Monday, December 8, 2014

Don't Blame eCharts for Anything

The grounding of Vestas Wind reminds us of basic issues in modern navigation.

We have long taught at Starpath that we should never blame echarts for anything that goes wrong. This is a fundamentally important rule that we see violated over and over again. Granted, there is much not to love about vector echarts—the ones called ENC (electronic navigation charts), as opposed to RNC (raster navigation charts), which are identical to their paper counterparts. Unlike the ENC, the RNC are updated weekly in print and echart at no charge. The rule applies to RNC as well, but they are not as often the brunt of criticism as are the ENC.

Vector charts are easy targets for blame. First, for general use in what is called ECS (electronic charting systems) there are no real standards in functionality and chart symbols, as opposed to ECDIS use (electronic chart display and information system, pronounced ek-dis) the professional system sanctioned by the International Hydrographic Organization. Sailors and other recreational mariners, however, do not often use official ECDIS software or charts.  We use mostly ECS, which means simply any combination of echarts and software we might have.  Some ECS strives for the ECDIS standards; others do not. The latest edition of the  Chart No 1 booklet now includes many of the ECDIS symbols, which is a valuable free download to have. The full range of ECDIS standards, however, is immense, as it also includes how the various aids and landmarks should be described, coded, and indexed for display in various layers. There are also standards on the ECDIS software functionality, which is often not as convenient as some ECS programs, which is in part why manufacturers do not use it.

A big difference between RNC and ENC chart symbols is the legends explaining the symbols are all in print on the RNC, whereas in ENC we usually need to right click it or highlight it and press Info to learn what it means. This has lead to several incidents. I recall a racing yacht grounding on the West Coast that blamed echart symbols, as well as a tanker hitting a bridge in San Francisco Bay because light symbols were misunderstood.

And of course there is the appearance of the land that differs between these two chart formats, sometimes tremendously.  All ENC start out as RNC that someone (or some software) manually (or electronically) digitizes. Eventually these might all be tested by satellite images, which is done to some level now.  Certainly at some point the ENC will be better than the RNC, and I would venture that someday all charts will be ENC. It is the logical direction to go, just as print on demand was a logical direction for paper chart production. But for now, we will find more errors in the ENC charts than in the RNC, and more of them blamed for navigation shortcomings.  ENC include all errors that might be in the RNC (from incorrect or outdated surveys) plus any new errors that come from the digitizing.

On the other hand, like any working database, the ENC will just get better over time as the errors are found and removed and new meta data describing the features are added. The attraction of the ENC comes from this great potential of including so much information, plus the fact that they are individually small files. Entire sections of an ocean basin can be included in one large file, including every harbor chart as well. Furthermore, we can have ENC charts for remote parts of the world where there are not many alternatives.

This last point came to mind immediately upon learning of the grounding of the Volvo Race boat Vestas Wind on a reef at the SW tip of Cargados Carajos Shoals on the 29th November at 1510 UTC. All crew are safe, which is quite a blessing in that the GPS track of the vessel shows 17 to 19 kts of speed within seconds of 0 kts. The grounding was just off Coco Island, meaning they were 5 miles off course to miss the hazard and at least 10 off for a safe passing at night.

This reef is in the quintessential middle of nowhere, about halfway through the 5,000-nmi leg two of the race, in the middle of the Indian Ocean, about 216 nmi NNE of Mauritius. But even though this is a remote location and a small group of reefs, islets, and shoals, it is not at all a secret place. In fact, the entire middle segment of the Indian Ocean has the Cargados Carajos Shoal’s name written across it in large capital letters in the graphic index to the US International Sailing Directions, Pub 171. It is Section 9. The Sailing Directions are the ocean counterparts of the Coast Pilots, and they have the same goal, namely to provide crucial navigation information that is not on the charts. For any long voyage it is fundamental that we check the Coast Pilots and Sailing Directions. In this specific case, however, we do not learn enough from the two paragraphs about this reef in the latest edition of Pub 171. In fact, the only chart of the area it refers to is DMA No. 61551, which has been out of print for about 5 years. It also has the confusing information that the SW tip might be 3 mi SW of where it is charted, then later says you can pass the SW tip within 1 nmi—presumably meaning from wherever it is.

