Depth surveys; Crowd sourced charts

Crowd Sourced Charts; History of amateur surveying

Overview:
I use a leisure-grade sonar and GPS to combine depth and position. I have two different objectives

Equipment:
Some amateur surveyors have used a tender/dinghy for surveying to get to shallower water without endangering their main vessel. MARLIN is a tough workboat and can take the ground - and I prefer to explore in the comfort of my wheelhouse. Also because of the big tidal range in my area (over 10m), I can safely explore areas at high water that will be drying at low water.

I use a Humminbird 898c SI sonar which has an external GPS antenna. This sonar has side-imaging (SI) which I have found to be difficult to use effectively in a marine environment (waves, current, turbid water,..). I have built a bracket to hold the SI transducer on a pole at a suitable depth over the side of MARLIN.
For maximum convenience I have also installed a conventional (puck) transducer which shoots through the (GRP) keel of Marlin. The GPS antenna is arranged directly above this. This can be used easily in any weather and while retrieving divers,.. (though note that bubbles rising from divers can pollute the sonar signal).
The Humminbird has a separate SD card slot for storing data and this allows sonar traces and GPS information to be saved every few seconds. I take this SD card home and process the data on my (Unix) laptop using codes I wrote myself.

Accuracy
Latitude and longitude are accurate to around 1 metre (s.d. value) and depth is accurate to around 0.1 metre. Attention must be paid to latency: depth and position information logged to the SD card are at slightly different times.
With a 200 MHz sonar transducer, Humminbird quotes a 20° cone angle for -10dB reduction. It is usual to quote cone angles for an intensity reduction of -3dB, then the transmitted beam cone angle is 11°. In practice, receiving the return signal will significantly reduce the spatial response region (to an 8° cone). This covers, for example, a region of diameter 2 metres at 14 metres depth. This smears the return from isolated or abrupt sea-bed features.
Altitude (see below) is less accurate but can be obtained to better that 1 metre by averaging over a long time period.

Tides:
Depth information needs to be evaluated relative to a useful datum: chart datum (CD) on hydrographic charts. This can be achieved by using the (freely available) raw tide gauge data (taken every 15 minutes at locations: Liverpool, Llandudno, Holyhead, Heysham, Workington,..) and quoted relative to CD. Tidal time differences to these ports are also needed for other locations than these ports.
The chart datum is given relative to Ordnance Survey datum (OD) (basically 0.2m below mean sea level: MSL) for ports on Hydrographic charts and more information can be obtained from the Admiralty Tide Tables. Note that in a tidal river estuary, the chart datum changes considerably as one heads inland. [also the tidal height prediction becomes less reliable].
 Chart Datum (CD) is basically specified to be at a level reached only by the Lowest Astronomical Tide (LAT). This is the tide that has the largest range (from HAT to LAT), theoretically. In practice, non-astronomical factors such as winds and air pressure can cause tide levels higher than HAT and lower that LAT.
A GPS gives altitude relative to the WGS84 ellipsoid. This can be related to MSL using the OS website OSGBN02 model. The altitude accuracy is usually worse (in metres) than that of latitude and longitude. However it is possible to average altitude over a period (say 15 minutes) since the sea-level changes quite slowly. I have explored this as a method to "act as my own tidal gauge" but found the resulting accuracy insufficient. Another drawback is that the Humminbird only outputs altitude information as NMEA so a computer/logger is needed to log it.

Codes:
The SD card can be analysed off-line. The data are stored in a binary form but this can be read and interpreted. I have written my own codes to do this (shell scripts plus Fortran 90 code).

Sonar information: the sonar trace is saved, so that one can reconstruct the "moving window" display as seen at sea. This is useful to look for anomalies, wrecks,... The Side-Imaging display (if stored) can be rerun also. The SD card also saves the depth determined by the Humminbird control head. This is usually reliable, except in shallow and or turbid water. Having data available from future as well as past times, one can re-analyse these traces to obtain optimum depth values.
For shallow water, I smooth the data over time, and over depth (since the sonar ping corresponds to an extent in depth) and then determine the location of the sea-bed. After checking that these depths are reasonably continuous in time, I get a robust depth value.
For ship-wreck studies, the sonar return at each time is explored to find the shallowest signal (top of wreck) and these values are then checked for reasonable consistency in time.

Position: The Humminbird unit exports to the SD card the latitude and longitude in Navionics format. (Navionics charts are the default chart displayed). These are integer values in "mercator metres": a "mercator metre" roughly corresponds to 0.6 metres at UK latitudes. [See GPSBabel module humminbird.c for more details.]

