Land Information New Zealand (LINZ) has the responsibility for the purchase of core hydrographic and bathymetric information on behalf of the Crown. This responsibility is outlined in the "New Zealand Hydrographic and Bathymetric Information Strategy" (NZHBIS). In that document, it is suggested that the Government can benefit from competitive services. As part of this, marine surveys are now tendered through LINZ and therefore a means of specifiying standards for such surveys is needed.
One of the Roles of LINZ in hydrography and bathymetry is considered to be the "development of operational policy and regulations, development and maintenance of standards, and monitoring of compliance with policies and standards". LINZ intends establishing new standards for Crown funded hydrography and bathymetry (Goal 1, Objective D within the NZHBIS). The new standards are especially necessary due to both the recent introduction of swath technology into New Zealand waters (HMNZS Resolution is now operational with an ATLAS MD2/30 ) and the fact that in the near future LINZ is likely to undertake swath data collection to define New Zealands legal continental shelf under UNCLOS.
While the NZHBIS defines the the "vision and direction" for this information, it provides no specific guidance as to standards for data collection.. For traditional single beam echo sounding, one may adequately fall back on the LINZ Hydrographic Survey Specifications (HYSPEC) or the Royal New Zealand Navy (RNZN) survey specifications which are modelled closely on the third edition of the International Hydrographic Organisation (IHO) Special Paper-44 (SP-44 Edition 3). With the advent of swath bathymetric sonar systems (most commonly multibeam echo sounder systems MBES), however, the IHO specifications have undergone significant modification. The latest release (IHO SP-44, Edition 4) seeks to propose new recommendations for survey standards which LINZ hydrographic specification document (HYSPEC) includes. At this point however, the contents of SP-44 Edition 4 are merely included in the document without any practical interpretation. The new IHO standards unfortunately contain significant ambiguity and are drafted for the sole purpose of data collection for nautical charting (a mandate much narrow than that of LINZ).
One example of this broader mandate is that, as of July 1997, LINZ has taken the responsibility for New Zealand's Continental Shelf Delimitation Project. This involves the "measurement and analysis of seabed information according to internationally agreed criteria developed by the United Nations Commission on the Law of the Sea (UNCLOS)". Unfortunately these criteria do not include any specificiations for the acquisition or delivery of data that might be acquired by MBES's.
This paper seeks to propose a set of specifications for those surveys which involve MBES. Only those specifications that are unique to the use of MBES are discussed. The usual specifications for positioning, tides and operator competence are not herein discussed. This paper is being developed for LINZ by John E. Hughes Clarke of the Ocean Mapping Group, at the University of New Brunswick in Canada.
While this document provides only a draft set of specifications to help clarify likely constraints on MBES mapping, the final approved document from LINZ " will be considered to be the authoritative basis of all core marine surveys, data collection and management, and charting. All service providers will be required to conform to these standards."
Swath bathymetric sonar systems have changed the way hydrographic and bathymetric surveys are run. The most common implementation is as multibeam echo sounders (MBES), but other methods including bathymetric sidescans and sweep sonars are used. In contrast to the older single, broad-beam echo sounders, all these swath systems provide a corridor of bathymetric data which describes both the alongtrack and acrosstrack topographic trends. Unlike the discrete two dimensional profiles derived from the single beam echo sounders (SBES), using MBES with inter-swath overlap, a near full description of the seafloor topography can be derived. This extra ability of MBES's over that of conventional SBES has necessitated a change in the International Hydrographic Organisation recommendations for minimum survey standards.
In order to understand the consequences of MBES one needs to compare in detail the differences in the way the old and new technology worked.
2.1 - Single
beam bottom search and target detection capability
Previously using SBES's, incomplete bottom search was accepted as a
practical limitation. Data density was defined by the line spacing, which
in turn was controlled primarily by the scale of the chart for which the
data was being collected. The single beam sonar was only expected
to find the shoallest depth within the beam footprint (immediately beneath
the vessel). Nevertheless, of course, any targets outside the beam footprint
(in general >95% of the seafloor) were left unsurveyed. As sidescan sonar
search technology became increasingly available, this was used by a number
of agencies to compensate for this inadequacy. Sidescan search, however,
was never actually specified in the IHO recommendations (and dosen't actually
provide a least height estimate, only an indication of a potential anomaly).
As routinely implemented, potential anomalies suspected from trends in
systematic survey lines or from sidescan had to be investigated as shoal
exams. For complex topographic regions this required significant
extra investment in shiptime.
Even though the beam footprints of single beam sounders were
relatively broad, due to the nature of the bottom detection method (first
arrival threshold), echoes from protuding targets within the footprint
could be picked up. This was possible even if the targets occupied only
a very small fraction of the ensonified footprint. Thus shoallest
point detection was excellent as long as the sonar was deployed immediately
above the target. Whilst the shoallest point was picked, the definition
of the target shape (of only secondary importance for hydrography, but
of increasing importance for other seabed survey requirements) was generally
poor as short wavelength feature definition was lost in the hyperbolic
echo trace. Shoaller points outside the beam footprint of course went completely
undetected.
2.2 - Bathymetric
swath sonar bottom search and target detection capability.
In contrast to near vertical incidence SBES, bathymetric swath sonars,
in general try to estimate the oblique slant range to the seafloor at ranges
beyond the first arrival using beams much narrower than those employed
in SBES (1-2 degrees rather than 5-25 degrees). This necessitates quite
different bottom detection algorithms (centre of mass or differential phase
commonly) which, while excellent for delineation of targets large with
respect to the beam footprint, will generally fail to resolve targets
much smaller than the beam footprint (the reasons are beyond the scope
of this paper but a discussion can be found in Hughes
Clarke, 1998). As these footprint dimensions range from ~1-10%
of the total waterdepth this will have a significant impact on the ability
of the sonar to detect very small seabed targets. If detected however,
in general the shape of the target will be better defined than with SBES.
Thus with the introduction of these tools, survey specifications need to
describe the minimum horizontal dimension of a feature that must be recognised.
Such specifications have been added to the new S44 document but again interpretation
of these requires clarification .
As bathymetric swath sonars provide information over a swath corridor that is, in general, several times the local water depth, for the first time it is now reasonable to be able to request some form of "complete bottom search/coverage/ensonification". Words such as these now appear in the new S44 document, but they are again open to interpretation. In its strictest sense, the term "complete ensonification" is almost impossible to achieve (Miller et al., 1997)
In addition to the improved bathymetric capability, MBES provide
an image of the seafloor based on changes in the seabed backscatter strength.
This potentially provides a means of delineating differing seabed types
(seabed characterisation or classification). Such a capability
allows MBES to deliver information beyond the bathymetric needs of
LINZ. As LINZ already has an additional requirement to characterise
the seabed based on bottom sampling, such a capability would allow more
selective (and hence less expensive) bottom sampling.
As part of the broad aim of generating new specifications to meet LINZ's needs (a process which is already underway as part of the Hydrographic Specifications (HYSPEC)), this paper seeks to expand upon those aspects of the hydrographic specifications that require either rewriting or better describing due to the availability of MBES. In general this includes five factors:
The NZHBIS outlines the Hydrographic and Bathymetric Obligation for New Zealand.
