How accurate are our charts?Jan 1, 2003
How accurate is the modern chart? This question must be a matter of fundamental concern for all chart users, both paper and electronic, since the Global Positioning System, especially differential-corrected GPS (DGPS) and Wide Area Augmentation System- or WAAS-corrected GPS, allows us to fix the position of our boats with more precision than was used to survey most of the details shown on our charts. What this means is that even if we have our GPS set to the correct chart datum and the position of our boat is accurately plotted, either manually or electronically, the surrounding coastline and rocks may not be; in which case, it might look as if we are in clear water when we are about to hit the bricks!
Before we are tempted to navigate narrow channels or sail close to any hazards, we need to know just how far out of position these charted bricks may be. Unfortunately, this is surprisingly difficult to quantify, because there is a large number of variables at work. Let's look at the key ones.Survey accuracy
No chart can be more accurate than the positioning accuracy of the surveys on which it is based. There is a significant amount of survey data still in use that was developed in the 19th, and even the 18th, century (for example, parts of the Pacific Ocean and Caribbean). When this data was collected, there were no recognized standards for survey accuracy, so the dependability of the survey results is largely a function of the skill and dedication of the surveying team and its leader. Both the horizontal accuracy (the position of one feature in relation to another, and the accuracy with which depth soundings are placed on a chart) and the vertical accuracy (the depth soundings, themselves) are questionable (not to mention the changes that have taken place since these surveys were conducted we came across one reef in the Bay Islands of Honduras that has grown vertically by 12 feet since it was last surveyed in the 1840s).
In the 19th century, various major hydrographic offices developed standards that were then adopted by other hydrographic offices. These standards reflected both the practical limits of accuracy that could be achieved with the available surveying and depth-sounding equipment and also the fact that it was not possible to navigate with pinpoint precision. Typical, until recently (mid-1990s), the U.S. government's general requirement was that positioning accuracy for coastal surveys be within 1.5 mm at the scale at which the survey was being plotted (e.g., if the survey scale was 1:20,000, the accuracy requirement was 1.5 x 20,000 = 30,000 mm = 30 meters, or approximately 33 yards). Prior to the satellite age, the same level of accuracy could not be sustained offshore. The charts compiled from these surveys were generally made at half survey scale (for example, a survey done at 1:20,000 is used to make a chart at 1:40,000), in which case, the theoretical error at chart scale becomes 0.75 mm x the scale of the chart (0.75 x 40,000 = 30,000 mm = 30 meters, once again).
In practice, errors may be considerably higher. In a letter written to me in 1996, the British Admiralty (BA) stated that most of their modern surveys have a positional accuracy of five to 20 meters; those done in the years after World War II are generally from 20 to 50 meters, and surveys from the early part of the century (pre-World War II) are from 50 to 500 meters to the andldquo;unknown.andrdquo;
All pre-1990 standards have been made obsolete by the advent of electronic navigation, especially GPS. Hydrographic offices have scrambled to catch up with this new technology. This has led to a set of internationally recognized surveying standards that, in 1998, were adopted by the worldwide community of hydrographers under the auspices of the International Hydrographic Organization (IHO). The standards are set out in IHO Standards for Hydrographic Surveys (IHO Special Publication #44, SP-44). Survey categories
SP-44 sets minimum standards for four different categories of surveys:
Special Order surveys: associated with harbors and channels that have minimum underkeel clearances for the ships that will be using the area and in which the highest degree of accuracy is therefore paramount. In general, horizontal-positioning accuracy must be within two meters and depths in shallow water to within 0.25 meters (approximately 10 inches), with the allowable depth error increasing marginally with increased depth.
First Order surveys: for less critical harbors, channels and coastal areas with depths down to 100 meters, which have not been surveyed to Special Order standards. In general, horizontal-positioning accuracy must be within five meters plus 5 percent of the water depth. In other words, in 6 meters (20 feet) water depth, positioning accuracy must be Â± 5.3 meters, or 5.8 yards. Depths must be to within 0.5 meters (20 inches) in shallow water, with the allowable error increasing with increased depth.
