Where’s the Confidence in Zone of Confidence?

This article was published in the February 2019 edition of the Nautical Institute Magazine “Seaways”

With the coming of ECDIS and vector charts came the concept of CATZOC (Category Zone of Confidence). The idea was that it would be an easily understood visual reference indicating the quality of the data on which the chart was based. CATZOC is expressed as a series of letters A to D, plus U (Unassessed), and also in the form of star ratings, with A1 (six stars) indicating the highest level of data quality, and ‘Unassessed’ (two stars) the lowest. More enquiring mariners may well have explored this new phenomenon, but for the most part, it was probably given scant attention – much like the survey source diagram on paper charts.

Ship operators are increasingly being asked to explain in detail how they use CATZOC data within their passage planning and monitoring processes. With that in mind, it is worth looking at CATZOC in some depth, and at the potential issues with the system.

The starting point

With paper charts, the more diligent mariner could refer to the source diagram to gain an appreciation of how old the survey data was, and how it had been obtained: this gave some idea of how accurate the survey might be. The experienced mariner knew that the surveys of certain areas, such as the South Seas, often dated as far back as the eighteenth century, and were obtained by magnetic hand-bearing compass and lead line from a small tippy boat, with longitudes often dependent on chronometer error by lunar distance observations. We were therefore not surprised to read comments like ‘This island is reported to lie five miles west of the charted position’. These were all evidence of what we would now call ‘data quality’.

We took this in our stride; day to day celestial navigation was generally even less accurate than the survey, and the prudent mariner made appropriate allowances for any errors in the chart.

The source diagram itself was easily understood but not particularly efficacious. Other than alerting the mariner to the possibility that the charted data was not very reliable or accurate, it offered no practical means or guidance to assess the degree of inaccuracy – certainly nothing of a numerical nature. It inherently recognised that the data, such as it was, was largely unqualified and unquantified.

By contrast, CATZOC seeks to define the undefinable and quantify the unknown. With the coming of electronic charts, those curious mariners who investigated their new vector charts would have been amazed to discover the abundance of data attached to even a blank area of sea floor, such as M_QUAL (quality of data), M_ACCY (the average shift of data), M_SREL (survey reliability information), QUAPOS (quality of position), and so on.

Assigning CATZOC values

S-57, the data standard which defines how an ENC is constructed, allows the quality of survey data to be recorded using many such metadata values. However, these are not easily assembled into a coherent mental picture. CATZOC is intended as a single composite indicator for bathymetry, taking into account vertical and horizontal uncertainty along with an assessment of the completeness of the survey. It uses an algorithm to combine all these features and assign a single rating indicating the quality of the bathymetry overall. This replaces the source diagram of the paper chart.

However, a questionnaire issued by the IHO Data Quality Working Group (DQWG) and completed by over 600 mariners worldwide clearly indicated that CATZOC was not well understood, nor liked, nor did it allow mariners to adequately make decisions based on data quality. The Working Group subsequently decided that CATZOC (in its currently recognisable form) will be dropped from the new S-101 chart standard. For the moment, however, the use of CATZOC remains a key part of the onboard audit process for some organisations – and is creating some problems.

A surfeit of data

In fact, much of the ENC metadata, other than CATZOC itself, is only relevant for data exchange between regional hydrographic offices. The philosophy appears to be that there is no particular reason to supress it, rather than any specific intention for the mariner to use it. Recent requests to hydrographic offices elicited little help on how this data could or should be practically used by the mariner. However, other organisations with arguably less knowledge or expertise have not been so reticent. This can lead to demands such as using CATZOC values to calculate underkeel clearances.

The problem seems to stem from the fact that, unlike the source diagram, CATZOC has numbers associated with it. When numbers are available, there is an unfortunate human tendency to try to coerce them into some kind of mathematical process. This probably reflects the equally unfortunate tendency to regard any information which is digital or numeric as somehow implying higher accuracy. We often see this manifested in unrealistic expectations of accuracy from electronic charts and other navigational equipment.

On paper charts, there was never any suggestion to try to make numerical depth allowances from the source diagram, and the mariner (and those in other positions of influence) might do well to remember that it is precisely the same data on the ENC.

Taken to its logical extreme, using CATZOC values to reduce underkeel clearance implies that a ship which is navigating on electronic charts could not load to the same draught as the same ship navigating on the equivalent paper charts, where no such values exist. Although there is no suggestion that this is actually happening, there is nevertheless some anecdotal evidence of ship operators retaining or even moving back to using paper charts as their designated ‘primary’ system, partly due to the problems surrounding the understanding and use of CATZOC.