The obligations are:
These obligations "expand beyond the current focus of maritime safety and defence to include a wider range of Crown needs" . In particular the first obligation is wide open to interpretation. Presumably a more geologicaly-oriented mapping program is called for, perhaps along the lines of that currently underway by the US Geological Survey (Gardner et al., 1988) which focusses on areas of high human impact, addressing environmental needs (dumping, substrate mining, fisheries habitat for example). Such an approach would be particularly demanding on the thematic end, requiring regional maps of the seabed substrate far beyond that required for normal nautical charting needs (which are traditionally limited to the delineation of anchor holding grounds etc..).
4.1.1 - Sounding Accuracy and Density requirements
S44 ( fourth edition, 1997) describes survey standards for four different survey requirements or "orders":
In brief the 4 Orders are specified by IHO as:
For these Orders, S44 specifies the depth accuracy for reduced depth to be a combination of a constant (a) and a depth scaling component (b). These are defined to be:
Special - a = 0.25m , b =0.75%
Order 1 - a=0.5m, b=1.3%
Order 2 - a=1.0m, b=2.3%
Order 3 - a=1.0m, b=2.3%
The horizontal and depth accuracy requirements are generally unambiguous for the actual observations over long wavelength topography. As explained in the preceding section, however, ambiguity exists in interpreting these requirements for short wavelength features. This is partly addressed in two new criteria These are :
4.1.2 - Specifications for seabed thematic information
S44 does have limited guidlines for bottom sampling campaigns: " the nature of the seabed should be determined by sampling or inferred from other sensors (..) up to the depth required by local anchoring or trawling conditions; under normal circumstances sampling is not required in depths greater than 200m.". Even allowing for the fact that the Orange Roughy (Hoplostethus atlanticus) and Patagonian Toothfish (Dissosticus eleginoides) trawl requirements in New Zealand extend beyond 1000m this definition is clearly not adequate to meet LINZ's needs.
The density of bottom sampling required would "normally be 10 times
that of the selected line spacing". Given that line spacing for bathymetric
swath sonars is likely to be significantly different from single beam surveys,
should the bottom sampling density then change? While these guidelines
might serve for the narrower needs of nautical charting work alone, they
fall far short of any systematic attempt to define the composition of the
seabed.
A copy of the existing LINZ hydrographic survey specifications may be obtained from their website (HYSPEC). In this document, there have been a number of modification to include the new wording from S44 Edition 4. This includes defining the Orders, their changing vertical accuracy requirements and mentioning the bottom search and target definition requirements (parts 0301 and 0322). The new conditions are added but in general without any attempt to explain how their ambiguity can be resolved. A few additional statements do exist to aid in their implementation:
4.2.1 - Swath Bathymetric Sonar Standards
6. Linespacing for Multi Beam or swathe sounding systems
(MBES) is to be interpreted as the distance between the
outer limits of swaths and is to be as follows:
Order 1 Surveys 3 x average
depth or 25m, whichever is the greatest
Order 2 Surveys 3-4 x average
depth or 200m, whichever is the greatest
Order 3 Surveys 4 x average
depth
Presumably Special Order have no gaps between the swaths. The fact that interline gaps are allowed for Order 1 suggests that the full bottom search is met by other means. Sidescan searches (see below) are presumably that means although not explicitly linked to the Order 1 specifications. It seems strange that MBES technology when specified, will rarely be used to achieve any real 100% coverage criteria.
4.2.2 - Sidescan Sonar Standards
1. The requirement to conduct a sonar search will be advised in the survey contract. Sonar searches are to be conducted along all leading lines and recommended tracks.
2. Sonar searches are to be conducted rigorously to ensure total insonification of the survey area. The search is tobe carried out in one direction and its reciprocal using sidescan sonar. The sonar lines are to be spaced to ensure that the seabed directly under the transducer, and at least 50m beyond it, is insonified by the adjacent sweeps.
4. Sonar searches should not be conducted at speeds in excess of six knots. At speeds greater than this, small, but significant features may not be sufficiently insonified to produce a trace. At six knots, all targets of two metres and greater in cross-section should produce a ?return?.
5. The optimum height at which to keep the fish above the seabed is equivalent to 10% of the range scale in use, i.e. using the 150m range scale the fish should be flown 15m above the seabed.
This implies that, sidescan may be required in some instances (presumably , although not stated, Special Order and Order 1). The overlap requirement is effectively 300% or greater as for example using up to a 150m range scale one would need to run lines at 100m spacing. This allows one to examine the poorly defined regions immediately below the adjacent swath. Interestingly sidescans are only constrained in speed to the point of guaranteeing the visibility of 2m dimension targets (i.e. not the full Special Order survey).
As part of the LINZ HYSPEC a discussion of bottom sampling procedures (0381) is provided:
1. Knowledge of the composition of the seabed, the collection of samples and interpretation of the sidescan sonar trace are important tasks during surveys on the continental shelf.
2. In depths less than 200 metres, the nature of the
seabed is to be obtained as follows:
.....
c. at regular intervals throughout the survey area, in depths under 100m
at interval of 10 cms on paper at the rendering scale. In depths between
100m and 200m the contract may direct
that this interval is increased to 20 cms;
This requires a systematic grid-like bottom sampling campaign. While the interpretation of sidescan is recognised as an aid in this, it is not allowed as a substitute. No specifications about the calibration of the sidescan are demanded ensuring that only subjective interpretation of sidescan records can be performed. Obviously this specification is not currently written with the potential of the backscatter information available from MBES in mind.
4.2.4 - Delivery of Bathymetric Data Products
0408 THE STANDARD SHEET
1.1 The Standard
Sheet has replaced the more traditional Fair Sheets as the means of rendering
final drawings to
LINZ.
10.1 Depths are to be drawn
in Univers Sloping style "North up" with about four to the centimeter in
general
sounding and five to the
centimeter where examinations have taken place. Where sounding lines run
in a due
east/west direction depths
may be drawn west up.
11.1 The depth contours
should be traced on to the Standard Sheet from the sounding tracing if
no suitable
automatic contouring tool
is available.
11.4 The following contours
(none of which are to be labeled) are to be inserted:
2, 5, 10, 15, 20, 30, 50,
100, 200, 500, 1000m and then at 1000 metre intervals.
1. All digital data obtained
during the course of the survey is to be rendered. This includes, raw,
corrected and
processed data.
2. A digital file duplicating
the bathymetry on the standard sheet is to be forwarded in the formats
agreed with the
Contractor.
4. LINZ will issue
additional guidelines on the requirements for digital data in due course.
Clearly standard sheets consisting of dense posted numerical soundings together with contours is the main deliverable. While this consists of a substantially reduced data set, however, all digitial records (and therefore the full density of data) must also be submitted.
The only agency worldwide that has already undertaken contract hydrographic swath surveys using a publicly available set of guidelines is the US National Oceanic and Atmospheric Administration (NOAA). Herein a brief review of their approach is covered.
A Sample "Statement of Work" (SOW) can be found off the NOS web site. This SOW extends beyond those specifications that are being covered here. The critical components that are relevant include:
In verbal discussions with NOS staff it is clear that the SOW provided is only one example. NOS reserve the right to negotiate the exact specifications with each contractor for each particular contract. The one provided on the web site was based on those agreed upon for areas that would be considered Special Order or Order 1 in the Gulf of Mexico.
