? Are you really getting "full bottom coverage" ?

A collection of thoughts and images
by John E. Hughes Clarke

© Ocean Mapping Group, UNB

Within the preliminary draft of the new specifications for IHO S-44, there are some poorly defined terms. These include:

"Full Bottom Coverage" and "100% Ensonification"

Hmmmm... are you getting 100%? Well it all depends on how you ping. This report tries to graphically illustrate how all the variables add together to decide whether or not you are truely getting 100%.
(Whether you need to get 100% is another matter entirely).

Technical contribution toward:

Contents:

WHAT IS ENSONIFICATION?

For any acoustic source, at least some energy is radiated in all directions. For a simple spherical source (such as explosives) the energy is distributed equally for all solid angles. For most sonar sources, however, there is directivity (uneven distribution of radiated energy). Many sources are described as having a certain "beamwidth". This might seem to imply that energy is only radiated in a constrained solid angular sector. However what is really meant is that the "main" part of the energy is radiated within that described solid angle. Less energy (but not none) is radiated at other angles.

For our purposes, when we say the seafloor is "ensonified" do we mean:

If we use the second definition we need to define what we mean by "main".

We use sources with directivity because we wish to constrain the angular sector in which we arrive at an estimate of two-way-travel-time (TWTT) to the target (in the case of hydrography, the seabed).

For simple single beam sounders, angular beamwidths of as large as 30 degrees + were common and depth measurements relied heavily upon assumptions about the seafloor being remarkably flat so that the minimum slant range would be a reasonable approximation of the vertical depth. In contrast, for sonars that attempt to arrive at an estimate of the TWTT for an oblique ray path (or more generally, for any target that lies at a slant range beyond the first arrival), we need to be confident that we are making an estimate of the TWTT within a narrowly constrained angular sector (because the incidence angle is no longer close to zero). In this case we rely on the fact that the energy we receive has been predominantly reradiated from within only a limited solid angle. To achieve this we require a beam pattern with a directivity that matches this requirement.

In order to come up with a standard to define what size solid angle we are referring to we need a definition. The common one is the 3dB limit.

What is the 3dB limit?

10log(1/2) = -3(.01)dB. The value of one half, expressed in dB (10log(power)) is equivalent to -3dB. When we quantitatively measure the beam pattern of an acoustic source we recognise an azimuth from source where the peak power is radiated (referred to as the boresite). If we look away from this axis we can find an angular offset at which the radiated power is a half of that radiated at the axis itself. This is the 3dB limit. In case the beam pattern is asymmetic, we measure this angle in the inverse direction along a chosen plane and the sum of the two angles is referred to as the "3dB level beamwidth" (for that plane).

For most multibeam sonars, recognising that a narrow beam is formed by the product of two planar-like beams (the transmit and one of the receives) we chose to define the beam width separately along these two planes and thus describe any beam with two 3dB beamwidth values.

Figure: Steered Beam Product

image showing the product of a transmit beam and a steered receive beam. The two transparent cones represent the axes of the transit and receive beam patterns. Note that the beam width is different along the two planes described by the cone intersections. This beam has a larger "across track" beam width than "along track".

For the new IHO standards, the statement 100% ensonification isn't defined. What is probably implied is a region of confidence within which any target would be recognised. There are two problems related to this desire:

This really depends upon the shape of the beam pattern outside the 3dB limits. If the remainder of the beam pattern lies at ~-4dB then a significant level of energy will be returned from other targets at equivalent slant ranges. In general however, we are dealing with beam patterns generated by approximating a line array. In these cases, the characteristic beam pattern is one of a main lobe and side lobes all separated by nulls. For this type of beam pattern. sidelobe levels are down generally < -20dB and the rate of drop off of power into the Null approximates a sinc function.:

Figure: Beam Power Pattern

image showing typical beam pattern for an unshaded line array of length about 20 wavelengths.
    vertical lines are 5dB apart and
    horizontal lines are 10 degrees apart.

