SO482A.  Historic shipwrecks: Science, History, and Engineering

GIS and Oceanographic Models in Search of the Bonhomme Richard

About the Presenter:

Professor Peter Guth (PhD, MIT) teaches oceanography, including geology and GIS, at the United States Naval Academy in Annapolis.  Dr. Guth’s publications include over 30 peer-reviewed journal papers and book chapters.  His research interests include geomorphometry from digital elevation models, algorithms using digital elevation models, and innovative uses of GIS in archaeology including the Battle of Big Hole (1877), 3D artefact distributions on the Swedish warship Vasa, and the search for the Bonhomme Richard.


Lesson Objectives:



Lesson Introduction--Click for WMV video clip.

We have used Geographic information systems (GIS) in several ways in the search for the Bonhomme Richard:

  1. Show spatial relationships of historic records of the battle and sightings of the ships involved.
  2. Show existing databases, such as wreck locations known to the United Kingdom Hydrographic Office, sonar seafloor surveys, and magnetometer surveys.
  3. Create models incorporating the tide and ship motion to track where the Bonhomme Richard might have drifted.
  4. Plan search operations.
  5. Manage search results.

In this lesson we will look at the principles of GIS and digital data, and see some applications as applied to evaluating the historical record of the final days of the Bonhomme Richard.  We will also look at some basic oceanography, and see how tides, winds, and currents might be incorporated into the GIS to model the drift of the ship.


GIS merges maps and databases, and has wide applications in the earth sciences, land use planning, and military operations.  GIS allows searching for spatial patterns and data visualization, and can be applied to historical studies.  Data layers provide the key feature of GIS, allowing integration of multiple data sets, and rapidly changing the display to reveal patterns and relationships.  Properly registered data, with geographic positions assigned, and standard data formats allow users to select from a wide variety of software.  At its core, GIS combines a computer database with a map display, and allows the user to go back and forth between the two. 

Google Earth provides many aspects of a GIS in a simple to use package that is free for personal use, and comes with free data streaming over the internet.  A wide variety of additional data sets, produced by government agencies and individuals, exist in the KML/KMZ format used by Google Earth.  If you do not have Google Earth on your computer, you should download and install it now.

You should now download several data sets related to the search for the Bonhomme Richard, and look at them in Google Earth.  You should click on the links for each of the data sets, and if you have Google Earth installed, they should open automatically.  The discussion forum and quiz will both assume you have looked at these data sets.


Table 1.  Data sets available as KMZ files for student use.

Shuttle Radar Topography Mission (SRTM) elevations.  This is an example of a raster data set, which has a regular matrix where each point has a specified value.  In Google Earth, the base satellite imagery is raster data.  If you try to blow up raster data too much, you will see the individual pixels. 

The SRTM data allows Google Earth and other software to show 3D visualizations.  For much of the world SRTM is the only data, and in other regions the only free data.  It has elevations values every three arc seconds (about 90 m), which is good for many studies.

Image: P. Guth

Drift and sinking locations for the Bonhomme Richard, based on simulations presented at the USNA Naval History Symposium in September 2009. This is an example of a vector data set, which is composed on line segments.  Points (like the sinking locations) are line segments composed of a single point, and areas (like the potential search area) are lines where the last point coincides with the first point).  In Google Earth, the roads or boundary layers are vector data.  If you try to blow up vector data too much, you will see the individual straight line segments.

Image: P. Guth

Range circles from Flamborough Head.  This is a vector data set with lines of equal distance from the spot that provided the name for the battle between Bonhomme Richard and Serapis.  Some of the historical records report the distance of the ships from the Head, and these allow us to assess when then ships could see land, or people on land see them.

Image: P. Guth

1794 British Chart of Flamborough Head.  This chart, from 15 years after the battle, is probably very similar to what Jones and the British would have had to navigate.  When you open this file in Google Earth, you will see that there is a file discussing where the chart came from, the steps taken to get it to appear in Google Earth, and any caveats on the accuracy and suitability of the information.  This is called metadata, and is important in GIS so that users understand what they can and cannot do with the data.

Image: P. Reaveley


Figure 1.  The GIS (here Google Earth) allows you to zoom and pan the map, and to easily turn layers on and off.

Previous  |  Auto/Stop  |  Next  

Image: Google Earth, displaying data from P. Guth



Figure 2.  One of the key strengths of GIS is the ability to go from a database to the map and back.  In you click on any of the symbols on the Google Earth map, you will see the full record for that point.  As shown in the table, you can see the conditions used for that particular simulation including the time, ship speed, tides, winds, and the course made good.  When you find the wreck, you will be able to get directions on how to get there.




Image: Google Earth, displaying data from P. Guth

Figure 3.  Unlike a full GIS, Google Earth lacks a few capabilities that can greatly enhance analysis.  In the map to the left, the colors indicate the hourly speeds of the Bonhomme Richard for one of the simulations.  The ship was sailing at 2 knots eastward, but the tide changes both the speed and the direction.  In addition, a GIS allows the database to be filtered, and only records matching the user's criteria will be displayed. 

Despite these limitations, Google Earth makes an ideal display mechanism for the display of GIS analysis, and results can be easily disseminated--with luck, each student has been looking at the data.

Image: P. Guth



One of the best "tricks" in Google Earth is its ability to show a three dimensional view of terrain.  This requires digital topography, most often supplied as a digital elevation model (DEM).  The most commonly used DEM is from the Shuttle Radar Topography Mission (SRTM), which flew on space shuttle Endeavor during a 10 day mission in 2000.  SRTM is an international project led by the National Geospatial-Intelligence Agency (NGA) and the National Aeronautics and Space Administration (NASA).  Data covers most of the earth's land area (it could not cover Antarctica or the higher northern latitudes) with a spatial resolution of 3 arc seconds (about 90 m), and this data is freely available for downloading on the internet.  In developed parts of the world higher resolution elevation data is available.  Much of the US has 1 m data, such as that was used for this>  model of hurricane flooding at Annapolis, but 1 m data requires 10,000 times the disk storage and computer processing.  In addition, outside the US most digital mapping data does not follow the US model of free distribution from government mapping agencies, and data can be very expensive.  For a wide range of analysis goals, the 90 m data provides adequate resolution, and in many areas, it is the only reasonable alternative. 