Practical issues

So much for the general concept of using CATZOC to determine underkeel clearance. When considering the practical detail, there are a number of problems which arise:

  • In many cases, and particularly for old and very old survey data, there is little or no recorded information on the accuracy of the survey data from which to determine the metadata from which CATZOC is calculated. Much of this therefore has to be estimated using broad generalisations. Footnote 3 to the CATZOC table states that ‘Depth accuracy need not be rigorously computed for ZOCs B, C and D but may be estimated based on type of equipment, calibration regime, historical accuracy etc.’
  • CATZOC is further degraded by certain hydrographic offices assigning unduly distrustful values to their published survey data. While this might assist them in reducing their liability in the event of a navigational incident, it is of no help at all to the poor Second Mate attempting to determine a sensible and workable underkeel clearance. Whatever the reason – technical or commercial – most of the world is currently rated as CATZOC D or Unassessed.
  • To further complicate matters, CATZOC cannot currently indicate temporal degradation. The start and end dates of the survey may be encoded, but these are not represented. This is illustrated by the Goodwin Sands where the banks have moved 1,500m between consecutive surveys (2.4m per week), resulting in drying heights where depths of 20m were available 12 years earlier. The ENC shows a CATZOC of A1 (i.e. accurate to within 5m), based on the quality of the latest survey, but the banks could be half a kilometre from their charted positions. Whilst ENCs generally warn of shifting sands, it is unfortunately the case that the human use will nevertheless often focus only on the numerical data.
  • CATZOC is a statistic and not an absolute. Footnotes 2 and 3 of the CATZOC table explain that the positional and depth accuracies quoted for each category are estimated only to 95% probability – i.e. there remains a 5% chance of the errors being greater. If we wanted an increase of the accuracy to, say, 98%, and assuming a normal distribution, the CATZOC margins would double – e.g. A1 would become +/- 1 metre, C would become +/- 4 metres, etc. Even this would only reduce the risk to 2%. Finally, if we demanded <0.1% risk of grounding – as well we might – the CATZOC values would become so preposterously large that they would exceed the charted depth.

Dangers of misuse

Long before CATZOC came on the scene, we made broad empirical allowances in our underkeel clearance calculations – 10% of draught for ports and approaches, 25% for coastal passage, etc. These broad-brush margins included the allowance for perceived charting inaccuracies. If we add a CATZOC allowance on top – as I have seen on some ships – we are unwittingly doubling these safety margins. When the navigating officer adds the ‘CATZOC D: worse than (2 metres + 5% depth)’ allowance on top of the 10% UKC allowance he has already made, it is little surprise that his vessel cannot enter the port.

So where does this leave the poor mariner? We can perhaps look to the guidance given in NP232 Guide to ECDIS Implementation, Policy and Procedures. Section 12.4, for example, recommends allowing a blanket 10% of the calculated vessel draught in CATZOC A1/A2 areas, 15% in CATZOC B areas, and so on. This can be subsumed into the traditional UKC allowances we would make for in-port, coastal and deep-sea stages to give a simple matrix giving the percentage UKC to be allowed in any combination of CATZOC and sea stage.

This could give a simple and easily-applied criteria to replace the staged underkeel clearances we formerly allowed. However, although simpler, this approach tends to be more pessimistic than the detailed CATZOC calculations, especially at deeper draughts.

Alternatively, we can use dedicated software to generate the appropriate CATZOC allowance, together with tidal height and all the other underkeel allowances we want to apply.

Coping with the unassessed

Whatever method we use, the big question remains. How do we decide on a suitable allowance for areas which are ‘Unassessed’, or even those rated D? And perhaps more importantly, what entitles us to do so, considering that the hydrographic organisations have conspicuously avoided doing so?

Of course, in the face of poor data, the only practical solution is to obtain up-to-date hydrographic information from the port. But the mariner is still left with a ship which  – at least on paper  – can’t get into the port. The SMS has a specific procedure for this: a risk assessment is prepared. This acknowledges, among other things, that the vessel, and other ships like it, have safely visited this port at such draughts for many decades before and so an ‘exception’ is allowed by management. The danger is that such ‘exceptions’ become the norm, and are no longer properly scrutinised – and that is how accidents happen.

With CATZOC, the industry has created a problem of its own making, where arguably none really exists.  I leave the last word to Explanatory Note 10 to the CATZOC table which states:

‘The depth accuracy guidance found in NP100 does not require the accuracy values to be deducted from the charted depth but for the mariner to be aware of the likelihood of a different depth within the accuracy values’.

This is sensible advice and puts us right back in the realm of the source diagram – and how we used it, all those years ago.

Richard Leedham

Master Mariner, B.Sc., FNI, FRIN