4.3.1 - Swath Bathymetric Sonar Standards
The method for defining alongtrack coverage is not dependent on 3dB coverage limits (as was imposed on the very first NOAA contract surveys). By defining merely the along track strike spacing, narrow transmit beamwidths are not unduly penalised (thus alllowing the narrow beamwidths which better meet the target definition criteria). Also the strikes per metre criterium is much easier to prove in the field.
The alongtrack coverage criterium within 30m requires a ping rate minimum of about half the speed (5Hz at 10 knots for example). This excludes the problem of yaw or pitch perturbations which would generally require high ping rates or reduced speeds. Based on observed ping rates for commercially available MBES (Hughes Clarke, 1997) this will limit the choice of sonar chosen.
Interestingly while full swath to swath overlap is required. Up to 100m is allowed between regions that meet the depth accuracy requirement. Thus having wider swath data that lie outside 4.3 is necessary but it need not meet any accuracy criteria. Only data that have proven capability to meet the depth error and resolution standards are of value. In discussions with contractors who are currently fulfilling these specifications, it is clear that these "reconnaisance" data are often not even edited for outliers.
The requirement for full bottom search with the swath bathymetry is limited to the reconnaisance level coverage and the ability to resolve 3m wavelength targets. This is because, 200% sidescan coverage is an additional requirement.
4.3.2 - Sidescan Sonar Standards
4.3.4 -
Delivery of Bathymetric Data Products
A Smooth Sheet (Sounding Plot with contours)
Note that, most recently, NOAA has released an online version
of the latest version of its Hydrographic manual. This can currently be
found at: http://chartmaker.ncd.noaa.gov/ocs/text/h-manual.htm.
A number of clarifications to the document discussed herein may be found
at that site.
A number of other agencies besides NOAA have used MBES in support of
nautical charting operations. Experiences from three of those agencies
are supplied below. These agencies were selected because they have had
long term exposure to the use of multibeam sonars for hydrography (Norway
since 1986, Canada since 1988 and Denmark since 1992). In none of
the listed cases are contractor surveys routinely sought however. In all
cases, their experience has been strongly influenced by the particular
sonars that they have operated. Thus their approaches should be seen in
the light of those instruments and used only as a broad guide for more
generic contractor specifications which must be hardware independent.
The following overview is based on a response to an email questionaire sent by the author:
The Norwegian Hydrographic Service have been routinely using multibeam data for nautical charting for 10 years. Up to 1998 the Simrad EM100 was the only sonar type used. In the 1999 field season, both EM3000 amd EM1002 are being introduced. For the EM100, it has been used to survey at depths ranges between 20 - 700 m.
The survey standard for multibeam sonar is to achieve full area bathymetric coverage on processed data. The overlap varies with the depth, but is normally 20% overlap to avoid problems with the outer beams. In uncharted areas, like Spitsbergen and in the Arctic, the overlap is 50 %, as there the outer beams are used as safety navigation control.
The latest IHO S44 changes have been incorporated into the national charting standards. The expected performance of the multibeam sonars used is: for EM100 and EM1002 Order 2, EM 3000 probably Order 1. Towed sidescan sonars are not used when collecting hydrographic data with multibeams due to the shape of the terrain in the Norwegian Coastal waters as this would be prohibitively expensive. A maximum speed of 6 knots is used for all hydrographic standard multibeam data collection.
For more information refer to:
Kjell Magne Olsen
tel. 51 85 87 00
Statens kartverk, Sjøkartverket
direkte 51 85 87 91
Leirviksveien 36
fax 51 85 87 01
Postboks 60
N-4001 Stavanger E-mail kjell.magne.olsen@sjo.statkart.no
http://www.statkart.no
The following comments are based purely on the authors experience working alongside the CHS, and thus should not be taken as the official CHS view.
The Canadian Hydrographic Service (CHS) has been using MBES in support of nautical charting surveys since 1988. The sonars used have been the Simrad EM100 (1988+), EM1000 (1992+) and EM3000 (1996+). The CHS does not usually contract out hydrographic surveys (the only exceptions are for the use of experimental technology such as airborne lidar and airborne electomagnetic induction methods). The multibeam sonars are being increasingly used to replace the old single beam charting methodology (which still continues in all regions at this point in time). Note that sidescan technology is not routinely used to support either SBES or MBES charting surveys.
At this time the CHS Standing Orders have not been updated to reflect the use of MBES. Prior to 1996, MBES data that was acquired for charting purposes was only used for issuing notices to mariners where anomalous shoals were identified. The CHS mandate, however, extends significantly beyond nautical charting (including cooperative surveys for science, offshore engineering, fisheries habitat, search and recovery and defense purposes). As a result, by far the majority of the CHS MBES surveying to date has been done for other non-nautical charting needs.
Depth Accuracy
The old IHO SP-44 Edition 3 standards remain in effect. These are applied for all sounding data used within the swath. Data outside these criteria (as determined from cross lines) should not be used for charting purposes (although this data is routinely used for engineering, scientific and other non charting purposes) .
Target Detection
Since the 1996 acquisition of the shallow water EM3000 sonars (4 operating
at this time) an increasing number of MBES surveys have been done in regions
of critical navigation where the new IHO target detection criteria might
apply. For all these operations, a common field policy has been implemented
(although not formally approved). This includes:
As part of cooperative surveys with other agencies (Geological Survey and the Navy) the target detection capablity of the sonar in this mode has been extensively tested (by comparing against deep towed scientific or mine hunting sidescan sonars).. Using the inner half of the swath, 2m boulders are seen to about 40m depth and 1m boulders to about 20m. Thus although not formalised in the CHS Standing Orders, the approach taken for the EM3000 sonars would appear to be acceptable to meet the target detection criteria for Order 1 surveys and perhaps Special Order surveys (but only in depths shallower than 20m).
The EM100 and EM1000 sonars, when deployed on nautical charting surveys are only used for the equivalent of Order 2 or 3 surveys. When those sonars are used for nautical charting work, 200 % coverage is required for both systems (equivalent to ~1x water depth for an EM100 and ~3x water depth for an EM1000).
Bottom Search
For all MBES operations, complete seafloor coverage defined by inter
swath overlap is required. As noted above, where the surveys are for nautical
charting purposes, at least 200% swath to swath overlap is required. As
the EM3000's are used at 12 knots, the EM100 at 10 knots and the EM1000
at 16 knots, in none of these cases would full alongtrack ensonification
be possible if one used the 3dB limits as the criterium. This is noted
in current operations, but is held subordinate to the target detection
criterium (which appears to be met using the EM3000, even though the 3dB
overlap clearly isn't).
Bathymetric Products.
For nautical chart production, conventional field sheet generation
continues. However as part of the quality control process and as a necessary
deliverable for all cooperative surveys, digital terrain models (DTM's)
are increasingly being created with resolutions appropriate to the sonar
beam footprints. These DTMs are routinely examined as sun-illuminated plots
to aid the hydrographer in quality control.
Thematic Data.
In recent years, with the loss of large platforms the bottom sampling
campaigns that used to be conducted in support of nautical charting have
been minimised. As part of the effort to compensate for this, for all MBES
surveys the bottom backscatter strength as recorded by the sonars is logged.