And thus whether we use the -3dB or -6dB levels has only a minor influence on our estimate of the ensonified area.

A critical issue is one of the horizontal scale of the target that is required to be found. IHO standards require about 1% vertical resolution, but state no horizontal resolution requirement. Presumably vertically mounted knitting needles/cricket stumps/telegraph poles must be located? We risk straying into the problem of feature resolution (see later discussion: effect of beamwidth on target detection) whereas we are only trying to address the "ensonification" issue at the moment.

Masking through shadows

It should be remembered that if there is any topography whose slope (in the direction of ensonification) exceeds that of the impinging ray, or has other parts of the seafloor between it and the source, it will not be ensonified, whatever the beam pattern.

This criteria becomes even more strict as the seabed gets rougher and as the beams go more and more oblique. Assuming that the source always lies above the targets of interest, it is impossible to shadow out a feature that is shoaler than the shading observed feature . Nevertheless, strictly if there are any shadows then 100% ensonification cannot be claimed.

TRANSMIT ISSUES

The prime thing to remember is that the area insonified by the transmit beam completely defines the region over which returns may be identified. Thus the orientation and heave at transmit are critical. No amount of receive steering can extract information about the seafloor for regions missed by the transmit beam footprint.

FORE-AFT BEAMWIDTH
The fore-aft beamwidth will be the prime control on the along track extent of the beam footprint. Most receive beams are much broader than this in the fore-aft direction (with the noteable exception of the EM950/1000).
In the absence of roll and pitch and with a flat seafloor, the fore-aft footprint distance is equal to twice the slant range times the tangent of the half beamwidth.
Strictly with transmit beam steering (for pitch or yaw stabilisation) the beam width will enlarge. But for common steering angle of less than 5 degrees this is is a small change (sec. of the steering angle).

PORT-STBD BEAMWIDTH
This has to be at least as wide as the required angular sector. Generally wider by a factor relating to the maximum expected roll.

EFFECT OF ROLL
As the roll is defined about the X axis of the body frame. The transmit fore aft beam width is unaffected. Potentially with extreme rolls, the port-starboard swath may be extended/truncated as the transmit beam width is rotated to the side.

EFFECT OF PITCH
This translates the footprint forward or aft (by the tan of the pitch angle). If pitch steering is applied, the footprint now reflects the intersection of the cone of the steered transmit beam and the seafloor:

EFFECT OF YAW
This rotates the line of ensonification. If the heading change between pings approaches half the transmit beam width, then inter ping gaps are going to start appearing in the coverage.

EFFECT OF HEAVE
In shallow water, heave (routinely between +/-0.1m and +/-1.0m) results in a significant change (1% to 10% in 20m of water) in the source altitude w.r.t. the seafloor.
Heave alters the instantaneous elevation of the source function and thus slightly enlarges or shrinks the ensonified footprint. But this is a secondary issue. The width of the swath is primarily controlled by heave at receive (during which time the effective swath width is set by the combination of angular sector and altitude at receive).

RECEIVE ISSUES

FORE-AFT BEAMWIDTH
While this is commonly quite large (to allow for changes in pitch between transmit and receive), this is not reflected in the fore aft dimension of the individual beam footprints as this is already constrained by the transmit ensonified area.

PORT-STBD BEAMWIDTH
For systems that form conventional receive beams, and which use amplitude detection methods, the port-starboard beamwidth defines the across track individual beam footprint (together with slant range and grazing angle). For amplitude detection methods, this width reflects the spatial resolution of the sonar (feature detection capability).
However a number of sonars use phase detection methods in various ways:

- Simrad "split aperture" in-beam phase method:
This method (see Simrad product literature for more details) relies on picking the zero differential phase point within the beam bottom strike window. Because the ability to differentiate angles based on phase is better than half the beam width (commonly (~0.1 degrees compared to ~1.7 degrees), the bottom detection spatial resolution across track reflects the phase resolution rather than the physical beam width. Thus although the physical beam footprint may be large, the system has the ability to resolve targets at across track distances finer than this (hence the tighter beam spacing than beam width used in EDBS modes).