Second Order surveys: for areas with depths to 200 meters that have not been surveyed to Special Order or First Order standards. In general, the allowable horizontal error is 20 meters plus 5 percent of the water depth. The allowable sounding error is one meter (3.3 feet) in shallow water, increasing with depth.
Third Order surveys: for offshore areas not covered by other surveys. In general, the allowable horizontal error is 150 meters plus 5 percent of the water depth. The allowable sounding error is the same as for Second Order surveys.
Then there are minimum standards for the surveying of fixed and floating aids to navigation and for how much of the bottom should be covered at what level of accuracy. On Special Order surveys, the entire bottom must be surveyed with equipment that is capable of detecting any feature greater than one cubic meter in volume (which doesn't mean all such features get detected
there is a subtle difference between capability and actuality!). For First, Second and Third order surveys, surveyors run survey lines at ever-wider intervals. These lines are typically run such that, when plotted, they are five millimeters apart. For example, if the survey is conducted at 1:20,000, the survey lines will be 5 x 20,000 = 100,000 mm = 100 meters (110 yards) apart. At a survey scale of 1:40,000, the lines will be 200 meters (220 yards) apart.
In the past, much of the seabed between the lines remained unsurveyed. The BA noted in the 1999 edition of The Mariner's Handbook: andldquo;Without sidescan sonar, on a scale of 1:75,000, a shoal one cable wide (200 yards) rising close to the surface might not be found if it happened to be between lines of soundings. In the same way, on a scale of 1:12,500, rocks as large as supertankers, if lying parallel with, and between the lines of soundings might exist undetected, if they rose abruptly from an otherwise even bottom.andrdquo;Interpreting the data
When a chart is compiled from survey data, the cartographer will interpolate between survey lines to develop isobaths (lines of equal depth, also known as depth contours and depth curves). This may be of some significance to mariners. For example, on a circumnavigation of Cuba, we found an area with several long, shallow spits that ran out from the coastline between the survey lines (a recent survey). These spits were not picked up on the survey, due to its scale, so they were not shown on the chart.
Unlike previous standards, the SP-44 standards are absolute in that, once a particular survey category has been chosen, the standards are unrelated to the scale at which the survey is conducted and plotted (for example, horizontal positioning with a Special Order survey must be within two meters, irrespective of the scale of the survey and how the data is recorded).
NOAA, which is responsible (via the National Ocean Service) for surveying all U.S. waters, has adopted the First Order standards for all its surveys, both inshore and offshore (see NOAA Hydrographic Surveys: Specifications and Deliverables, published in June 2000, which can be downloaded from the NOAA website at www.chartmaker.ncd.noaa.gov). Note that an off-the-shelf DGPS or WAAS-corrected GPS may still position a vessel with a higher degree of precision than that of these survey standards!
The National Imagery and Mapping Agency (NIMA), which is responsible for U.S. charts of overseas waters, has this to say about its current level of accuracy (which is likely to remain for many years): andldquo;The NIMA-specified accuracy for harbor, approach and coastal charts is that features plotted on a chart will be within one millimeter (at chart scale) with respect to the preferred datum, at a 90-percent confidence level. For a large-scale chart of 1:15,000 scale, a one-millimeter error equates to Â± 15 meters (16.2 yards), which is the same order of magnitude as the absolute GPS error. For a smaller-scale chart of 1:80,000 scale, the chart error is Â± 80 meters (86.4 yards), which will become the limiting factor in position plotting accuracy. The reverse can be true for large-scale (small-area) charts, such as a harbor plan inset at 1:5,000 scale. In this case, the navigator's plotting accuracy is limited by the absolute accuracy of GPS, rather than the chart; however, features on this chart should be accurate to Â± five meters.andrdquo;
Of course, adopting these standards, and getting survey data that meets them, which requires new surveys to be conducted (the old data cannot be tweaked), are two completely different things. In recent years, most hydrographic offices have been under budgetary constraints at a time when there has been a great deal of pressure to convert the existing paper charts to electronic versions. The net result has been, in many cases, a reallocation of resources to the digitization program and a cutback in surveying. It will be many, many years before even the inshore areas of the world are re-surveyed to contemporary standards. Some areas may never get surveyed. In any event, the first to be covered will be priority areas for commercial shipping, which are often not of much interest to recreational mariners. Many of those areas of primary interest to sailors will be well down the priority list.