For those subset of cooperative surveys for which it is required, MBES
sidescan mosaicks are generated to aid in bottom classification.
The following overview is based on a response to an email questionaire sent by the author.:
The Royal Danish Administration of Navigation and Hydrography (RDANH) are responsible for nautical charting surveys in Danish waters. They have been using ELAC Bottomchart multibeam sounders in support of nautical charting operations since 1992. The RDANH do not yet contract surveys out to private organisations and have no plans in the near future to do so. The data they have collected has varied in quality over the years (reflecting the early evolution of swath sonar technology). For most of the early data, only that close to nadir was used for nautical charting. Since about 1996 a wider swath has been used.
At this time the BottomChart data is in general used only for Order 1 (SP44) survey. The BottomChart has been used in the depth interval 4 - 150 meters. They have only done one Special Order (SP44) survey so far and during that survey the speed was reduced to 4 knots and only measurements covering ± 45 degrees were used to provide the 100% bottom coverage. It is now stated that in general, at least 200 % independent towed sidescan sonar (SSS) data must be collected at the same time for both Special Order and Order 1 surveys. In Order 1 surveys the sidescan is only towed to 30 meters of water depth.
For any given order, using the error budget they calculate backwards in order to find the maximum error permissible for the multibeam, the error of which is estimated by the procedure in [J.Eeg: On the estimation of standard errors in multibeam sounding]. The goal is to get 5 - 10% overlaps between adjacent survey lines. If this goal is not achieved inter-lines are run in order to get 100 % coverage. At this point the latest IHO S44 changes have not been incorporated into the Danish national charting standards.
Minimum depths found with multibeam or SSS are to be verified with a singlebeam echosounder. Run up time before starting the data collection is minimum 1 minute due to HRP settlement. The weather conditions have to be very fair (wind < 7 meters/second and sea state < 0.2 meters). The maximum speed allowed for bathymetric data collection is stated to be 6 knots for Order 1 surveys for the BottomChart system and 4 knots for Special Order surveys for the BottomChart system.
More information may be obtained from:
Morten Sølvsten
Hydrographic Surveyor
Royal Danish Administration of Navigation and Hydrography
Ovengaden O.Vandet 62 B
P.O.Box 1919
DK-1023 Copenhagen K
E-MAIL: mns@fomfrv.dk
Phone: +45 32689609
Telefax:+45 32541012
S44 Edition 4 has provided a number of new criteria as well as redefining some existing ones . The most critical ones are:
5.1 - Interpretation of Coverage
Within the new S-44 Edition 4 document there is the requirement for "100% Bottom Search" for Special Order and Order 1 surveys. During the evolution of this document a number of terms appeared and dissapeared including "full bottom coverage" and 100% ensonification. This led to some uncertainty and a number of studys were performed (Miller et al., 1997) to quantify the difficulty in truely achieving these ambiguously defined criteria.
As adopted herein, four ways of defining coverage are differentiated:
5.1.1 - Swath to Swath overlap
This is seen as the degree to which swath corridor overlap. A swath
is defined as the corridor bounded by the outermost beams on either side
for a succession of sequential swaths. Note the outermost beams are considered
here which only excludes those editing out as grossly erroneously but notably
including those that may not meet the depth accuracy specification.
The swath may be further subdivided based on the predictions of sounding
uncertainty. It is proposed that LINZ encourage complete swath overlap
for all surveys. However, the amount of confidence
in the data will vary with the survey requirements. Using a model similar
to that of Hare et al., 1996, differing subsets
of the swath may be deemed to be within certain survey specifications.
5.1.2 - Ping to Ping Gap density
A much harder problem is that of along-track ping to ping gaps. Initially
this was not considered a criterium for coverage. But subsequent modelling
has shown that in typical seastates at speeds above 10 knots, almost all
of the available systems leave gaps in shallow water. The amount of alongtrack
coverage varies with a number of parameters:
The question of whether inter ping gaps can be tolerated is open to debate. Can small regions of the seafloor that have not been ensonified be tolerated? This of course depends on the survey requirements. Recognising that there are finite dimensions to the size of a feature that swath sonars can resolve, really the requirement is to ensure that the minimum size of target be confidently detected (the size being a separate issue). However, detection and coverage are separate. If you don't cover the target you can't detect it, but the inverse (if you do cover it you then automatically detect it) is not guaranteed. Does a single strike ensure detection? In the absence of convincing answers to these questions a series of guidelines are proposed to define how successfully and evenly the seafloor is searched.
Rather than attempt to prove or disprove complete seafloor ensonification,
it is proposed that LINZ follow the NOAA approach and use the minimum target
size specifications as a proxy for coverage requirements. In order to convincingly
define a short wavelength target bathymetrically it is necessary to provide
multiple strikes on that target. Herein it is proposed that at least three
strikes on the minimum target dimension be provide in both the along and
across track dimension.
One could ask the question: "Why three, why not one, two or four?". As will be shown below, this proposed sounding density will act as one of the most constraining (and thus potentially expensive) demands imposed by the new standards. The choice proposed herein is driven by three concerns:
More generally, if one is interested in describing seabed topography down to a lower horizontal wavelength cutoff (as effectively the target dimension is forcing us to do), the conventional Nyquist sampling theory would force us to sample at at least half wavelength or better spacing in order to adequately describe that feature. If we can increase this just slightly to one third wavelength or better, the feature definition, and our confidence is improved.
If we accept that we are limiting our search to features only above
the horizontal dimension of three beam spacings, this has the pleasant
side effect of allowing us to accelerate our data cleaning through the
use of more robust statistical methods.With an expected 5-9 strikes on
a target in total, one can eliminate those doubtful targets using the bathymetric
data alone without recourse to other, normally slower (and thus more expensive)
methods such as simultaneous sidescan search or shoal exams.
A more practical consideration that now has to be considered is how
does one define "three strikes". Do three entire beams have to be
within the target dimension? If so then beam width has to be considered
(and which level -3 or -6 dB?). In this case you would have to shoot alongtrack
(and acrosstrack) at, at least a third of the target dimension (even possibly
tighter for broader beam sonars). Or does just at least part of three beams
have to be within the target? In which case you only need to shoot
only slightly more than once over the distance of the target (assuming
your beam footprints overlap). Looking to the NOAA specifications, one
observes that they call for the vessel speed to be adjusted so that that
required number of beam footprints centre-to-centre fall
within the required dimension. If we adopt the phrase centre-to-centre
and stick with our three strikes criterium, it imples that the beam boresite
spacing must be no worse than half the target dimension. This interpretation
is recommended.
5.1.3 - The contribution made by overlapping swath corridors.
Overlapping swath data provide a valid means of contributing additional sounding density and thus may be used as a way to meet the coverage criteria. Note that the competing demands of a longer TWTT to get overlap, and the resulting lower ping rate, will generally cancel out the advantage for many instances. However, in critical survey areas, overlapping data provides particularily crucial supporting information as they represent a reimaging of the seafloor at both a different time and different aspect. In this way, anomalous soundings that might be a result of water column disturbances (fish, cetaceans, algae, bubbles, tidal fronts etc..) would unlikely be repeated and thus would be seen to be erroneous. It is proposed that for Special Order surveys, where MBES is being used without sidescan aiding, 200% bathymetric coverage be required.