Thus when planning a trip near this region we have to work harder. The natural next place to look is the BA Pilots, which are the British equivalent of the US Sailing Directions, though famously more thorough and famously more expensive. Vol 39, South Indian Ocean, is only $60 but other volumes are twice that. They are in good libraries, however.  Here we learn that  BA chart No. 1881 has pretty good detail, and No. 4702 is a smaller scale showing the extensive gamut the yachts must face headed north across the Indian Ocean. The BA Pilot devotes two full pages to these Shoals.

We do learn from both the US and BA sailing directions that the Cargados Carajos Shoals should not be approached from the east (breaking waves obscure the location of the reef, which itself is only poorly known) and that the Shoals should not be approached from any direction at night. The four boats that passed the Shoals well to the west some 5 hours earlier did so in daylight, and at least one may have been close enough to see them, which are described as clearly visible with some hills to 70 feet with vegetation and a few low-lying buildings. This can be seen on Google Earth (GE). There are even some permanent residents there, including two Coast Guard personnel—very isolated, but in a very pretty place.  Vestas Wind unfortunately came in from the east at night.

One other yacht passed well below 1 mile of the reef tip in daylight (too close for a navigation instructor, but not a racing tactician), and could have been lucky in light of the NGA warning about the SW tip.  Figure 1 shows the boat tracks. In daylight the closest may have seen the reef and its limits as they passed.



Figure 1.  Boat positions just after the time of the Vestas Wind grounding (blue track). The white boat is 111 miles ahead of this Vestas Wind position.  The yellow boat and white boat are 4.5 miles apart on this scale.

BA charts are available from all BA outlets worldwide (there are a dozen in the US), and remote  charts are probably in stock at many of them as it is part of a BA distributor requirement. The chart 1881 is the one of most interest here, being the basis of all other charts of this area, and the navigation charting story begins with it.

This chart is essentially the same as it was in 1941, based on a survey taken in 1894, with periodic updates over the years, but no real surveying improvements. There is the discussion mentioned in the sailing directions and much online chatter about whether or not the shoals are where they are plotted, but we can compare with GE to see that it is indeed very close (Figure 2). This is not an issue.  There is always the issue of having your GPS set to the chart datum of the chart, but that should be standard knowledge these days, and could account for hundreds of yards and rarely a mile. The out of print US chart 61551 is essentially identical to the BA chart 1881 (upon which it was based), but there is one important difference.

Figure 2. A Google Earth image overlaid onto chart 61551, showing good agreement. The approximate location of the grounding is shown with a red X. The segment shown is about 5 nmi across. The light showing on this outdated chart does not now exist. Soundings in meters.

If you compare 61551 with a paper version of 1881 purchased from an authorized BA distributor, you will notice one main difference in the vicinity of the grounding. The BA chart will have the notable light on Coco Island crossed out by hand in ink (Figure 3). All BA distributors are obligated to update the charts with pen and ink by hand, and this light was no longer functional after a BA notice to mariners in 2012. This light (nominally visible for 12 miles) is just  a few miles from where Vestas Wind went aground.

Figure 3. Section of the latest printed edition of BA chart 1881, hand corrected by Captains Nautical in Seattle, an official BA chart distributor. The correction is dated in the bottom left with the notice number. The BA RNC of this chart, on the other hand, has the light removed, which shows the value of official RNC. Soundings in fathoms

Which brings us around to the echarts. Unless you are purchasing official BA ENC or RNC (called ARC), which are very expensive, then you are buying from companies that do not necessarily have the same BA standards for keeping their charts up to date—and I must add NOAA standards as well these days, as all NOAA charts are updated weekly now. You can’t buy US litho charts anymore that may have been sitting in drawers for years. They are print on demand, and the masters, print and electronic, are updated weekly. 

But we are considering foreign charts, or charts of foreign waters made by commercial companies of the US or elsewhere. In the case of these shoals some of the electronic charts in use at the time of this grounding were not up to date in that they did indeed still show this navigation light, which could in fact (if there) have saved the day in this case. Hopefully it did not mislead any navigator.