- ISIS "multi row" interferometry:
In this method, the receive beams are all broad and overlapping in the across track direction. Angular resolution in not achieved by using narrow (across track) receive beams, but rather by looking at differential phase between vertically displaced transducer rows. As with the previous method, the ability to resolve angles is based on the ability to measure differential phase. This will depend on: the quantitisation of the phase; the element spacing; and the noise. Phase resolution is quoted at about 0.1degrees. As a result this system provides a time series of angle estimates (up to 1000 per side) rather than a series of TWTT for predetermined angles.

- ATLAS Fansweep 20 "phase detections":
Although not well explained in product literature, the method for bottom detection used by the Fansweep 20 for the outer part of the swath appears also to rely on differential phase. Again a very large number (up to 1440 solutions per swath) are presented which are not related to spacing of individual preformed beams, but appear to be similar to the previous method in that they represent a non-uniform sampled time series of angle estimates.

BEAM SPACING
Following on from the discussion above, it is easy to see how beam spacing of conventional amplitude detect solutions can be used to estimate the spatial resolution capability of a sonar. In contrast, for those systems that are basing their angular discrimination on phase, the feature detection capability may be significantly better than the physical beam spacing (if any at all).
There are two end member types of sounding solutions:

For the second case, it is very hard to define criteria to go into hydrographic standards which adequately state whether or not a particular system meets the feature detection requirement of that particular survey.

EFFECT OF ROLL

EFFECT OF PITCH
As most of the sonars considered have receive beams that are broad in the fore-aft direction, pitch (more strictly change of pitch w.r.t that at transmission) has little effect on the resultant beam location (pitch at transmission has already determined the fore-aft displacement of the resulting beam footprint).

EFFECT OF YAW
Little effect on the receive beams, again because they are generally broad in the fore-aft direction. Note for sidescans (including interferometric sidescans, yaw can have an effect as the receive beams are narrow fore aft, and any change in heading between transmit and receive can misalign the two beam footprints resulting in "striping" in the backscattered data time series.

EFFECT OF HEAVE
Heave will change the altitude of the sonar w.r.t the seafloor. The effect of this depends on whether the system is:

EFFECT OF FORWARD PROPAGATION

To graphically demonstrate the effect of motion during forward propagation on swath coverage patterns, a series of coverage plots have been generated for specific conditions of motion (with/without roll/pitch/yaw/heave and with/without roll/pitch/yaw compensation).

For all the models a constant velocity of 12 knots is used. The vessel actually propagates in a straight line (i.e.: no surge or sway), but is free to rotate about any of the three axes and to heave up and down.

The water depth selected is 20m and the sonar model used covers a 6X water depth angular sector with a Tr. beam pattern 1.2 degrees wide. The chosen repetition rate is 0.3 seconds.

The image of the seafloor exhibited is a patch of dimension 200 x 200m. The image presented is a 1200 by 1200 pixel representation of the seafloor with a pixel dimension of 17 by 17cm.

Areas coloured blue were not ensonified within the 3dB limits. Areas in greyscale were ensonified, the greyscale representing the range -3dB (black) to 0dB (white).

These sample images are not intended to represent any particular system, merely to show the effects of the differing motions (we thus choose a narrow beam, wide swath, system that wont achieve 100% overlap at these speeds in order to best illustrate the sounding distribution.

Figure: 32 second Motion Time Series
Figure showing the 32 second time series of roll, pitch yaw and heave used while the model sonar propagates over a 200 by 200m square of the seafloor at 20m depth. (The time series is in fact a real sample generated from a 36m vessel in sea state 4).