5.1.4 - Occasional Data logging hiatuses
An operational reality of swath sonar data acquisition is occasional short periods during which data acquisition is corrupted for a number of reasons, some of which include:
5.1.5 - Coping with Cast Shadows
If at any point the slope of the seafloor in the plane of ensonification
is steeper than the ray path, then a shadow will be cast. This strictly
violates the condition of 100% Bottom Search. Should one then demand that
all shadows be imaged from the opposing direction? Assuming that the ray
paths are below the horizontal, it should theoretically be impossible to
shade a shoaller point from a source that lies above the seafloor. If the
target detection criteria are taken seriously, targets of 1 or 2m could
easily be lost in the shadows. Realistically, for nautical charting purposes,
if a target lies in a shadow, it cannot be the shoallest point. But for
other survey purposes (minehunting for example) it might perhaps be important
to examine everywhere. For most of the other wider range of LINZ
needs , targets lost in shadows would not be a critical loss. Where
however the shadow extends to the edge of the swath (and thus the bathymetric
swath draws in) the gap must be filled from the other side.
NOTE that for none of these coverage definitions is target discrimination guaranteed. The target may be guaranteed to have been ensonified, but, whether it will show up as a bathymetric anomaly depends on the scale and aspect of the target, the bottom detection algorithm and the size of the beam widths used. This effect is addressed in the next part.
5.2 - Interpretation of Feature Definition
With the acceptance of the new S44 standards, for the first time minimum target specifications have been quantified. These include:
For Special Order surveys, no depth is stated? Implicit in the description is that the areas that Special Order is called for are likely to be those that will be a challenge to shipping (<40m?). As the resolution of hull-mounted systems generally scale linearly with depth, meeting this specification at the ~40m limit might only be reasonable for a deeply towed sonars that maintains a constant elevation above the seabed. This is all further complicated by the fact that most sonars show a strong variation in target detection capability as a function of incidence angle. Thus the target detection criteria can limit usable swath width just as effectively as roll or refraction uncertainty.
It is herein proposed that for the purposes of LINZ interpretation, target detection may be achieved in one of two ways:
Should the contractor choose to use a sidescan as an aiding tool, the target detection critieria would apply to the sidescan (3 sidescan strikes in the equivalent target dimension). In this case the MBES sounding density can be relaxed for regular survey lines. The penalty for this approach, however, would be that any target not heighted confidently in the first pass of MBES would require a follow up shoal exam.
It is proposed that for Special and Order 1 surveys the 1 and 2m feature dimensions, already in use in the IHO definition, be used. Note that while detection of a "cube" is stated in the IHO document, this should more realistically be expressed as the ability to recognise 1 or 2m horizontal wavelength features. whatever the amplitude (to a lower limit of the depth accuracy specified below). Also in the IHO standard, Order 1 survey target resolution requirements suddenly drop to 10% from 5% below the 40m contour. For LINZ purposes it is proposed that the 5% criteria be maintained instead as there is no good reason why any particular sonars capability should degrade so suddenly (as the beam footprints scale linearly with depth). Should the drop in resolution be acceptable for the purposes of the survey, LINZ may state that the Order is to change from 1 to 2 for depths deeper than 40m.
While IHO Order 2 and 3 have no target or coverage requirement, for
LINZ purposes it is proposed that a 10% and 20% of depth wavelength target
detection criteria be imposed when MBES are used (and thereby implicitly
a coverage requirement is included also). This should not be terribly challenging
for most of the sonars on the market. Nevertheless it acts as strong constraint
on the coverage criteria (see above). Should
LINZ wish to relax the target criteria below a certain depth, a change
in order will be explicitly stated for that particular survey at that depth.
5.3 - Interpretation of Depth Accuracy Requirement
Given that we have imposed a high pass cutoff on the sonar spatial resolution, it now makes sense to consider the vertical accuracy specification to apply only to that longer wavelength part of the seabed roughness spectrum (in general > 20% of the water depth).
IHO S44 Edition 4 has 4 different requirements for vertical accuracy.
These successively degrade from Special Order to Order 3. As this sequence
of orders is designed to move from the critical inshore region to the less
crucial offshore deep regime (free of hazards to navigation) this might
seem reasonable. However, the ability of swath bathymetric sonars to achieve
vertical accuracy actually increases with depth. A number of things scale
quite linearly with water depth including bottom detection uncertainty
and orientation-error-induced depth uncertainty. However external factors
such as tide and heave are depth independent and thus become increasing
significant in shallow water. In addition water column stability is often
worst in the most critical areas (harbour entrances often are colocated
in river mouths). Accuracy as good as 0.2% have been reported for
deep water swath bathymetric data at high aspect ratios (<+-45 degrees).To
only demand 2.3 percent is simply to ignore the great potential of these
systems and is actually a step backward from the old S44 Edition 3 which
has always demanded 1% for all deep water measurements.
Another factor that should be noted is that the quality of the swath
sounding solutions will systematically and predictably vary as a function
of grazing angle. Whilst single beam sounders provided a consistent accuracy
level for a given depth range, the accuracy level of swath sonar
within a given depth range will normally degrade quite predictably as the
solution becomes more oblique. Exact performance specifications will,
of course, vary for any particular sonar type. A typical swath
sonar, however, should provide bottom detection uncertainty in the range
40-60% of IHO-SO specifications for the innermost beams, yet in the same
depth range, will exceed these specifications at incidence angles greater
than about 60 degrees. Bottom detection uncertainty is, of course only
one component of the total propagated error. Whilst tide, heave and draft
errors are common to the full swath, refraction and roll uncertainty again
grow non linearly with incidence angle. Thus whilst a given accuracy level
may be required for a particular survey., the reality of swath sonar surveying
is that, in order to meet that accuracy level for all solutions, the sonar
will, most likely actually significantly out perform that level for the
near nadir data..
Therefore in this specification, tailored to LINZ needs, it is proposed that the depth accuracy criteria be significantly altered. It is well recognised that swath sonar performance generally degrades as the solution become more oblique. In the most critical situations, the finest performance is demanded throughout the area covered and thus IHO Special Order depth accuracy requirements (IHO-SO-DA) are appropriate for the whole swath. But as the survey requirements are relaxed in deep water (or more generally in less critical areas), it makes sense to allow greater uncertainty which will creep in at the outer part of the swath (while continuing to be IHO-SO-DA level in the inner part). Herein, rather than a single accuracy criteria being specified, it is proposed that a percentage of the coverage is required to be to IHO-SO-DA standards (which may usually reflect the near nadir region, although not explicitly stated). But an increasing percentage of the swath will be allowed to have larger uncertainty (previously in the single beam situation there was total uncertainty as the inter line region were unsurveyed).
