IHO SP 44 Standards for Hydrographic Surveys and the demands of the
new century.
Dave Monahan, CHS, and D E Wells, Ocean Mapping Group, UNB
Keeping any standard up to date is a never-ending task. For example, rapid advances in
multibeam technology and increasing interest in deeper areas of the ocean driven by UNCLOS
mean that IHO SP 44, Standards for Hydrographic Surveys, 4th Edition, needs to be reviewed
and expanded. SP44 provides 4 different requirements for vertical accuracy, successively
degrading from the critical to navigation inshore region to the offshore deep regime. Requiring
lower standards in deeper water has some justification for navigational charting, yet ignores the
needs of a wide spectrum of other ocean users, and ignores the standard that MBES are capable
of achieving. Article 76 of UNCLOS, for example, requires the accurate location of the 2500m
contour, a location which the current version of SP44 allows hundreds of meters to kilometres of
horizontal uncertainty. Furthermore, the ability of swath bathymetric sonars to achieve vertical
accuracy actually increases with depth. For these reasons, it is proposed that SP44 be revised or
expanded to include those cases where accuracies similar to those in shallow water are required
well below navigation depths
Introduction
Establishing standards in any field is never easy. Establishing standards that
must be agreed to by 70 nations is difficult. Establishing standards in a field that
is undergoing rapid technological transformation is a constant game of catch-up.
Doing all these while establishing standards for measuring something that can
not be seen and in which the measurement that the standard is to be applied to
in fact generates the standard is - hydrography.
The International Hydrographic Organization (IHO) has issued standards for the
accuracy of echo sounding since 19xx. (2nd edition 1982, 3rd 1987, 4th 1998)
under the identifier SP44 or, more recently, S44. The preface to the 3rd edition
tacitly recognizes the rapidity of change with a brief statement that SP 44 will be
updated every five years, a worthy goal that has proven impossible to achieve. It
is very instructional to compare the 3rd and 4th edition, and try to provide
guidance on how the 5th edition can be improved.
Although the standard was produced by the IHO for use by hydrographic surveys
in support of navigation, since it is the only International standard for survey
accuracy and carries the imprimatur of a highly respected International body, it is
referred to or cited by a number of users who are not concerned with navigation.
Description of the standard
The third edition is entitled “IHO Standards for Hydrographic Surveys,
Classification Criteria for Deep Sea Soundings and Procedures for Elimination of
Doubtful Data”, treating these as three separate subjects. The fourth edition
spends one page on doubtful data and combines the first two, hydrography and
deep sea sounding into one table. The combination of two quite different
requirements is not necessarily an improvement and should be examined more
1
17/04/00
Dave Monahan
closely. In its Preface, the 4th edition states that it is an attempt to broaden its
outlook to meet the needs of “a much more diverse group than previously recognized.
Hydrographic data is also important for Coastal Zone Management, environmental modeling,
resource development, legal and jurisdictional issues, ocean and meteorological modeling, and
engineering and construction planning, and many other uses.” Most would agree with this
as reflecting the actual conditions in most offshore areas. However, the
Introduction on the next page seemingly contradicts this by stating that the 4th
edition specifies “ different accuracy requirements for different areas according to their
importance for the safety of navigation”. We examine this further below under “bias”.
Rather than dividing areas to be surveyed into the 3rd edition’s two zones of less
than 200m depth and greater than 200m depth, the 4th edition provides for 4
zones:
Special Order - "specific critical areas with minimum underkeel clearance
and where bottom characteristics are potentially hazardous to vessels"
(generally less than 40m)
Order 1 - "harbours, harbour approach channels, recommended tracks,
inland navigation channels, and coastal areas of high commercial traffic
density" (less than 100m)
Order 2 - "areas with depths less than 200m not covered by Special Order
and Order 1 "
Order 3 - "areas not covered by Special Order, and Orders 1 and 2 and in
water depths in excess of 200m".
For each of these it specifies Horizontal Accuracy, Depth Accuracy, 100%
Bottom Search, System Detection Capability and Maximum Line Spacing. It does
not address how these are to be combined. (include Table 1 from SP44, 4edition)
In this paper we focus on the issue of depth accuracy, leaving an examination of
bottom search for another report (jhc in progress).