The most critical parameter to determine is the alongtrack distance that the vessel moves over during a shot repetition cycle (a ping). This will be controlled by the shot repetition rate and the vessel speed.

EFFECT OF PING PERIOD
Two-Way Transit Time:
The time taken between successive estimates of across track profiles is referred to as the "ping period". For those system that achieve single ping ensonification (only one transmit pulse is used to ensonify the entire across track profile footprint) the ping period must be greater than or equal to the time taken for sound to propagate to and from the most distant target. This will depend on the water depth and the obliquity of the measurement. Because the sonar does not know the depth before measurement (it can guess based on the previous depth) it will in general wait slightly longer than the TWTT to allow for increasing depth along track.

Compute Time
After reception of all the energy within the desired ensonified footprint most systems require a finite time period to do the bottom detection solution and (depending on whether this is performed by the same or parallel machine) the transformations necessary to compensate for orientation and water column.

Single Ping Total Period:
For most sonars the maximum ping period in shallow water is limited by the compute time rather than the TWTT. Until the early 90's this was about 0.5 seconds. It has since dropped to less than 0.1 seconds.

But not all sonars do single ping ensonification...

Multi Ping Ensonification
Some sonars (BottomChart, Fansweep 10), do multiple transmits, using beams that are narrow in the athwartship direction also. They require 7 and 14 pings (respectively) to acquire a full across track profile. Thus the TWTT fraction of their ping rate increases by this number. While this is a disadvantage from the point of view of coverage, the manufacturers would argue that the increased sidelobe suppression achieved (generally -36dB rather than ~-24dB) by this method more than offsets this disadvantage.
Other sonars ping alternatively to port and starboard to avoid mutual interference (Seabat 9002 and ISIS Submetrix). Thus the full coverage requires double the TWTT.

*** UPDATE *** for real examples of how the ping rate varies as a function of water depth for a number of commercially available sonars, follow this new link (01/98).

EFFECT OF SPEED
Simply the distance propagated between pings. But this distance on the seafloor can be modulated by variations of the orientation and heave of the transducer from ping to ping .....

EFFECT OF ROLL

As you can see the disadvantage here is the periodic lateral displacement of the swath. If the operator requires 100% coverage between swaths, under these conditions, the line spacing must be closed together to guarantee overlap under the worst condition of the vessel rolling away from the overlap on both swaths. As you can see, this essentially removes the effect of lateral swath displacement. The only major disadvantage is that it requires digital beamforming to allow time varying beam steering for every beam in the swath. To avoid this, some systems choose to preform beams over a larger angular sector than required and select only those subset of beams that fall within the vertically referenced beam sector required.
A lesser problem is that for receive beams that are already steered obliquely, the steering angle may have to increase with a resulting enlargement of the beam width and increased sensitivity to surface sound speed errors.

EFFECT OF PITCH

The visible disadvantage here is that the along track profile spacing becomes variable, resulting in either gaps or overlap. Indeed as the water depth increases it is possible to invert the order of profiles (sequential swaths displaced successively aft as the vessel nose drops). While pitch steering of the transmit beam can ensure equidistant along track spacing at nadir, This is not the case off nadir. This is because the transmit steering angle is relative to the long axis of the array so that the angle subtended with respect to the horizontal plane varies with the receive steered angle.

EFFECT OF YAW

In this case a simple periodic rotational displacement of successive swaths makes it difficult to guarantee 100% coverage in the outer beams. In the absence of roll and pitch problems, the minimum requirement is that the yaw offset over a ping cycle is less than half the transmit beam width.

  • What about Gyro Stabilisation?
    This can be done for a single sided system (and indeed has been done for many years with the IOS GLORIA system). For a single transmit array, one cannot achieve the same required steering for both sides at the same time.
    For those systems that use dual arrays, independent yaw compensation is possible. Note however, that to yaw compensate a beam in the far range is equivalent to pitch compensating a beam in the nadir region.