Even worse than that, some of the ECS chart packages did not layer the scales of this region very well, and in fact this point was brought up in the chatter about this region by several sailors on the race on several boats. Namely on one scale the Shoals did not appear at all and on the next scale up they filled the screen (see Figure 4).  Thus if you did not zoom in at very near the right place, you would not see them at all­—that is of course if you did not know ahead of time that they were there from your homework, and that this was indeed the primary issue of the navigation for at least a day or so, and as such have all sorts of back up contingency navigation in place, including radar and depth watch, besides GPS waypoints and tracks. In short, we are back to our rule.

Figure 4. Example of one echart defect. This is a full screen view.  The top view (A) alerts us with the white patches that there are large scale charts available in that region. (B) shows zoomed into that region with multiple zoom-in steps, (C) is the very next scale step in from the view above it, which does show the light. (D) is next zoom in, and now the light is gone from all further zooms. (E) next zoom in.  (F) shows slightly different zoom from a different start that does show the light. The presence of the light and which zooms show it depends on where you start from the white region and how you move the screen before the next zoom.  But there is no way to watch your vessel approach this reef on a view that actually shows the reef. This is from an iPad app with poor  chart presentation. It might look different in computer software or GPS units running these same echarts.

No matter how bad the echarts and echart display might be, we cannot and should not blame them for anything. It is frankly our job as navigator to know such limits and work around them. When all is going well, echarts are an invaluable aid to navigation and racing tactics, but when the navigation is crucial, we need to back these up with other info. Figure 5, for example, shows a section of another commercial echart brand used on one of the race boats that also shows this light—I think.



Figure 5. Another commercial brand of echart used on some of the boats. It does show the light—or something on Coco Is (called South Is); it is not clear what that symbol means without right clicking it! We always have two issues: did the company up date the chart, and then did the navigator update from the company. Often the update option is a paid subscription.  It appears that this navigator had defined a custom boundary around the reef, which often has the virtue of scaling more nicely so it stays in view, as well as offering danger bearings and ranges. This chart shows much better sounding data than the BA chart. We assume the navigator highlighted the depth contour IDs (meters), again good practice underway.  We do not know which boats used which brands of echarts. Some I would assume have several.

Another flaw of at least one echart program in use at this time was when right clicking and getting the data for this (non-existent) light it gave us all the information except how high it was. This height is crucial for predicting visible ranges of lights, so I would consider that a major charting error. Its official Light List number was also missing.

But suppose this is the situation: You have this echart on shore before the race as you plan your route, and you look at the light and find the height is missing. You know how important this light might be, so you must find this height somewhere. A first place to look is the NGA Pub 122, List of Lights. It covers all lights worldwide and it is free.  In the latest 2014 edition of that book on page 566, sure enough you find that light. It is US No. 32892, International No. D6681.6. It is listed as Fl(3) W, 30s, 27 ft above MHW, with a nominal range of 12 nmi. All this info is on the echart except the height 27 ft.  That would mean standing on deck at 9 ft height of eye, you would see this light from about SQRT(9) + SQRT(27), or about 8 or 9 miles off. The 12-mi nominal range means it would show bright at that distance when it first comes over the horizon, and you might even “bob the light” by standing on the boom.

So you are better prepared... or so you would think. But not at all. This official reference book happens to be wrong. That light is not there; gone since 2012 as best I can tell.  It is an error in the 2014 Pub 122. So we are back to our rule.  This false confidence came from overlooking the main issue. The echart itself.  When things are crucial, and you are using echarts from a commercial company that are not certified by the Hydrographic Office of that country, then the first step is check the echart with a printed chart or an official echart.

At this stage, just a few days after the incident, we have no idea what led to the error. I have not heard of anyone directly involved with the boat blaming any echart, or in fact blaming anything.  We are all human, and humans make mistakes. The extensive online chatter from distant observers, however, went straight to the charts. 

The discussion here has been just one scenario. As it turns out some commercial echarts of this region are based on charts from the Indian Hydrographic Office, which happens to have an excellent chart of this area No. 2503 (Figure 6). Though much the same as BA 1881, they do they the light removed so users of this brand of echart would not be confused by this issue. Chart 2503 also has improved soundings data from their own surveys. I have not been able to test any echarts based on 2503 and what meta data appear in these echarts. I am trying to get samples of other brands for comparison. Also we have to assume that whoever makes the echarts (regardless of source) do keep them up to date. Chart 2503 was new in Mar, 2014.  Print on demand nautical charts from all nations are available online from EastView Geospatial.