Bias
All measurements are subject to bias, some controllable, some a function of the
equipment in use at the time. Standards can seek to eliminate these biases or to
enshrine them, and the wise users of standards seek to understand which of
these alternatives have been followed. The controllable safety-of-navigation bias
is expressed as shoal-biasing every uncertainty source for safety’s sake, but that
does not necessarily serve the best interests of the non-navigational users of
hydrographic data specified in the introduction to edition 4. An example of
equipment-dependant bias could be the in older-generation radiopositioning
systems, in which measuring the risetime on a noisy, stretched-out pulse will
2
17/04/00
Dave Monahan
always result in a too-long distance measurement, presumably placing all depths
it measured further from the control point.
Not all uncertainties are biased, of course. For example, errors caused by
beamwidth or sound speed error effects are as likely to occur one way as the
other.
The standard does not address the issue of bias. Presumably, it is left to the
hydrographer to work around the biases and still meet the requirements of the
standard.
SP44 standards are “a priori” depth error requirements
The standards define the accuracy that the various elements of a survey are
required to meet. They offer no instruction as to how one would assess whether
these requirements have been met, nor what to do if a survey falls short in one
category or in all of them.
Conventionally, there are two general approaches to error estimation, a priori and
a posteriori. Before a survey is conducted, an application of the a priori approach
tries to assess how well each piece of data could be collected, based on a
theoretical appreciation of the equipment and methods used, the geometry and
geomorphology of the area, the physical characteristics of the water mass, and
other practical considerations. After conducting such an analysis, it might be
concluded that the proposed survey would not meet the standard, leading to a
change in the plan, and a reassessment to determine wither the new plan would
meet the requirements of the standard. It might also be determined that the
proposed survey could exceed the requirements if conducted as planned,
possibility leading to a modification of the survey. Used this way, SP 44 provides
a useful measure against which planned surveys can be compared.
Once results are in, the a posteriori approach to error estimation attempts to
determine what accuracy really was achieved. The data are examined and tested
resulting in an indication of how well they behave. How closely the real data will
approach or exceed the standard will depend the detection and elimination of
blunders and systematic biases and short lived anomalies in the water (unusually
high water level, for example) on the factors included in the a priori approach
plus some new factors introduced by the real world. The biggest of these is the
slope and roughness of the bottom. Achieving agreement between checklines is
extremely difficult over flat seafloors, and over extremely rough seafloors. Of
course, in a previously un-surveyed area, these can be extremely difficult or
impossible to predict, and that may be reason enough to not try to include them
in an a prioiri error estimation. However, once a survey has been conducted, they
can strongly influence the results, and there should be a feed-back loop that
would incorporate the slopes detected. It is possible that two surveys, identical in
all respects except the roughness of the seafloor beneath them, would be
3
17/04/00
Dave Monahan
classified into different orders. SP44 Edition 4 does not instruct that any
assessment be carried out. (Edition 3 did so to some extent, through the
requirement to run checklines and to examine the results where checklines
disagreed by more than twice the accuracy requirement (.3m to 30 m. 1% deeper
than 30).) However, can it be claimed that a survey “met” the standard if no post
survey check was carried out to verify that it did? It could of course be claimed
that surveys were planned to meet the standard, but planning and realization are
not always the same thing.
This begs the question of what happens if a post-survey analysis reveals that the
survey has not achieved the planned standard. Can the survey be rated at a
lower order? Can parts of a survey be rated at one order, other parts at another
order?
Application to measurements
Standards are something to which real measurements are compared. Preparing
standards for hydrographic data accuracy is complicated by the fact that each
sounding represents the sum of several measurements, each containing errors.
How these interact and accumulate is critical, yet not always solvable. How to
determine whether a sounding in the middle of a bay is within some specified
accuracy is close to impossible. SP44 has taken the approach of separating the
measurements that make up a sounding and classifying them individually. While
this does produce a standard, it leaves many questions unanswered when trying
to evaluate real-world results. Does one component have precedence over
another, or does a survey fail to meet standards if it does not meet the criteria in
one component or in all of them?
Measurement theory indicates and good survey practice demands that surveyors
attempt to minimize errors by
a) repeating each measurement several times with the expectation that
errors will cancel one another (sometimes we even discard maximum
and minimum measurements completely) and
b) designing the sequence of the series of measurements in such a way
that some measurements can be compared to other measurements
taken at a different time in the sequence (for example, leveling circuits
must close).