Mapping: combining the position and depth information gives a 3-dimensional representation of the sea-bed. I prepare files to be displayed by various Gnuplot options. The data taken will lie on a curved line (the vessel's track) and one can check (after tidal correction) for consistency when the line crosses itself in position. To interpolate between the track segments, I use a triangulation (Delaunay triangulation) to cover the area. This allows 3-dimensional images and also allows contours of constant depth to be determined.

Examples:
Three dimensional images of wrecks (scales in metres): Calcium ; Tug Nelson; Alarm ,
Depth tracks (scales in metres) Upper Mersey.

Crowd-Sourcing


Crowd-sourcing depth and position data is potentially a way to give up-to-date charts, especially of areas which are no longer surveyed by the Hydrogaphic Office (HO). I have experience of two such attempts:

TeamSurv:  I have submitted a lot of data to TeamSurv which is freely available with graphical display. They correct the data for tidal height and present it as tracks with spot depths. They show drying heights on sandbanks also. In some areas (S of England) there is sufficient data that they can present a chart with depth contours.
[I note that up tidal rivers, the chart datum varies and also the tidal height prediction becomes less reliable: TeamSurv is still working on this].
[As of 2019-2020, Teamsurv seems to have stopped updating their displays]

Navionics:  I have submitted data to Navionics for use in their SonarChart crowd-sourced addition. Much of my data was in shallow water and so above CD (LAT). This they initially showed as a narrow strip crossing a drying area (which is a sensible way to treat depths along a track). Later versions just ignore this data: they fix the shape of the drying areas (above CD or 0 of depth) as given by the HO and just add extra contours in areas below this. Their extra contours seem to be evaluated by interpolation keeping the zero-depth contour fixed. This is not a reliable procedure in an area with sandbanks and channels that move from year to year.
Even worse; I have noted that the SonarChart addition sometimes interpolates an earlier version of the HO chart than their "Navionics" version: so SonarCharts are actually more out of date.
I stopped submitting data to Navionics - since their treatment of sandbanks in an area of big tidal range is not adequate.
After discussion with Navionics - they admit their current offering has the limitations I presented above, but are planning to improve their product and will use all submitted files to achieve that. So I am now sending them data again.

I now also have a Raymarine MFD which can display charts. It can accept depth data (from Humminbird unit) by NMEA and has its own (internal) GPS. I experimented with Navionics Sonar Chart Logs - which are stored as .sdf files on the Navionics card in the MFD. This stores lat, long, depth and time but in a proprietry format (compressed binary by the look of it) which I cannot read.
I prefer to store data in a manner readable by me - so continue to use NMEA log and Humminbird stored track and sonar record.

Another Navionics option (available on the Raymarine MFD) is Sonar Chart Live. This stores depth (with option to correct it tidally within the MFD), lat and long. The files are stored on the Navionics card in lsdcache as .lsd files. This data is displayed on the MFD (in real time plus previous data). A single track is shown as a 50m wide swathe. Multiple tracks nearby allow contours to be drawn (depth interval between contours is selectable).
I have checked this out myself and the contours agree well with my own codes using my data and tidal corrections. In the case of an isolated wreck - the Sonar Chart Live smooths out the peak of the wreck - presumably because of a bias towards suppressing spikes in depth sounding records.
  Three features currently lacking which would be useful are (i) ability to display contours of drying areas (above CD) (ii) ability to enter a depth offset (eg transducer to waterline) and (iii) ability to display existing data without overwriting it with new data.
These .lsd files are binary files (uncompressed) which is an efficient way to store data and I have worked out how to read/write them myself. This provides a way to present my own data files on the Raymarine MFD. At present the limitations of Sonar Charts Live mean that I have to add 10 metres to depths (effectively using a new chart datum 10m above the real one), so as to be able to display drying areas.
  I have also investigated whether Sonar Chart Live .lsd files are still displayed after the Navionics "Freshest Data" has expired. Indeed contours (at 2m separation) are displayed with 0% transparency - but (for Raymarine MFD) with no option to change transparency or contour separation or turn off SCL (except by deleting/renaming the .lsd files).

Also:
For other "DIY" surveying examples see Antares and Mother of Perl[link broken].

History

There is quite a tradition of amateur (or self-taught) surveying of the coast of Wales.

Lewis Morris was resposible for the first detailed charts of the North Welsh coast: see here for more about his Plans of harbours, bars, bays and roads in St. George's Channel of 1748.

A later important input was from F.H.(Henry) Glazebrook who sailed from the west coast pf Anglesey in his 14ft boat MERMAID and lovingly charted every rock and recorded every eddy and inshore passage. His legacy is the Anglesey and North Wales Coast Pilot (1961).

Back to Index