Figure 6. Section of Indian Chart 2503, showing improved soundings over BA 1881.
With special thanks to the Hydrographic Office of India

Remember too that of the half-dozen or so commercial companies that make worldwide echarts, some are known to be better for some regions than for others.  We cannot learn which is best where from advertising; we have to check with experienced users.

The Volvo boats had been in difficult conditions for a couple days negotiating a tropical disturbance with winds up to 30 kts. Approaching the Shoals the wind direction was changing rapidly along the fringe of this Low (with clockwise winds in the Southern Hemisphere).  The grounding took place about 2 hours after sunset, which at this low latitude was well past nautical twilight, meaning dark enough that the horizon was most likely not visible.  Even in clear weather, you cannot see the reef downwind with waves breaking over it. There was a half moon about halfway up the sky to the east, which may have offered some light, but that near the Low this was likely hidden by clouds. I would have to guess that current was not a factor.  Both the OSCAR 5-day average and the latest RTOFS models gave current flowing to the west at about 0.5 kts at that time.  They were off to the east so this current would have helped, if anything.

I want to stress that vector echarts certainly have much virtue and not just limits. Besides the unique values mentioned, they offer custom displays and features that can indeed enhance safety not threaten it. Furthermore, all of the limits can and will eventually be removed.

For now we have the issues above in at least some systems, but these might not appear the same in all platforms—that is, in a computer versus in a tablet app, or integrated into the GPS display. On the other hand, it could be that even the best echart navigation software has limits on this that are inherent to the specific echart package. There is also the question of how this might vary between ECDIS standard ENC and commercial vector echart products. Though in use now for quite a long time, there is still a lot to learn and a long way to go.

PS. There is some discussion online, in print and audio, regarding work on the boat at high and low tide etc. But we must remember we are in the middle of the ocean here, so tide is not much of a factor. See Figure 7.

Figure 7. Tidal data near the grounding site. S=spring, N=neap.



A version of this note will appear in the Jan, 2015 issue of Blue Water Sailing Magazine, an issue that includes other topics of electronic navigation.





15 comments:

Anonymous said...

As a final thought, recall that vector echarts are just a database, so the zooming issues are really more an issue with the presentation software than with the echart itself. The same brand of echart could behave differently in different software or tablet app.

Not quite correct. C-Map etc SDK does all the drawing, not the charting program, so the various software packages will look much the same

David Burch said...

Thanks. I made some assumptions here. i will reword that, so the above wording will shortly be different. I might still guess that viewed in an iPad say might be different from viewed in a computer in this regard... ie how would the pinch and zoom translate. The sample i studied changed with pinch and zoom differently from using + and - buttons. Also what about viewed in one of the high end charting programs compared to small display from a dedicated GPS display? If this also the same in a nice charting program then the issue is even more crucial.

Anonymous said...

A major point is missed with much of this discussion. Cross check your assumption with your depth! If you think you are in the middle of the ocean, and get ANY depth on the sounder, much less a depth like 120 feet, you KNOW you missed something!

This works in almost all waters. Even when navigating into a harbor, I use the same approach.

I actually like vector charts because I can adjust the "safety depth" number. So I lay my plot, then adjust up the safety depth, confirm that I'm clear all the way in at x feet, then if I see something different I slow the boat way down. Sometimes something is on the chart, sometimes not!

David Burch said...

You have a good point here. In making several posts about this article i described the subject incorrectly. I will try to go back to those and fix them, if i can. The actual text is not at all saying we do not like vector charts, it is just stressing their limits...they have many virtues, as you point out a couple, and we refer to a couple more as well in the note. The message here is we should know their limits and not blame them if something goes wrong, and we would like users and makers of these to report and fix known errors as soon as possible. In some of these you can actually edit the meta data yourself and add comments, then send off a note to the maker. For the ENC to achieve the potential they have we must all take part in the feedback.

David Burch said...

After looking into this some more, i believe my original statement was correct. Even in different navigation programs using the same chart data the zoom and other features might differ in appearance. You mention one particular brand, but i am not addressing individual brands of charts or software. There are quite a few of each. Even for the same charts and software there may be different licensing levels that affect these issues.

David Burch said...