In hydrographic surveying, these principles are followed when establishing
vertical control and position fixing; sounding, of course, presents particular
logistic problems which militate against rigorously applying the above principles.
To repeat each sounding, even once would reduce speed to a small fraction and
magnify cost exorbitantly, a totally unacceptable state of affairs considering the
size of the task confronting us. However, condition b) can be satisfied to some
extent by running check lines approximately perpendicular to the main survey
lines. At the intersection of two checklines, the hydrographer was attempting to
repeat both depth and position measurements. That is, it was thought that the
4
17/04/00
Dave Monahan
vessel was at the same location as previously, and were that true, the water
depths observed should be the same. Edition 3 of SP 44 recognizes this and
states that intersections at which the difference in depth exceeds twice the
specifications should be investigated. Edition 4 states only that crosslines should
be run, but gives no instruction on how to use them.
A closer look
Balance in importance between fixed and variable errors
SP 44 correctly divides errors into two contributing types, fixed and variable.
Constant errors are those that have the same value no matter the water depth,
while variable (depth-dependent) errors are a fixed percentage of water depth
and thus grow larger (vary) with deepening water. Each type of error is summed
with other errors of the same type. The two types are the combined as RSS, that
is the square root of ((Fixed errors) squared + (variable error multiplied by depth)
squared). RSS is a very powerful tool, with a few important characteristics. One
is the way the relative sizes of the numbers being summed interact with one
another. When they are similar, they each have an effect, but when one is
significantly larger than the other, the larger one dominates to the extent that the
contribution of the smaller diminishes to the point of being negligible. This
captures the real world, and should be carefully allowed for in any standard.
It is interesting to examine these numbers as currently expressed in Edition 4,
and to compare them with Edition 3. The third edition specified that for depths
less than 30m, the error could be 0.3m, and for deeper depths it could not
exceed 1 % of depth. The 4th Edition’s Special Order, the fixed error of 0.25m is
dominant to a depth of about 34m. At 40 m, which is the max depth for Special
Order, total error = .39m, slightly more stringent than edition 3. . Moving to
deeper water and relaxing the standard, for Order 1 the fixed error of 0.5m is
dominant to a depth of c38m, at which point it and the variable error make the
same contribution to the total. At 100m, the max depth for Order 2, the total error
= 1.4m, not as tight as the 1m demanded by the 3rd edition.. The remaining
orders, 2 and 3, share the same errors, a fixed error of 1m and a variable error of
.023 x water depth. At about a depth of c43m fixed and variable make equal
contributions, but after that the variable error quickly comes to the fore. At 200m,
the max depth for Order 3, the total error is 4.7m while at 2500m, the depth
important for UNCLOS, error is 57.5 m, to which the fixed error contributes a
mere 8mm! Errors allowed in these zones are more than twice (2.3) those that
were allowed under edition 3.
Much more interesting in this table is the observation that required accuracy
actually increases with increasing water depth. This seems to contradict the
overall intent of sp44 Table 1, wherein Horizontal Accuracy, Depth Accuracy,
100% Bottom Search, System Detection Capability and Maximum Line Spacing
5
17/04/00
Dave Monahan
are all arranged to lessen the accuracy required as the bottom becomes less
likely to interfere with navigation.
Depth Accuracy as expressed in Table 1 must be applied with care. Although
the Table lists values for fixed and variable accuracies, they must be combined
and the result is the depth accuracy. It would seem logical that where one is so
small in relation to the other that its effect is negligible, that the smaller one be
allowed to exceed the values in the Table. Otherwise, a data set could be failed
for missing the fixed error target by a few mm, yet that miss would be totally
meaningless in terms of overall accuracy.
Beam width errors
Sound energy emitted from an echosounder transducer face spreads as it travels
through the water column, and this spreading effects the manner in which the
returning signal describes the seafloor. These effects included the introduction of
horizontal displacement when the seafloor is sloping, the smoothing of the shape
of large features and the obscuring of features whose wavelengths are less than
twice the ensonified area. Great efforts on the part of sonar engineers have
produced transducers and arrays that control this spreading and allow the
focussing of the sound energy. In fact, the history of sonar development is one of
continual progress towards finer and finer resolution. Cone widths of 2 or 3
degrees, as used in MBES, resolve a great many of the issues that plagued deep
sea sounding in the era when cone widths were a order of magnitude greater.