The IHO has another publication, Status of Hydrographic Surveys Worldwide (IHO S-55), which takes a broad look at the quality of worldwide hydrographic survey data from the perspective of contemporary accuracy requirements. It is a sobering document that shows we have a long way to go before this data is considered adequate.Data storage, retrieval and output
Before the development of electronic databases, the standard method for preserving survey data was in a written format. In other words, as the survey progressed, the findings were plotted onto a chart of the area. Various corrections were made to the data (such as correcting soundings for tide changes) until a final version of the plot was derived. This was known as a smooth sheet. The smooth sheet was submitted to the hydrographic office for approval, after which it became the official record of the survey and represented its limit of accuracy. Printed charts were (and often still are) derived from smooth sheets.
This raises some interesting issues. The finest line that can be drawn is about 0.1 mm wide, but such a thin line is hard to see, and as such, it is not recommended, sometimes forbidden, to be used for drawing features such as coastlines and other critical objects. As a result, various hydrographic offices have adopted 0.2 mm as the finest line to be used on a chart. Let's say the hydrographer decides to plot the smooth sheet at a scale of 1:20,000 (in other words, one millimeter or inch on the smooth sheet represents 20,000 mm or inches on the ground). At 1:20,000, a line that is 0.2 mm wide represents 20,000 x 0.2 = 4,000 mm on the ground. This is four meters. Even if the survey is accurate to within inches, this plotting accuracy has now become the limiting condition in the accuracy of the final product. If the smooth sheet is plotted at 1:50,000, the plotting accuracy is Â± 10 meters. If the pencil used to plot the data is not sharp and draws a line that is 0.5 mm wide, the plotting accuracy at 1:50,000 goes down to 25 meters, or 27 yards!
In the old days, paper was used for smooth sheets. Paper is notorious for being somewhat unstable. In a humid environment it will absorb moisture and stretch, though not uniformly, and in a dry environment it will dry out and shrink, once again not uniformly. This has the potential to add more errors to the stored data. To combat this, hydrographic offices often used cloth-backed paper, which is dimensionally more stable. In time, paper was replaced by plastic, notably Mylar (beginning in the 1960s), which is much more dimensionally stable.
In order to compile a chart for sale to the public, a cartographer would use (and in many cases still does use) these smooth sheets, along with shore-side surveys of landmasses (generally the province of another government department) and other sources of information. It's a fascinating process to watch. First, the hydrographic office must make a decision as to the area to be covered by the chart and the physical size of the chart itself. This will determine its scale. Next, the cartographer collects the most up-to-date and largest-scale (most detailed) maps and smooth sheets available, reduces these to the same scale as the chart and outputs them on sheets of Mylar (which may then be raster scanned if the chart is to be compiled electronically, as all now are in some hydrographic offices). Now the process of composition takes place as the cartographer uses his or her accumulated experience, refined through the collective experience and traditions of the hydrographic office and regulated by its standards, to create the chart.Determining adequate detail
It is often the case that the most recent and accurate surveys of an area do not have adequate detail, while there are older surveys with more detail. This is especially the case for soundings in shallow water, which used to be of interest to commercial and navy ships when their draft was relatively shallow. For this reason, they were often thoroughly surveyed in the past, but they are no longer of interest to these users, so they are low on the re-survey priority list. Nevertheless, they are clearly of great interest to recreational mariners. The cartographer may well use the recent surveys to establish the boundaries of landmasses and whatever other detail is available and then begin to extract additional data from older surveys to fill in missing pieces.