The following specifications are proposed (all quoted as multiples of
IHO-SO-DA):
|
|
|
|
|
| Special Order | 100% 1x | 1m or 2.5% | 3 strikes per 1m or 2.5% of water depth along track and across tack |
| Order 1 | 50% 1x 50% 1.5x | 2m or 5% | 3 strikes per 2m or 5% " " |
| Order 2 | 33% 1x 33% 1.5x 33% 2.0x | 4m or 10% | 3 strikes per 4m or 10% " " |
| Order 3 | 25% 1x 25% 1.5x 25% 2.0x 25% 2.5x | 8m or 20% | 3 strikes per 8m or 20% " " |
Thus one can see that IHO-SO-DA is demanded for at least a subset of all order surveys (operationally this will probably easiest be achieved by the near-nadir solution). But for higher orders, an increased tolerance is accepted for a growing percentage of the swath. Although not explicitly stated, this will probably mean that larger angular sectors will be employed. Note however, that LINZ does not specify the angular sector to use. Rather LINZ leaves that decision to the operator who will make a decision based upon a model of the sonar uncertainties, backed up by analysis of checkline data.
Whilst these depth accuracy specifications reflect the likely performance of the sonar, they are quite complex. Any such specifications will have to be demonstrated by the contractor to LINZ under operational conditions.As proposed so far, these specifications would be hard to practically implement and prove to the LINZ shipboard representative. In order to simplify the application of these specifications in the field it would be easier if a single number could be used. Toward this aim, two possible methodologies are proposed:
Method A: When the statistics of all sounding solutions that are deemed acceptable are combined, the average must meet the following specs:
NZ-SO
- 1.00 X IHO-SO
NZ-O1
- 1.27 X IHO-SO
NZ-O2
- 1.55 X IHO-SO
NZ-O3
- 1.84 X IHO-SO
Method B: A plot showing the uncertainty as a function of either
grazing angle or swath width must be provided and the outermost (or worst
sector) beams
should meet:
NZ-SO
- 1.00 X IHO-SO
NZ-O1
- 1.50 X IHO-SO
NZ-O2
- 2.00 X IHO-SO
NZ-O3
- 2.50 X IHO-SO
Because any final chart product will have to be presented with
a statement of the data uncertainty, Method B would represent the most
conservative case. If the numbers used for Method A were stated, they could
be misleading. The average mariner would assume that all data has that
uncertainty, (as would generally be true for single beam surveys). The
fact that the data uncertainty is incidence angle dependent would be both
hard to represent on the chart, and confusing to the navigator.
One extra point: if there is a periodic nature to the uncertainty (a
ribbing appears in the sun-illuminated swath data), indicating an imperfect
motion reduction. Even if the magnitude of the uncertainty is within these
specifications, all reasonable attempts should be made to isolate the cause
for the systematic dynamic error. This is because, even if within the DA
specifications, it is detrimental to the interpretation of seafloor morphology.
5.4 - Interpretation of Line Spacing
This report deals only with overlapping bathymetric swath corridors. The line spacing will depend on the angular sector used and the depth. The sectors are not specified by LINZ, only the required achievable depth accuracy and target detection capability within the sector delivered. Thus the concept of line spacing is not valid. The operator will choose an angular sector over which the data can be collected within the requirements demanded for that level of the survey. If the operator has a remarkable system that can provide both :
5.5 - Interpretation of Thematic Sampling Requirements.
The thematic sampling requirements within the existing IHO standards fall far short of the broader needs of LINZ. The ability of MBES (and related swath sonar technologies) to provide maps of the seabed backscatter intensity variations provides one, perhaps more costs effective, means of meeting these needs. Thus rather than propose a modified dense sampling program based on conventional seabed sampling methods, a set of requirements are proposed for the acquisition of seabed backscatter intensity information. These are outlined in the following chapter.
6.0 - Proposed LINZ Backscatter Imaging Requirements
6.1 - What, rather than where is the seafloor
To meet the needs of a significant subset of LINZ surveys an estimate
of the composition of the seafloor is particularily useful. Traditional
bottom sampling methods could be employed to meet these needs. Such approaches
however are very time consuming and expensive. As backscatter data is acquired
from almost all swath sonars without added hardware requirements, it would
be of advantage to LINZ to ensure that these data are retained in a useable
form to aid in assessing the composition of the seafloor. Recognising that
swath sonars operate with a variety of methods to extract backscatter data,
this proposed requirement aims to be sufficiently generic that calibrated
backscatter data may be delivered from most commercially available swath
sonars.
At this point in time, whilst seabed backscatter strength is used as
an aid in seabed classification, it is not a proven robust tool for use
in isolation. IHO S44 requires that "any inference technique should
be ground-truthed by physical sampling" reflecting this reality. However,
if a judicious sampling program is used based on interpretation of the
regional pattern of seabed backscatter strength variations (acquired through
the use of MBES) rather than sticking with a dense regular grid of physical
samples, less sampling should be necessary (with a resulting cost saving
to the Crown). It is thus proposed that LINZ require the collection and
delivery of seabed backscatter information with the following specifications.
6.2 - Methods for acquiring seabed backscatter data
All sonars are required to log the received backscatter intensity. No restriction is based on the method in which this is derived. For example:
6.3 - Backscatter data reduction
Conventional sidescan sonar imagery has been used as a qualitative tool for seabed classification for many years. The appearance of such data however, is commonly quite unique to a specific hardware set up. The received intensity (and its graphical representation as a greyscale) is not usually an inherent property of the seafloor. Changes in the source level, receiver gains, towfish aspect ratio, pulse length etc. invariably alter the gross appearance of the sidescan image. Indeed for the purpose of optimal target recognition all of these parameters are routinely altered as a function of time to optimise the contrast of the image.
Such unreduced sidescan images, while useful for target recognition and local geologic interpretation, are of little or no value for assessing regional trends in the the composition of the seabed. In order to be able to use these estimates of received backscatter in a more quantitative manner, they need to be routinely reduced to an inherent property of the seafloor. Such an estimate is the seabed backscatter strength. This dimensionless number represents the ratio of the backscatter intensity as a function of the incident power per unit area of the seafloor. This is then an inherent property of the seafloor (a function only of the frequency, the grazing angle and the azimuth). In order to make such measurements, however, the logged data must be properly reduced for the following:
6.4 - Backscatter Strength Estimate Repeatability
At this time, as there are few affordable means for confirming the absolute
estimate of local seabed backscatter strength estimate, the emphasis
is placed on repeatability. That is to say the sonar must be demonstrated
to provide a repeatable estimate for the same region, for the same grazing
angle within 2dB over the time duration of a survey. Such repeatability
may be compromised at any time by changes in hardware. Thus any time a
transmitter or receiver board is changed, or a software algorithm is updated,
this must be logged in a digital record provided to LINZ. Where possible,
a small subset of the data, before and after the change should be repeated
and logged as a separate repeatability test.. If multiple instruments
are used either simultaneously, or over a period of surveys, then inter-instrument
repeatability tests need to be done and estimates made of their differences
and all intercalibration information be made available to LINZ.
6.5 - Limitations imposed by aeration
One of the major operational concerns is the drop in signal level
recorded in marginal seastates when bubble wash down start to blanket the
arrays. If significant data striping or drops out occur in the received
backscatter images, then the data will be considered unacceptable.
Defining the onset of "significant" is going to be a tricky issue and at
this point, until a better means can be found, it is proposed that this
assessment be made qualitatively by the LINZ shipboard representative.
As a result, contractors are encouraged to use installations that minimise
these problems. Interestingly backscatter attenuation due to bubble effect
is normally a very sensitive indicator of imminent degradation of the bathymetric
data and thus both data logging streams should be terminated.