Perhaps because of this progress, and because it does claim to be a standard for
future data collection, SP44 make no mention of cone effect.
Unfortunately, not all surveys in the immediate future will be MBES based, and
sounders with wide cone widths will continue to be used for some time to come.
It is instructive to examine the limitations indirectly imposed on beam angle by
the specifications. Table x shows that, if the fixed error is taken as the maximum
for each order, and that error due to beam spreading is considered to be the only
source of variable error, then for Special Order, the maximum allowed beam
angle is 14 degrees, for Order 1, 10 degrees, and for Orders 2 and 3, the
maximum allowed beam angle is 24.6 degrees. Survey plans must consequently
incorporate sounders with appropriate beam widths.
Sampling along Tracks or Lines
Edition 4 includes “Maximum Line Spacing” as a category to be applied in the
standard. Special Order is excluded since 100% bottom search is compulsory,
presumably meaning that the line spacing will be dictated by the characteristics
of the sonar system used. For the numbered Orders, line spacing is given as
function of depth. (The old hobgoblin yet again) However, it says nothing about
the number of soundings required along track. Soundings could be spaced to
match the line spacing, spaced at some arbitrary distance possibly related to
6
17/04/00
Dave Monahan
sizes of figure at plotting scale, or spaced to capture the bottom profile. The latter
would include:
a)
Preservation of small features;
b)
Preservation of significant corner angles
c)
Preservation of any included frequencies;
d)
Preservation of the location of significant points; and
e)
Absence of features not contained in the original seafloor.
Capturing the bottom profile in this way has many advantages, but the 4th edition
makes no reference to any form of along track sounding spacing
Beyond depth – the bathymetry model
Another section of SP44 that addresses errors is the section “Bathymetric Model
Accuracy”. Despite the somewhat obscure wording of SP44 on this point, the
term “Bathymetric Model” can be interpreted to mean contours derived from the
measured depths. In other words, the seafloor is measured at points with error
parameters as described in SP 44, Table 1, while between the measured points
the seafloor is modeled by contours. Logically, these contours cannot have as
high an accuracy as do the measured points, and Table 3 lists constants for the
fixed and variable terms in the formula used to determine error values of the unmeasured depths. These produce a vertical error 2 to 3 times tat at the measured
locations. SP44 does not indicate how this is to be combined with the results
from Table 1. Note that there is no weighting for the distance from the measured
points, nor the measured points’ isolation or clustering. SP44 appears primitive in
comparison with the results produced by other methods (for example Vanicek,
1999, Kielland and Dagbert, 1992) and it appears that SP 44 needs some
revisions in this area.
Other uses
United Nations Convention on Law of the Sea (UNCLOS)
Ratification of United Nations Convention on Law of the Sea (UNCLOS) in 1994
brought with it the possibility for Coastal States to claim a juridical continental
shelf beyond the 200 nautical mile-wide Exclusive Economic Zone, provided
certain conditions are met. Of direct relevance here is the 2500m contour whose
mapped position will, in certain cases, determine the outer limit of jurisdiction of a
Coastal State. The 2500m contour is to be used as the line from which a
constraining line can be constructed, at a distance of 100nautical miles seaward
of the 2500m contour. This 2500m+100 line is one of two lines that must be
combined to form a constraint on the outer limit of the area that a State can
claim. The other constraint line is one “350 nautical miles from the baselines from
which the breadth of the territorial sea is measured”. The two constraints are
blended together by choosing sections of whichever is most seaward. Potentially
then, any miss-location of the 2500m contour can substantially effect the area
that could be claimed by a State beyond 350nautical miles. Like other depth
7
17/04/00
Dave Monahan
measurements and contours, the 2500m contour will be displaced horizontally by
an amount equal to the uncertainty in depth measurement divided by the cosine
of bottom slope.
