changes are often indicative of errors e.g. due to crossing bubble wakes, and should be removed at early stages of the processing. Other changes due to bottom slope must also either be removed or checked for acoustic data stability and reliability. For vessel speeds of about 4 to 5 knots, and a classification about every five seconds, ‘large’ changes may be 0.8 m or less, according to RoxAnn data from Sydney Harbour obtained at beamwidths of 50 °. This equates to a bottom slope of 4.5 degrees, quite a low value. Some confirmation is provided by a detailed analysis of slope effects on the QTC View system (von Szalay and McConnaughey 2001). They found slopes above only 5-8 ° caused misclassifications for two 38 kHz QTC View systems with beamwidths of 7 °x7 ° and 9 °x13 °.

Acoustic bottom classification systems ensonify different areas at different depths, so depth changes may change results even for the same bottom types. A postulated example from Rukavina (1997) is as follows: "it is important to note that where the bottom variability is at a smaller scale than the footprint, because RoxAnn integrates over the footprint it cannot distinguish e.g. … clay and boulders from a uniform gravel with the same average acoustic properties. Also the footprint size varies with depth".

Signal to noise ratio decreases with increasing depth, so that a wide range of depths in an area may cause poor classifications. A wide depth range can also affect the reference depth corrections.

Acoustic bottom classification systems are subject to bottom slope effects, especially for second echo methods. They may not provide reliable results near the sides of channels, over deep holes, or outcrops.

Back to top of page


Calibration

For classification the systems rely on establishing empirical relations between ad hoc acoustic parameters and sediment sample properties. System calibration and classification then become a function of the bottom sampling strategy, a key point which cannot be over-emphasised. Classification can also depend on the purpose of the user e.g. a mapping of fish habitat could produce a different classification from a mapping allied to grainsize. Video or similar groundtruthing is of central importance notwithstanding the restricted field of view associated with such techniques.

Acoustic systems are subject to noise and variability. Because of their empirical nature, classifications made using different acoustic bottom classification systems have an unknown relation to each other. Even for the same system and vessel, classifications could differ over time with changes in transducer characteristics with age or fouling, or in background noise, regardless of any changes to the environment.

Calibration methods may be classed as direct or indirect. Direct methods are applied by classing particular portions of the acoustic bottom classification parameter space, and generally seek quantitative calibrations: explicit correlations of portions of the parameter space are sought with bottom properties such as grainsize bounds or vegetation indices obtained at calibration sites. Indirect methods may classify in parameter or geographical space e.g. the RoxAnn space may simply be arbitrarily classed by rectangles of equal size, and the geographical class distributions so formed are then examined for obvious trends. A second example of an indirect method is that of applying image processing methods to RoxAnn data in geographical space, and then using groundtruth to assign meaning to the geographical classes (Greenstreet et al., 1997; Fox et al., 1998). The geographical classes so formed should be transferred back to RoxAnn space to check for outliers and errors. Indirect methods may be more appropriate for habitat assessments, where explicit separation of classes or groundtruth might not exist. For indirect methods Geographic Information Systems (GIS) could be used to overlay acoustic classes and groundtruth to check for correspondences or otherwise.

Back to top of page