Wednesday, December 17, 2014

Questions about electronic charts in a twenty-first century ship grounding

 On Saturday, November 29, Team Vestas Wind’s boat grounded on the Cargados Carajos Shoals, Mauritius, in the Indian Ocean.
Fortunately, no one has been injured. 
photo : NCG Operations Room – MRCC Mauritius​

We showed in our previous GeoGarage blog article that the grounding of the Vestas had no relationship with some positional discrepancy in the nautical charts, contrary to a similar grounding such as the USS Guardian in the Tubbataha reefs (Philippines) which occurred last year (see GeoGarage blog).

A spokesman for the Volvo Ocean race said that it would be reasonable to assume a large reef formation would be easily detected early on and avoided by changing course as 4 other boats did before Vestas grounding but during the day time not in the darkness.
At a press conference on Monday, Team Vestas admitted that the crash appears to have been caused by “a simple human error”: the navigator, Wouter Verbraak, did not zoom in enough on his charts, worried by a a high-stress situation caused by approaching bad weather.
Positions of the tropical cyclone 02S,
located East-Northeast of Mauritius island 
(Joint Typhoon Warning Center) 

 Meteosat 7 IR picture (courtesy of MTOTEC) 29-11-2014 09:00Z
Low pressure forecast from South African weather service / other satellite pictures 1 & 2

“Prior to the crash in the preceding 48 hours, Wouter and I in regard to our normal duties of looking where the boat was going with the routing, noticed that there would be some seamounts. When I saw those I asked what the depths and the currents and the wave conditions would be." said Skipper Chris Nicholson.
“Wouter’s reply was that the depths went from 3000m to 40m, (which) were the extremes of the depths, the current was negligible and we would monitor the wave state as we approached...”
Wouter explained the reason for the accident to the media:
“In hindsight we would’ve continued to zoom in on the area much more, on the electronic charts. Not doing so is the big mistake that I made, but the good thing is that we didn’t make any more.”
(see his own statement / SailingAnarchy / ZeilenNewYorker / SailingScuttlebut)

Racing navigation isn't ordinary navigation.
Before they left Cape Town in 40 knots of wind, the teams had already been warned of a possible cyclone. They then skirted an iceberg exclusion zone in the Southern Ocean, dodged a tropical depression and had to keep clear of a new Indian Ocean exclusion zone (released one day before the start of the leg) in case of pirates, which lead to set the Cargados shoas in the direction of the Abu Dhabi final destination.

Note that an independent report into the grounding of Team Vestas Wind's boat has been set up by the Volvo Ocean race.

David Burch in his own blog writes some interesting articles 'How a big, well-run, high-tech race boat, can go aground... almost' & 'Don't blame eCharts for anything' completing our first article.
Following to his different comments, we must add that private vector charts displayed in ECS software or in some plotters or tablets call for comments regarding cartographic generalization issues.

 zooming animation on the Marine GeoGarage with NGA raster charts layer :
showing a continuous vision of the Cargados Carajos shoals,
with 'exaggerated' shallow (land) parts on zoomed out views
Other example : same continuous vision of raster charts zooming from general to detail 
with British Admiralty charts :
see zooming animation on the Marine GeoGarage with UKHO raster charts layer

With official vector ENCs, we have also the same continuous vision zooming from low scales to high scales.
In our case, 3 ENC charts coming from 3 different Hydrographic Offices (UKHO, SHOM and NHO) are today available in the worldwide ENC catalog of charts :

respectively overview (GB104702), general (FR274880) and approach (IN42503A) charts

  • GB104702 Indian Ocean - Seychelles group to Nazareth Bank and Agalega Islands (overview) scale 1:3,500,000 (updated 31-05-2012)
  • FR274880 Ile de la Reunion to Rodrigues island and Ile Tromelin (general) scale 1:1,700,000 (updated 21-03-2014)
  • IN42503A Cargados Carajos shoals (approach) scale 1:45,000 (updated 25-03-2014)
GeoGarage ENC webmapping with s63 vector ENCs (GB104702, FR274880, IN42503A)
showing continuity in the display of the Cargados shoals during scaling

But this continuous display of the shoal is not showed in commercial vector electronic charts such as Jeppesen C-Map :

Jeppesen (C-Map) scaling animation
showing that the shoals are not correctly displayed on one of the chart levels.
Note : the same behavior occurs on other commercial nautical vector charts : Navionics / Transas 

 Plan2Nav for iPad (Jeppesen C-Map) zooming animation
showing at high scale depth soundings (on the South East part of the shoals)
whose source are probably coming from the Indian official IN42503A ENC

These above behaviors -with commercial electronic charts- underlines the issues in matter of cartographic generalization of vector-based data especially with computer automation.