To examine claims, the Convention established the Commission on the Limits of
the Continental Shelf (CLCS). This body has issued Guidelines (UN, 1999) on
the evidence it will accept. The CLCS requires that submissions include
estimates of error using the formula and values for fixed and variable error in
SP44. Since it is unlikely that a lot of older data can meet the standard, and
indeed the Standard proclaims itself to be for the future, the Guidelines may thus
be requiring all Coastal States to collect new data for their claims.
The impact of technology
Positioning Capability
Moving from inshore to offshore, there is a progressively looser positioning
accuracy demanded for each Order of survey described in SP 44. This appears
anachronistic in the era of positioning satellites which are obliterating distance
from shore as an issue (ad this week of 40cm position accuracy at ranges of
900km). The hydrographer’s traditional safety net of being able to position data
better than the mariner who used it could position his vessel is long gone:
modern clients can use equipment that located them as well as the hydrographer
was located. Each year sees the accuracies of these systems improved, to the
point where even last years data are suspected of not being good enough. Why
then have varying requirements for positioning?
Position with respect to WGS
84, what are charts on?
Mbes
Recommendations
When work commences on the 5th Edition of SP44, consideration should be
given to the following recommendations. Indeed, it is hoped that publishing them
will help move the IHO towards the goal expressed in the 3rd Edition of issuing
SP 44 every 5 years.
1. The w/g that rewrites the standards should remember the point underlines in
the ISO Technical Committee 211 (TC211) definition of quality: The totality of
characteristics of a product that bear on its ability to satisfy stated and implied
needs. (emphasis added) SP44 Edition 5 must satisfy the implied needs that the
standard applies to all measurements of depth, whatever their use.
8
17/04/00
Dave Monahan
2. Bottom slope can be more important than depth measurement to overall
accuracy achievable, and should be included
3. Instruction should be provided for along track sounding spacing in areas not
having 100% Bottom Search
4. The standard should be altered to accommodate the many users want the
same accuracy everywhere, not accuracy that degrades with depth
Refs
Hare, Rob, 1997, Procedures for Evaluating and Reporting Hydrographic Data
Quality. Canadian Hydrographic Service, Internal Report, 176pp.
Hare, R. and D. Monahan, 1993. A modern quantification of historic hydrographic
data accuracy. Lighthouse, no 48, pp. 1-14. and FIG XX International Congress,
Melbourne, Australia
Kielland, P., and P. Dagbert (1992). The use of spatial statistics in hydrography.
The International Hydrographic Review, March, Vol. LXIX, No. 1, pp. 71-92.
International Hydrographic Organization, 1987. IHO Standards for Hydrographic
Surveys, Classification Criteria for Deep Sea Soundings and Procedures for
Elimination of Doubtful Data. International Hydrographic Organization, Special
Publication No 44, 3rd Edition, 32pp.
International Hydrographic Organization, 1998. IHO Standards for Hydrographic
Surveys, International Hydrographic Organization, Special Publication No 44,
4th Edition, 23pp.
UN
TABLE 1
Summary of Minimum Standards for Hydrographic Surveys
___________________________________________________________________________________
ORDER
9
Special
Examples of Typical Areas
Harbours, berthing
areas, and associated
critical channels with
minimum underkeel
clearances
Harbours, harbour
approach channels,
recommended tracks
and some coastal
areas with depths up
to 100 m
Horizontal Accuracy
2m
5 m + 5% of depth
17/04/00
2
3
Areas not described in
Special Order and
Order 1, or areas up to
200 m water depth
Offshore areas not
described in Special
Order, and Orders 1
and 2
20 m + 5% of depth
150 m + 5% of depth
Dave Monahan
(95% Confidence Level)
10
Depth Accuracy for
Reduced Depths (95%
Confidence Level)
a = 0.25 m b
= 0.0075
a=0.5m b =
0.013
100% Bottom Search
Compulsory
Required in selected
(2)
areas
System Detection Capability
Cubic features > 1 m
Maximum Line
Spacing
Not applicable, as
100% search
compulsory
a =1.0 m b =
0.023
Same as Order 2
May be required in
selected areas
Not applicable
Cubic features > 2 m
in depths up to 40 m;
10% of depth beyond
40 m
Same as Order 1
Not applicable
3 x average depth or
25 m, whichever is
greater
3-4 x average depth or
200 m, whichever is
greater
4 x average depth
17/04/00
Dave Monahan