The original IHO specifications were restricted by water depth range, reflecting the declining requirement for accurate bathymetry as the seabed deepened . Herein, the proposed specifications are expanded to allow each order to cover the full range of depths should the broader needs of LINZ require it. This reflects the fact that there may be applications for which more demanding bathymetric criteria are needed even if the region does not present a potential hazard to surface navigation. Examples of this might include:
In order to facilitate the interpretation of the IHO guidelines
for the purposes of bathymetric swath sonar surveys commissioned
by LINZ, the following summaries are presented:
7.1 - NZ-Special Order
While originally designed for <40m, specifications are extended to include all water depths encountered. Recognising that some contractors may prefer to meet the target detection criteria through the use of sidescan sonar rather than rely on bathymetric detection alone, two options are proposed.As can be seen, Option 1 closely follows the CHS or Norwegian approach whereas Option 2 follows the NOAA or Danish approach.EITHER: (Option 1: target detection through bathymetry)
- 1m horizontal wavelength features to 40m 2.5% of depth below that using :
OR: (Option 2: target detection through sidescan)
- 100 % bathymetry interswath coverage at 1x SO vertical accuracy specs.
- 100% bathymetric interswath coverage at up to 1.5x SO vertical accuracy spec.
(imples the need for 200% coverage bathymetry)and for both
- 200% towed sidescan imaging with
- 50 % bathymetry interswath coverage at 1x SO vertical accuracy specs.
- 50% bathymetric interswath coverage at up to 1.5x SO vertical accuracy spec.
- (imples the need for only 100% coverage bathymetry)
- Shoal exams on all targets recognised in sidescan but not bathymetry.
- calibrated backscatter imaging required to 100% interswath.
7.3 - Order NZ-2
- 2m horizontal wavelength features to 40m, 5% of water depth features below that.
- 50 % bathymetry interswath coverage at 1x SO vertical accuracy specs.
- 50% bathymetric interswath coverage at up to 1.5x SO vertical accuracy spec.
- calibrated backscatter imaging required to 100% interswath.
- 4m horiziontal wavelength features to 40m. 10% of water depth below that.
(implies most natural shoals seen but that anthropogenic targets may be imperfectly resolved)- 33 % bathymetry interswath coverage at 1x SO vertical accuracy specs.
- 33% bathymetric interswath coverage at up to 1.5x SO vertical accuracy spec.
- 33% bathymetric interswath coverage at up to 2.0xSO vertical accuracy spec.
- calibrated backscatter imaging required to 100% interswath
7.4 - Order NZ-3
While traditionally only applied to data deeper than 200m, specifications are clarified to include non critical shoaler areas that are discovered in the area.
- 8m horizontal wavelength features to 40m 20% of water depth below that
- 25 % bathymetry interswath coverage at 1x SO vertical accuracy specs.
- 25% bathymetric interswath coverage at up to 1.5x SO vertical accuracy spec.
- 25% bathymetric interswath coverage at up to 2.0x SO vertical accuracy spec.
- 25% bathymetric interswath coverage at up to 2.5x SO vertical accuracy spec.
- calibrated backscatter imaging required to 100% interswath
7.5 - Technological Implications of NZ Orders
It is recognised that these proposed MBES specifications will strongly influence the choice of sonar and ancillary sensors that can be used. Much work has already been done on estimating the contribution of the various error sources on vertical depth measurement accuracy (e.g. Hare et al., 1996). These type of studies, however, only address the longer wavelength bathymetric features. Most sonar manufacturers would claim that their systems today would meet IHO special order accuracies for long wavelength targets over swaths of at least 120 degrees and arguably up to 150 degrees (with careful alignment, high quality motion sensors and sufficient water column knowledge). The constraint of target detection, and coverage, however, turns out to be far more demanding.
In these proposed specifications, it is required that sounding density be sufficient to provide at least three pings along track and 3 beams across track within the dimension of the target for any particular Order. As currently, all swath bathymetric sonars ping rates are constrained to transmit only after the last echo of the previous ping has returned, the three strikes criteria places strict limitations on vessel speeds and swath widths for any given order. To illustrate this, an example set of calculations are presented. In this example we use a vessel speed of 10 knots and assume that the sonar pings at a rate of 1.2 times the two way travel time to the outermost beam (the 20% wait is to allow for the internal processing time). Results are calculated showing :
* Note that swath angular sectors over 150 are trimmed due to
other constraining factors (roll accuracy and refraction uncertainty).
Also note that the degrading effect of yaw instability on along track density
is left out (which would require either active yaw stabilisation or even
higher ping rates to compensate).
| Operating Conditions: | ||||||
| speed | 5 | m/s | 10 | knots | ||
| ping rate efficiency | 1.2 | (of minimum TWTT) | ||||
| alongtrack strikes | 3 | |||||
| acrosstrack strikes | 3 | |||||
| Order name | Special | Order 1 | Order 2 | Order 3 | ||
| target size (% of depth > 40m) | 2.5 | 5 | 10 | 20 | ||
| depth (m) | 10 | target size (m) | 1 | 2 | 4 | 8 |
| required ping rate | 10 | 5 | 2.5 | 1.3 | ||
| angular sector (deg) | 150 | 150 | 150 | 150 | ||
| swath width (m) | 75 | 75 | 75 | 75 | ||
| min # beams | 149 | 75 | 37 | 19 | ||
| depth (m) | 20 | target size (m) | 1 | 2 | 4 | 8 |
| required ping rate | 10 | 5 | 2.5 | 1.3 | ||
| angular sector (deg) | 149 | 150 | 150 | 150 | ||
| swath width (m) | 145 | 149 | 149 | 149 | ||
| min # beams | 289 | 149 | 75 | 37 | ||
| depth (m) | 40 | target size (m) | 1 | 2 | 4 | 8 |
| required ping rate | 10.0 | 5.0 | 2.5 | 1.3 | ||
| angular sector (deg) | 116 | 149 | 150 | 150 | ||
| swath width (m) | 127 | 289 | 299 | 299 | ||
| min # beams | 254 | 289 | 149 | 75 | ||
| depth (m) | 80 | target size (m) | 2 | 4 | 8 | 16 |
| required ping rate | 5.0 | 2.5 | 1.3 | 0.6 | ||
| angular sector (deg) | 116 | 149 | 150 | 150 | ||
| swath width (m) | 254 | 578 | 597 | 597 | ||
| min # beams | 254 | 289 | 149 | 75 | ||
| depth (m) | 160 | target size (m) | 4 | 8 | 16 | 32 |
| required ping rate | 2.5 | 1.3 | 0.6 | 0.3 | ||
| angular sector (deg) | 116 | 149 | 150 | 150 | ||
| swath width (m) | 508 | 1157 | 1194 | 1194 | ||
| min # beams | 254 | 289 | 149 | 75 |
8.1 -
Requirements for Delivery of Bathymetric Data Products
It is proposed that all data collected onboard the chartered vessel be considered the property of LINZ and should be provided in a specified manner. These specifications apply to the swath sonar and peripheral data. For all surveys it is proposed that the following be required:
8.1.1 - "raw data"
In general this will include all the output of the sonar during acquisition.