    *** UPDATE *** for a more up to date discussion for Active Yaw stabilisation, follow this link (03/98).

    EFFECT OF HEAVE

    In shallow water, the heave displacement translates into a noticeable change in the seabed-relative sonar altitude. Because the swath generally subtends a fixed angular sector, the swath width increases as the vessel heaves up and decreases as the vessel heaves down.
    On the other hand, for a system that receives data for a predefined fixed time interval (e.g.: Submetrix ISIS), the swath width illuminated will vary in the inverse sense (now, heaving up will reduce coverage). In general however, for wide swath systems, the swath distortion for the fixed time method will be less severe than that of the case of a fixed angular sector.
    Short of a mechanical heave compensator, neither of these effects are avoidable.

    COMBINED EFFECT OF ALL MOTIONS

    The worst case. How bad will depend on the scale of the periodic rotations and heave displacements which it turn will depend on the seastate and the vessel stability.
  • With Pitch and Roll Stabilisation
    While improved coverage is generally achieved for the inner part of the swath, it has been argued by some manufacturers that there is little gained in the outer part of the swath. The amount of coverage efficiency gained will depend on the seastate and water depth.

    SWATH FOOTPRINTS OF SYSTEM SPECIFIC MODELS

    In all cases the model is run for 20m water depth at 12 knots. No roll, pitch, heave or yaw is present. The images viewed are 200m wide by 16.7m high (pixel size of 16.7cm).
    The regions represented by a greyscale indicate those areas that lie within the -3dB limits of the product of the transmit and receive beam patterns. All areas beyond the 3dB limit are coloured blue.

    RESON 9001 (1.5)

    60 beams 1.5 deg FA and 1.5 deg. PS at 1.5 deg spacing. Shot timing for 20m of water is commonly around 0.1 seconds.

    RESON 9001 (10.0)

    The special case when using a 10 FA Transmit beam width.

    RESON 9002 (1.5)

    Twin 9001 heads mounted at +/- 30 degrees. Transmit beamwidth is 1.5 degrees for both. A shot period of 0.2 seconds per side is common. Note the sides fire alternately.
    The case when the 10.0 degree Tr. beam pattern is used.
    (NOTE: this is actually not quite right as I've modelled the receive beams as if they were steered w.r.t the individual 9001 head vertical axes (hence the apparent bowed footprints)). Nevertheless I think it demonstrates the danger of trying to meet unrealistic 100% coverage specifications!.

    RESON 9003 (1.5)

    40 beams 1.5 degree FA and 3.0 degrees PS spaced at 3 degrees.

    Simrad EM950/1000 (EA 150)

    60 beams per ping. 3.3 degrees PS and 2.4 degrees FA (the 2.4 is achieved because the FA beamwidth of both the Tr. and Rx. are both 3.3 degrees and thus the effective beamwidth is the product of both of these).
    This is an example of the "beam-hopping" philosophy whereby on alternate pings the entire receive beam pattern is shifted by 1.25 degrees. In this case the beams are spaced at equal angle (EA) intervals of 2.5 degrees. The shot interval is about 0.25 seconds per ping (0.5 for the two ping sequence).

    Simrad EM950/1000 (EDBS 150)

    Same beam hopping philosophy applied. However, in this case the beams are spaced in an "equidistant" (EDBS) manner. This results in an equal horizontal distance between each of the beam centres. As you will note, however, because the across track beam spacings subtended are variable (even though the actual beam widths aren't), the across track coverage is incomplete. 0.25 seconds per ping, 0.5 seconds for cycle.

    Submetrix ISIS 100

    1 degree beam width is specified. The system alternately pings port and then startboard. As the system provides a continuous series of angle estimates with time, there are no discrete receive beam patterns across track. A 0.23 second period per side is observed for operations in 20m of water. This is equivalent to a 170m slant range. Under these conditions, the useable bathymetric data extends out to about the equivalent of a 150-160 degree sector.