In looking a bit more into this, i see that several chart companies are indeed encouraging this feedback with online forms to submit proposed errors, and others have essentially a private notice to mariners page that list specific corrections that should be made to each echart, with instructions on how you change the meta data. Thus when using a particular brand of chart, we should not only check the official notice to mariners, we should check our source to see if they have private updates. If you have not tried this, check your own source to see if there is anything of interest there. They do report errors that are not at all related to official changes. Maybe someday there will be a twitter feed we can sign up to get all of these notices… might be an actual productive use of twitter.

Anonymous said...

What a fascinating review David, very well done and thanks.
I still get the feeling that World class seamen, particularly a designated navigator, would know all about these echart deficiencies and have the best and most up-to-date electronic and paper charts available for the routes. I have used a basic plotter for a few years as a weekend sailor and I am well aware of the loss of detail when zooming out using Navionics Gold UK charts. No way would anyone set up an Expedition route without applying chart constraints for the hard bits (aka land). Back on board Vestas - the sun went down rapidly, the boat was steaming along at 19knots quite clearly going to pass over shoals. When preparing for the night, when human eyeballing no longer works, nobody would go off watch without double-checking a shoal right on the route. A competitor yacht Team Alvimedica is just a few miles to windward altering course around the shoals, there are 9 lives at stake, one can see from the video that visibility conditions were poor and if crew were looking for the non-existent lighthouse then someone would have zoomed in on their position on a plotter to check it. The crew were obviously nervous and twitchy, I suspect they eventually saw breaking seas and probably heard something too. If your are Grand Prix ocean racing for days and weeks then you have to set a depth alarm on the sounder to remind you that the ocean is not bottomless.
Prophetic quote from the Volvo website before the race: "Much more importantly, the 2014-15 Volvo Ocean Race will be won and lost by the athletes, out on the water. There will be no one else to blame, and the crews are embracing the fact that the buck will stop with them."
Yes there is a lot of admiration for the actions of the skipper and the crew after the incident but then they were able to step off the boat onto a sandy beach in the warm Indian Ocean. Welcome to Coco Island, the island that 'wasn't there'. I see there is a wreck marked just South of their grounding, so if they don't salvage all of the boat maybe there'll have to be a new 2014 correction.
What price the Admiralty chart 4702 at $38 and Admiralty sailing directions NP39 at $92 for a $6million boat when the crew sunglasses were probably $300 each.

David Burch said...

Thanks for your note. You bring up another good point here that I only alluded to, but is important to navigators, using celestial or GPS. In the tropics the sun goes down fast (short twilight) so it gets dark very quickly. Just like in the morning, "the sun comes up like thunder on the road to Mandalay." It is dark then a short time later it is hot out.

PS... we have a few follow up notes on these echarts that will be posted in a few days. Some presentations are indeed much better than others, even if details of the nav aids may still be outdated.

Anonymous said...

David, The rapid sunset was from a quote by the Vestas on board reporter Brian Carlin where he ponders what the night would bring just a few hours before the grounding: http://tinyurl.com/n5s836t
Here: http://tinyurl.com/n68fd8t you can see the Volvo 65 nav station is a narrow desk (at 0:32) and all the nav is done on laptops. The navigator fills out logbooks on his knee as the desk doesn't have room, so paper charts would not be easy to refer to in such tight space. On Alvimedica the charts can be seen tucked behind the laptop at 0:43.
Chris Nicholson interview with Scutlebutt http://tinyurl.com/k3utwxk
"How much do you supplement the electronic mapping with paper charts?
We have paper charts on board of which — don’t know, you’d have to ask Wouter, but normally we haven’t been using them very much.
Did you have paper charts for this area?
That, you’d have to ask Wouter. We certainly had them. I knew five days out, we certainly had all the paper charts."
If the navigator was off watch there would still be somebody on the on-watch crew who was looking after basic navigation who could zoom into that patch of blue stuff on the plotter that the 65ft 11 ton lightweight carbon missile with 16ft draft and a 3.5Ton bulb hanging off it was heading straight for at 19knots.

Andrew said...