Extract from NOAA document (1988) :

All charts are reductions of some part of the environment.
The reduction of the environment to a more comprehensible scale concomitantly yields a variety of undesirable consequences.

These include:

  • a decrease in the distances separating features on the chart;
  • a loss of visual clarity due to overcrowding; and
  • a shift of visual importance from the specific to the general.
In order to depict the important aspects of the Earth's surface at a more reasonable scale, features must be reduced in size and some detail of features must be omitted.
Also, entire features might have to be eliminated, enlarged, combined, and or displaced to fit within the graphic constraints of a typical chart.
To this end, the cartographer must apply a series of manipulations to the chart data in order to depict the important information at the reduced chart scale.

Example of cartographic generalization with coastline with ENCs,
showing how the chart accuracy and display of a coastal area
will be varied related to the selected scale range.
Large-scale charts encompass relatively small areas and show a great amount of detail.
Small-scale charts cover a large area, but show less detail.

The establishment of rigid guidelines for generalization has heretofore been a cartographic enigma.
Regardless of the apparent disparity in the definition of the term, cartographic generalization will be defined as the selection and simplified representation of detail appropriate to the scale and/or purpose of the chart.

Well known cartographic expert M. Eckert (1921) considered cartographic generalization as an subjective process dependable only on cartographers skill.
Today with the some advanced computer aid tools, automated algorithms are used to help the cartographer in the generalization processes especially for the coastline and bathymetry (soundings and isobaths).
The ENC data creation is the process to design and plan an ENC database, taking into account digital generalization factors.
Multiple representation of ENC data is controlled by SCAMIN attributes in the s-57 norm.
The SCAMIN (abbreviation of Scale Minimum) value of an object determines the display scale below which the object is no longer visible on an ECDIS, in order to reduce the amount of clutter displayed and to assign priority to the display of the different objects.
e.g SOUNDG objects -depth soundings- with SCAMIN=10,000 are not be displayed when viewed at smaller scales (>10,000)

So when presented in shipboard compatible Electronic Chart Systems (ECS) or Electronic Chart Display and Information Systems (ECDIS), the user can zoom in to the ENC and be presented with an increasing number of features, which only become visible once an assigned scaling minimum is reached, the SCAMIN s-57 attribut.
Note that in the next version of the ENC norm (s-101), an ENC should ideally be displayed between two scales: Minimum Display Scale and Maximum Display Scale. (see IHO)

But some spatial objects in the usual s-57 norm such as natural coastlines (COALNE) may not be included within SCAMIN.
For example, a lower scale coastline is a generalized version of a higher scale coastline achieved by automated line simplification and smoothing techniques.

Following to some Email discussion with Nick White (navigator and Expedition software developer), it must be remembered that the 3 navigation software used in the Volvo Ocean Race (Expedition, Adrena or B&G Deckman for Windows) are ECS dedicated at first to weather routing, each using commercial electronic charts (C-Map MAX in this case, but are not compatible with raster ARCS from UKHO) and not ECDIS systems (with ENC charts) fulfilling the requirement under SOLAS regulation so they can not be used to replace official paper nautical charts which are available onboard (Use of private produced charts is not officially approved for navigation).
As ECS are at first dedicated to routing (weather optimization function for racing and performance), so usually displayed at low scale charts and are not really related to navigation as such : they might be only used as additional navigation aids to complement the usual job with the paper maps.
But it's also true that today, modern race boats have rudimentary room for A0 paper charts : so it's not obvious to always get all the relevant charts accessible on the chart table.

General coastlines for land determination in the area
GSHHS World Vector Shorelines file (WVS) in black spots (general & detailed) + the one from Google in yellow

By the way, routing software on the markets offer the possibility to launch routing calculations with land avoidance for a performance point of view, so really used for large land masses as sailing can be done quite close to land/reefs.
Generally, the land avoidance routing algorithm uses high resolution shoreline files in order to check the route does not cross the land, or any shallow areas (via depth/hazard avoidance).

Vestas in blue / Dongfeng in red closer of the South West part of the reef
see zoom in the South of the Cargados
(courtesy of Volodia) 

Another interesting point is the fact that the Volvo organization (which receives satellite tracking positions at a couple of minutes of interval) didn't alert the Vestas team once it became quite obvious the boat were heading straight to the shoal (even if Vestas might have been going to sail up to half a mile off the reef and gybe to sail along it -see the other tracks-).

Knut Frostad, CEO Volvo Ocean Race watching C-Map/Jeppesen nautical charts
Crédit : A Sanchez

In conclusion, it's quite common nowadays -especially with the newer generation of sailors using electronics as a primary means of navigation-, to see that all these charting systems (plotters, software and mobile applications) lull people into an excessive sense of security, with the risk of encouraging them to rely too much on these tools they have come to trust, but which can't be the only key for safety of life at sea.

Links :