    ELAC BottomChart (SeaBeam 1180)

    This system forms 8 beams on both transmit and receive for each of seven pings. For each ping in the cycle, the 8 beams are successively shifted over. Over the 7 ping cycle a complete across track profile is obtained. On the eighth ping the cycle is repeated, and thus it appears as if beams are cyclically swept across the seafloor.
    The beam widths are specified as 5.2 degrees FA and 2.7 degrees PS (56 beams in total). A total period for the 7 ping cycle in 20m of water is observed to be about 0.7 seconds (0.1 seconds per ping).

    ATLAS Fansweep 10 (4X)

    This system forms 4 beams (on both transmit? and) receive for each of 14 pings. For the first ping the four beams are clustered at +/-45 degrees and they are successively stepped out from there to form a resultant W shaped pattern over the full 14 ping cycle.
    Although not explicitly described in the manuals, I have assumed that the beam spacing is equidistant. 56 beams are formed over the 14 ping cycle with stated beamwidths of 6 degrees FA and 3.0 degrees PS. The system requires about 1.0 seconds to complete the 14 ping cycle in 20m of water.

    Simrad EM3000 (single)

    This system provides nominally 127 beams per ping. The FA beam width is 1.5 degrees. The PS receive beam width varies (as all are steered from a single horizontally mounted line array) with a minimum of 1.5 degrees at nadir and increasing with the sec. of the steering angle. A 0.25 second ping period has been observed at this water depth.
    The beam spacing is controlled by the FFT beam forming, and is approximately 0.9 degrees at nadir and increases again with the sec. of the steering angle.
    The exact number of beam formed and their exact steering angles are controlled by the surface sound speed. For surface sound speeds of ~1480m/s one commonly receives about 120/122 beams over a 130 degree sector.

    ATLAS Fansweep 20 (100kHz) (6x)

    This system has a 1.2 degree FA beam width. No PS beam width is supplied. Commonly between 128 and 1440 depth solutions are provided across the swath. Without exact information, I have merely modeled the system as if it provides a continuous time series of angle estimates over a predefined angular sector.
    While this system has been convincingly demonstrated to operate to 12X the water depth, for this case the 6X water depth mode is used as it quoted as the mode in which the best performance is provided. An observed ping period of 0.5 seconds is seen with the 6X water depth mode in 20m.

    *** UPDATE *** for a more up to date discussion including the since -released : RESON SeaBat 8101 and the new ELAC BottomChart Mk II (SeaBeam 1180) follow this link (01/98).

    WITH MOTION AND FORWARD PROPAGATION

    RESON 9001 (1.5)
    Figure: RESON Seabat 9001 (1.5 Tr.)
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.10 obs. ping rate
      - 90 degree angular sector.
      - 1.5 degree Tr.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp Off
    Pitch Comp Off

    RESON 9001 (10.0)
    Figure: RESON Seabat 9001 (10.0 Tr.)
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.10 obs. ping rate
      - 90 degree angular sector.
      - 10.0 degree Tr.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp Off
    Pitch Comp Off

    RESON 9002 (1.5)
    Figure: RESON Seabat 9002 (1.5 Tr.)
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.40 obs. ping rate (P+S)
      - 150 degree angular sector.
      - 1.5 degree Tr.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp Off
    Pitch Comp Off

    RESON 9003 (1.5)
    Figure: RESON Seabat 9003 (1.5 Tr.)
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.10 obs. ping rate
      - 120 degree angular sector.
      - 1.5 degree Tr.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp Off
    Pitch Comp Off

    Simrad EM950/1000 (EA 150)
    Figure: Simrad EM950/1000 (equiangular beam spacing 150)
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.30 obs. ping rate
      - 150 degree angular sector.
      - 2.4 degree Tr./Rc. prod.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp On
    Pitch Comp Off