OK, my last posting on this, thanks for the forum David.
This video http://tinyurl.com/lg2htvz shows that the on-watch crew knew they were passing over a shoal and the depth was called at 40 metres (video time 0:07). The crew then saw some things ahead that made them concerned but it's a shame that they didn't monitor depth at this time as it must have dropped to 4m by 0:46.
Maybe, with the benefit of hindsight the first job, after crew safety, would be to get the water ballast off the boat. It could have been as much as 2 tonnes.
It's a fantastic and terryfying piece of footage. My heart goes out to these guys and I'm so glad they are all safe. It was the ultimate nightmare for the reporter Carlin to be on the toilet at the time of impact! Thank goodness it wasn't the Southern Ocean. It's a testament to the strength of the boat (my favourite design/build combo Farr/Green Marine) that the canting keel wasn't ripped off and the hull stood up very well to pounding on the reef all night. A 3 month old boat, what a heartbreaker.
Poignant human reaction at 1:26 to see the helmsman still trying to steer the boat despite the fact that she was parked on a reef. Those wheels became dangerous spinning objects after the boat grounded 2:05.

Anonymous said...

Interesting article, covering a lot of good points about the different types of charts available today. The human factor in interpreting any and all types of nautical charts is clearly still a key factor in safe navigation today!

I don't fully agree with this point: "... for now, we will find more errors in the ENC charts than in the RNC". ENCs are increasingly becoming the primary charting product of many national hydrographic offices, being produced and updated before paper charts or RNCs. They may even contain more information as they are updated than those other products, as it takes less time to update ENCs than it does to update paper charts and RNCs. I'm sure you are right in saying that ENCs will become even more important in the future.

Finally, one small point - second paragraph "World Hydrographic Organization" should be "International Hydrographic Organization" (www.iho,int).

Anonymous said...

Honestly at 19 knots they wouldn't have had much time to do anything at all. The terrain changed very rapidly. The surrounding dept is several thousand meters. Also, since they were expecting to pass an area of 40 meters they probably would not have been alerted to a change to that depth. They ran up on a wall of a reef at 19 knots. I know of several areas I've sailed around where it comes from several thousand meters to 3 feet deep over a distance of about 10 feet. Your checking the depth sounder wouldn't do you one bit of good, nor them in this situation. My issue is that since they were expecting to go across an area of 40 meters I probably would have zoomed in on the chart to a lower scale to check the depth further in that area. Someone slacked off and the result is they put a boat on the reef.
End of story.

David Burch said...
This comment has been removed by the author.
David Burch said...

Thank you. I think you refer to official IHO ENC of some nations. I agree with you on the future of ENC. In the US, all paper and RNC charts are updated weekly (i.e. checked for changes and changes made if present), and it seems the ENC are also checked frequently, but i refer above to the commercial ENC, made and sold by private companies used outside of professional shipping. Perhaps we should have a difference name for these. These “ENC” are in some cases not updated very often, and the updates are not always thorough, and it is left to the user to then update their own.

But this too will change, and one day there will be commercial subscription services that keep all mariner's charts up to date automatically. It is one of the virtues of the electronic charting. Nevertheless, for now it is my experience that in the ENC i am familiar with there are more errors in the ENC than in the RNC—under the assumption that the latest edition of the RNC are correct. In other words, if the RNC is correct, meaning the paper chart is correct for US charts, then one single error or missing information in the ENC makes this point, and that is true even if in fact on some other aids there is more information on the ENC than on the RNC. The ENC inherently can contain more information, but they should in my mind include at least all of what is on the RNC.

For example, if i see a + with 4 dots on an RNC, i know this is a rock is awash just at the surface at tide height 0. This is crucial information for small craft navigators. If the ENC simply calls this rock awash, i would call that a serious error… despite the fact that on this same RNC i might see a symbol for a tower and know from that nothing more (unless i read the coast pilot), but on the ENC i might learn very much more about this tower. I would not call than an error in the RNC. So that is my biased approach to this.

it is on the list here to make an article to illustrate this point, using up to date official ENC charts, but again, the errors we might find should be sent immediately to the source, so on next check they are not there. In fact, we have an article like that in our course work somewhere showing glaring errors in nearby waters from several years ago, and i would bet if we look at the same place now these ENC errors will be gone.

And thanks for IHO correction. It seems i am more often referencing the WMO in writing, and hence the slip. I will fix that in the original.

David Burch said...

Interview with skipper and navigator at http://www.news.sail-world.com/Volvo_Ocean_Race__Chris_Nicholson_on_the_Team_Vestas_Wind_disaster___1/129959