This data may be provided in a hardware specific format on the strict
condition that full format specifiers together with a simple
sample reader program is provided. This format description and the sample
reader program will be freely distributed by LINZ to all parties wishing
to provide value added products to the data. This data is essentially
unedited (all invalid or corrupt data acquisition need not be provided,
only that used to derive the next two levels of product are essential).
This data will also include ancillary information not already integrated
into the sonar data stream including full orientation and position
data (if not already included), SVP information, tidal information and
the integration parameters used (installation offsets, misalignments angles
and clock time differences).
8.1.2 - "processed
full density data"
As part of the field processing sequence all the raw data, including
sonar and ancillary integrated sensors should be examined for blunders.
At the conclusion of this process, the data should be provided in a cleaned
fully integrated form (reduced for position, elevation, orientation
and water column). All seafloor soundings and thematic solutions should
be provided although quality flags should be added to all data that indicate
whether the data has been rejected or deemed beyond deliverable survey
specifications. Note that the imaging geometry must still be retained in
this format which would include, for example, the ship relative geometry
and original imaging angles and two-way travel times.
As with raw data, no exact format specifications are enforced. Again,
however, is is necessary that the full format specifiers together with
a source code level library for reading the data be provided. At a later
time LINZ reserves the right to specify a format that must be generated
via a provided filter.
Depending on the type of survey specific reduced data sets will be required. In general a final hard copy map product will be delivered (for example fieldsheets for nautical charting surveys). For the case of the Standard sheet, specifications will remain similar to that already existing in the LINZ HYSPEC. Following the NOAA model, however, these data will be delivered in a digital form that includes full traceability for every selected sounding retained.
Standard Sheets are designed to be a legal document for use in the compilation
of nautical charts (on which only contours and the apex of shoals will
normally be represented). However, field sheets produced at common scales
provide can exhibit only a fraction of the data (usually selected on the
basis of a shoal biased algorithm) and thus other means are now proposed
to better provide the full information content on the MBES data.
8.1.4 - Seabed Elevation Data Derivatives
With the higher data density implicit in bathymetric swath sonars and the provision of thematic information a variety of more intuitive plots will also need to be provided. Standard plotting sheets only allow an interpretable data density of a little larger than the font size (resulting in ~4-5mm spacing of solutions on average). At the same scale, if the solutions are presented as pixels in an image one can have interpretable relief at resolutions of better than 100 per inch. Useful images can thus be generated that would reveal both bathymetric data quality and seabed geomorphology.
To be most useful, image resolution should approach the size of the smallest resolvable target (in general, for a given sonar, this will be a specific percentage of the water depth). To reveal the morphology, while suppressing the uncorrelated sounding noise, mean surfaces will be essential rather than the shoal biased approach used for nautical chart derivatives. Mean surfaces derived by gridding at the appropriate resolution allows the user to rapidly visualise the the information content (and artifacts) within the resulting data. It is thus proposed that LINZ require that mean surface grids be provided as a digital deliverable. To aid in rapid assessment of the deliverable, a number of hard copy images (with digital originals) would be required including:.
- colour coded depth images
- 2 orthogonally sun-illuminated plots.
(provides
the means to rapidly QA the data)
At this point, no constraint is placed on the format of the derived
gridded, fully georeferenced products as long as full format specifiers
and sample code for reading the data is provided. In the future, though,
LINZ reserve the right to specify a particular format as a deliverable.
In all cases, a common image format (such as a tif or gif) should be used
to provide quick look images showing the correct look of the data.
8.2 - Delivery of Seabed Backscatter Data
8.2.1 Calibrated backscatter strength estimates.
For all data, for every ping, reduced estimates of the seabed backscatter strength should be provided at at least 2 degree intervals over the angular sector employed. This may be achieved by reducing the average response from beams spaced at these intervals or by extraction of the equivalent time interval from the sidescan time series to the angular section desired. The data reduction described in part 6.0 should be used with all assumptions stated. The data should be provided in a fully documented digital format with sample reader software to allow development of custom translators for future client needs. The main use for these estimates is envisaged to be seabed classification.
8.2.2 - Strip plots/images of received backscatter strength as a function of across track distance and time.
The aim here is to provide a digital equivalent of a traditional sidescan paper roll. These digital plots should try to preserve the highest justified across track resolution achieved (for the purpose of target recognition). A maximum of 1024 pixel however per side need be retained. Note that this requires that the sonar provide a sampling interval that, in general, will be higher than the beam spacing and thus the older, one solution per beam methodology provided by some sonars will not be acceptable.
In this case we are attempting to provide a map of the targets, superimposed on the regional seabed backscatter strength variation. To that end, methods for minimising the effect of grazing angle are encouraged. In general this will involve a gross compensation for the dropoff in mean backscatter strength with grazing angle. The aim being to provide a "cosmetically" pleasing produce that is faithful to the regional trends of sediment (not unduly biased by the continuously changing grazing angle) while preserving the detail available in the data. The method used to achieve this should be fully documented so that the assumptions used are know to subsequent users during their interpretation of these products.
8.2.3 - Regional maps of mean backscatter stength (grazing angle compensated)
Using the products outlined in 8.2.2, a third product required is a
geographically registered mosaick of the seabed backscatter estimates.
This image needs to be provided at a resolution appropriate to the time/angle
sampling capability of the system. As a guide, however, the larger of the
pulse width or the along track beam footprint should be used as a reasonable
resolution. Thus in general, the target detection capability of this presentation
will be reduced over that of the prior product. As this product will generally
be used for qualitative inference of the distribution of surficial sediments
it is preferrred that the variation in backscatter strength, due to changing
grazing angles, be minimised. All the products described must be
provided as digital images in formats that are fully described and easily
to translate into common derivatives.
Hare,R., Godin,A. and Mayer,L., 1995, Accuracy estimation of Canadian Swath (multibeam) and Sweep (multi-transducer) sounding systems: Canadian Hydrographic Service, internal report.95pp.
Gardner,J.V., Butman,P.B., Mayer,L.A. and Hughes Clarke,J.E., 1998, Mapping U.S. Continental Shelves: Sea Technology, v.39, no.6, p.10-18.
Hughes Clarke J.E., 1997, A comparison
of swath sonar systems demonstrated at the 1997 US/Canada Hydrographic
Commission Coastal Multibeam Surveying Course: http://www.omg.unb.ca/~jhc/uschc97/
and
Hughes Clarke J.E., 1996, A comparison of swath sonar systems
demonstrated at the 1996 US/Canada Hydrographic Commission Coastal Multibeam
Surveying Course: http://www.omg.unb.ca/~jhc/uschc96/
Hughes Clarke, J.E., 1998, The effect of fine scale seabed morphology and texture on the fidelity of swath bathymetric sounding data: Proceedings Canadian Hydrographic Conference 1998, Victoria, p. 168-181.
IHO, 1987, International Hydrographic Bureau, Monaco, Accuracy Standards, Special Publication #44, 3rd Edition.
IHO 1997, International Hydrographic Organisation, Standards for Hydrographic Surveys, Special Publication #44, 4th Edition, Draft after 2nd meeting of S-44 WG.
Miller, J., Hughes Clarke, J.E.
and Patterson, J., 1997, How Effectively Have You Covered Your Bottom?:
Hydrographic Journal, no.83, p.3-10. and see technical web page at:
http://www.omg.unb.ca/~jhc/coverage_paper.html