    Simrad EM950/1000 (EDBS 150)
    Figure: Simrad EM950/1000 (equidistant beam spacing 150)
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.30 obs. ping rate
      - 150 degree angular sector.
      - 2.4 degree Tr./Rc. prod.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp On
    Pitch Comp Off

    Submetrix ISIS 100
    Figure: Submetrix ISIS 100
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.46 obs. ping rate (P+S)
      - 150 degree angular sector.
      - 1.0 degree Tr.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp On
    Pitch Comp Off

    ELAC BottomChart (SeaBeam 1180)
    Figure: ELAC BottomChart BCC-SEE 28 (Seabeam 1180)
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.70 obs. ping rate
      - 120 degree angular sector.
      - 5.2 degree Tr.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp On
    Pitch Comp Off

    ATLAS Fansweep 10 (4X)
    Figure: ATLAS Fansweep 10 (4xWD)
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.98 obs. ping rate
      - 128 degree angular sector.
      - 6.0 degree Tr.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp On
    Pitch Comp Off

    Simrad EM3000 (single)
    Figure: Simrad EM3000 (single)
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.25 obs. ping rate
      - 130 degree angular sector.
      - 1.5 degree Tr.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp Off
    Pitch Comp On

    ATLAS Fansweep 20 (6x)
    Figure: ATLAS Fansweep 20
    Coverage pattern for:
      - 12 knots
      - 20m depth
      - 200m x 200m area
      - 0.5 obs. ping rate
      - 140 degree angular sector.
      - 1.2 degree Tr.
    Roll On
    Pitch On
    Heave On
    Yaw On
    Roll Comp On
    Pitch Comp On

    EFFECT OF BEAMWIDTH ON TARGET DETECTION

    In order to detect a target bathymetrically one needs to:

    If the target is large w.r.t the beam footprint then it may not be necessary to completely ensonify it (if you get 10 strikes on a 5m sphere, does it really matter that you don't cover every inch of it?).

    The problem is one of spatial wavelengths....

    (1) have you ensonified the small target?

    This is the problem of coverage. For an infinitely small target you have to truely have 100% coverage. If one the other hand you specify a minimum spatial dimension for the resolved target then you need only guarantee that you have a maximum inter beam footprint gap that is smaller than this dimension.

    A commonly defined dimension is the "metre cubed":

    The following set of images are closeups of a 12 by 12 m square of the seafloor. The image (of dimension 1200 by 1200) now has a pixel dimension of 1cm x 1cm. The sonar model for a RESON Seabat 9002 system (1.5 degree beamwidth) is run using the same parameters as previously (12 knots, 0.4 seconds for both sides in 20m of water).

    The minimum footprint dimension is naturally 2*tan(0.75)*20 = 52cm. and the distance propagated between shots is 1.2m (between sides) or 2.4m (on one side). The seafloor is conveniently inscribed with 1m square lineations to show the resultant sounding density.

    As you can see, under these conditions, one cannot guarantee a strike every metre square. However, one does generally have about 144 strikes in the nadir region in this 12x12m square. Is that enough?. If one changes ones definition to having part of a beam (rather than the beam boresite) in each metre square, one can do a little better. But what it all comes down to is semantics. The IHO has to define more clearly theirs.

    If we wanted to do better (at getting the elusive 100%) we could either:

    **** NOTE, I'm still applying the same motion time series, but at 6 knots you use more time to travel 12 metres (hence apparent different motion history recorded). All models are for the sonar travelling from left to right, starting at time zero.

    Is this really worth the extra expense?

    The other aspect to consider in target detection is :

    (2) size of beam footprint w.r.t. target dimension?

    (Ignoring for the moment the problem of defining an "equivalent" beam dimension for a phase detection system rather than an amplitude detection system)

    Just because you ensonify a target does not mean you detect it. Take for example a broad beam sounder which ensonifies a patch 10m across. Within that patch there is an off-axis, 1m proud, boulder whose top lies beyond the the distance of closest approach. It will therefore not be detected bathymetrically.