Due to surveying errors, these older surveys may not line up (register) properly with the new surveys. In such a situation, I have watched a cartographer place the Mylar with an old survey containing the chart under development beneath that and then slide the top Mylar around until the best fit is achieved, at which point details are taken off the old chart and added to the new one (if the chart is being compiled electronically, the same methodology can be used with the raster scans by manipulating the on-screen images). For example, if soundings are being added to a bay, the top Mylar may be slid around until the headlands on the chart under development more or less match the headlands on the old survey, and then soundings may be added from the old survey. Moving around to the next bay around the coastline, the cartographer will repeat this exercise. Clearly, such methods will introduce an unknown degree of error in addition to the original (unknown) surveying errors.
The cartographer also has some discretion (cartographic license) in the placement of labels and other details. This varies from one hydrographic office to another. If, for example, a sounding ends up overlapping an aid to navigation, a U.S. chart compiler may displace the sounding a little in order to make the chart easier to read (although, most likely, the sounding will just get left off). Whereas, a BA compiler will move the buoy if necessary (on the basis that the position of floating aids is not fully reliable). Or, if a channel is so narrow that it is difficult to draw, the U.S. cartographer may widen it to improve clarity (for example, in the event that a channel, channel buoys or markers plot at less than 0.5 mm in width, it is U.S. policy to plot them at 0.5 mm). Whereas, the BA compiler will never widen it and will most likely simply close it up if it is too tight to draw clearly. After all the science, there is quite an art to creating charts, the best of which are themselves works of art in their own right. Registering individual sheets
Until recently, when a chart was finished, it was outputted to a series of sheets of Mylar, one for each of the colors used in the printing process. These sheets were used to make the printing plates. The printing itself used a process called offset lithography. Given that each color had to be printed from a different plate, if the plates were not perfectly lined up (registered), further inaccuracies crept in. Once the chart was printed, the quality of the paper on which it was printed affected its stability, just as with paper-based smooth sheets.
This long-established process is still widely used, but it is changing. Increasingly, survey data is kept in electronic databases. These databases can be maintained electronically at any scale up to and including a theoretical 1:1 relationship with the surface of the earth (there has been talk of a worldwide effort to develop just such a worldwide database, known as the Worldwide Electronic Navigational Chart Database, or WEND). What this means is that the level of accuracy with which the data has been surveyed can be preserved in the database without further errors being introduced. When it comes time to use this data, it can be electronically scaled down to whatever scale is appropriate for the chart that is being created. This chart can then be used either in an electronic format or output to paper, printing directly from the electronic file without the use of lithographic plates. With this method, many of the errors formerly introduced by data storage, retrieval and printing processes are eliminated. Of course, all the errors inherent in the original survey data are still present.User beware
In the end, the absolute accuracy of the average harbor, approach or coastal paper chart is generally not less than one millimeter with respect to the chart datum, which is to say that the charted positions of features should almost always be within one millimeter of where they would be if the chart was completely accurate. Put another way, most of the time the cumulative errors will not exceed one millimeter x chart scale. For a 1:40,000 chart, this is 40,000 mm = 40 meters, or 44 yards. There will, of course, be many specific bits of data that have been derived from older andand#8260;or less accurate surveys than the norm, and which, as a result, fall outside these parameters, sometimes by a wide margin, as will almost all charts of areas beyond the immediate coastal belt.
This is all a long-winded way of saying that the user of any chart should not be lulled into a false sense of security about its accuracy. Before GPS, there was always a degree of uncertainty about a boat's position. This led navigators to give a wide berth to hazards. In general, the techniques used to position hazards on a chart were more accurate than the nav tools available to the mariner.
Since GPS, this situation has been turned on its head. The equipment with which we navigate now has a positioning accuracy greater than that underlying the charts we use (including electronic charts, which are usually based on paper charts and old survey data).
Before you go shaving any corners, you need to have a thorough grasp of, and a healthy respect for, the limits of chart accuracy! n
For more information on chart datums visit www.OceanNavigator.com. Click on the web extras button.
This is an excerpt from contributing editor Nigel Calder's latest book, How to Read a Nautical Chart, to be published in August by International Marine Publishing.