    If we narrow our beam width, we lower the likelyhood of loosing the boulder echo within the surrounding seafloor echo. We now need more beams to cover the same area but we are unlikely to miss targets that are larger than the beam dimension. Nevertheless we remain insensistive to topographic anomalies whose spatial dimension is small w.r.t the beam footprint.

    Thus just enlarging the beam width to meet ill defined 100% coverage criteria can actually lower your probability of detecting small targets!

    SIMPLE SIDESCANS

    While we have focused primarily on the new generation of swath sonars that provide bathymetric solutions (slant ranges and depression angles for a given azimuth), the conventional or simple sidescan (which provides only slant range for a given azimuth) has been the tool of choice for hydrographers for many years. By examining the modulation of seabed backscatter (defined only in slant range and azimuth) the experienced hydrographer has been able to recognise the likely presence of off-nadir seabed targets. Because the seabed generally closely approximates a flat surface, the confidence with which the hydrographer can constrain the location of that target in horizontal position is good enough for him/her to subsequently investigate the target using single beam technology to establish a minimum depth.
    Thus while the "simple " sidescan provides no direct bathymetric measurement, it is an efficient tool for confident off-nadir target identification. How efficient will depend, as with the swath bathymetric systems on the imaging geometry (beam widths, speed, repetition rate, orientation, heave). Sidescans have conventionally been deployed as towbodies both to isolate the sonar from surface noise and motion and to achieve a better imaging geometry (lower aspect ratio). This results in a much higher likelyhood of casting shadows (which thus strictly violates the 100% ensonification criteria) while aiding in target recognition.
    Sidescans commonly transmit and receive on the same array and thus the beam footprint is defined by the square of the transmit beam pattern. Along track beam widths less than 0.75 degeees are common and the angular sector covered extends from nadir to a maximum slant range generally limited by a predefined shot repetition rate (rather than a predefined angular sector as is the case with swath bathymetric sonars). For a fixed shot repetition rate,the total swath coverage will decrease with towfish altitude. Also because a constant towfish height is maintained w.r.t the bottom, the swath width is generally water depth independent.
    Because the sidescan only samples in slant range and not depression angle, the result in not sensitive to roll or refraction uncertainty. As a result much wider angular sectors are commonly acquired, and thus a wider swath width than bathymetric sonars is common in shallow water (<20m). Inversely, again because the system only can differentiate slant range, it has little practical resolution close to nadir where the slant range changes only slowly.

    Looking at a model example of such a sonar:

    **** NOTE we are having severe aliasing problems here. The beam footprint is less than 16.6 cm wide and thus the image is undersampling the beam footprint. The only way to do this, is to go to smaller pixels and thus corresponding large images. I will try running this for a smaller subset (perhaps the 12x12m windows used for feature detection?).
    Also should I mask out the nadir region to show where we effectively have no definition (+/-20 deg off nadir)?

    We see (for 0.75 degree, 75m slant range max per side (0.1 seconds rep rate) in 20m of water (but with the towfish at 10m off the bottom) at 12 knots) that these sonars also do not strictly achieve 100% ensonification. This observation is well know and most conventional sidescan are not towed at speeds above ~5knots (for reasons of towfish stability also):

    second example, same conditions, but at 6 knots.

    There is an (expensive) way around this problem. That is the use of "multibeam sidescans". These systems (Huff, various references) form multiple parallel receive beams (all with the same orientation as the transmit beam, rather than orthogonal as is the case for bathymetric multibeam sonars)) displaced along the length of the towfish. This provides effectively multiple, successively displaced, ensonified areas rather than a single one, thus maintaining 100% ensonification even at much higher speeds. Such sonars, however, while affordable for the military minehunting community are (to-date) too prohibitively expensive (due to extra mechanical towfish stabilisation, and signal processing) for most Hydrographic services.

    Last modified November 08, 1996, by John E. Hughes Clarke (jhc@omg.unb.ca)