We have used the SRTM topography to calculate the locations from which Serapis and Bonhomme Richard could see the cliffs at Flambouough Head on the British coast.  Because intervisibility is reciprocal, these are the same locations from which observers on land could see the ships.  The computations consider earth curvature, and we have repeated the calculations for various heights above the sea surface--sailing ships would send lookouts into the ship's rigging because of the increased distances they could see.   The computations can be done as profiles showing the intervisibility, as perspective views of what a person would see at a location, or as maps (called viewsheds or weapons fans) showing visible and masked terrain.  The military uses these for direct fire weapons, radios, and radars which all require line of sight; civilian uses include cell phone tower coverage, or preserving the environment in parks and other areas threatened with development.  We mentioned in the navigation lesson that ship's officers had rules of thumb to estimate visibility, but with the GIS we can account for the added range with the 130 m tall cliffs


Figure 4.  View of the cliffs at Flamborough Head from the presumed location of the Serapis and Bonhomme Richard at 1500 on Friday 24 September 1779.  This view considers the curvature of the earth, and contains significant vertical exaggeration. The point considered is 10 m above the sea surface.

Image: P. Guth using SRTM data

Figure 5.  Locations that can see the cliffs at Flamborough Head from locations 5 m (red), 10 m (yellow), and 15 m (green) above the sea surface, which correspond to different locations on a ship and its rigging.  Conversely, someone along the cliffs can see that portion of a ship higher than 5, 10, or 15 m.

Image: P. Guth using SRTM data


Tides in the North Sea


The tides respond to the gravitational attractions between the sun and earth and the moon and earth.  Most places have two high and two low tides a day (the actual periods are a little over 24 hours).  Spring tides, with larger tidal ranges, occur with the new moon and full moon.  The tidal range off Flamborough Head in the North Sea is about 5 m, but away from land the vertical tide motion decreases and has minor impact on most things.  The tidal currents required to move water horizontally will increase as the tidal range increases.  The currents will approach zero at slack water at both low and high tides, and reach a maximum in the middle of the cycle.  There will be four peak currents daily, approximately every 6 hours, and 4 minimums when the tidal currents approach zero.

Tidal currents in the North Sea can have speeds exceeding 2 knots, which according to the logbooks is as fast as the Serapis moved in the days immediately following the sinking of Bonhomme Richard. We have tidal hindcasts from the High Resolution UKCS Model: Surface Currents (CS20-15HC_S), Proudman Oceanographic Laboratory.  The tidal current predictions cover the region show in Figure 6 for a three day period from 1800 Thursday to 1800 Sunday, which covers the period of the battle until after the sinking of the Bonhomme Richard and the Serapis moving away eastward.  The North Sea has diurnal tides with a period a little over 12 hours, and consequently the tidal currents have about 4 maximums and 4 minimums per day (Figure 7; the hourly computations do not capture the zero velocity when the tide actually reverses).  Maximum current velocities were increasing during this time period of the battle and its aftermath, with the full moon on Saturday 25 September and the maximum spring tides two days later.  We use the Proudman estimates at the 4 nearest points to perform a bilinear interpolation to estimate tide currents at any location shown in Figure 8 for our drift models.  Using a time step of 1 hour, to match the tidal predictions, we do not have to do any time interpolations.



Figure 6.  Tide predictions for N54°10.498' E0° 0.750' in late September 1779.

Image: P.Guth using Proudman Oceanographic Laboratory data


Figure 7.  Predicted tidal currents for N54°10.498' E0° 0.750' in late September 1779.

Image: P.Guth using Proudman Oceanographic Laboratory data


Figure 8.  Animation showing the tidal currents off Flamborough Head in late September 1779.  The vectors show the predicted direction and magnitude of the tidal current at hourly intervals.  Note that the tide is not at the same phase everywhere in this map area at the same time, and that the maximum speeds also vary with location.


Image: P.Guth using Proudman Oceanographic Laboratory data



Drifting Bodies and Ship Motions


We have tried to model the motion of three things for what they can tell us about the sinking of the Bonhomme Richard:

  1. The motion of the Bonhomme Richard from 1500 Friday afternoon, the last reliable and precise sightings from land, until it sank at 1100 Saturday.  There are no logs or other accounts giving speed or heading of Serapis and Bonhomme Richard which we interpret to have been sailing together with other ships in the squadron. 
  2. The drift of the mast and rigging of the Serapis which was cut away as John Paul Jones worked to get the ship seaworthy, probably Friday afternoon.  This was sighted by a British search squadron about noon on Sunday, and we can take estimates of that position as the ending point of drift simulation with the tides and wind currents to work out where Serapis would have been at different times.  Even if we do not know exactly when the mast was cut loose, we can constrain where Serapis must have been depending on when the mast was cut loose.
  3. The path of Serapis to the Texel, Holland, for which the log exists and starts an hour after the sinking of the Bonhomme Richard.

A drifting body will respond to motions from the tides, and from wind induced currents.  If the crew is actively sailing a ship, there will be a third motion caused by the force of the wind exerted on the sails. The tidal current will generally move back and forth, and is the simplest and easiest component on the historical BHR scenario to compute.  The wind induced current will show a lag with the initiation of the wind, and will be at an angle from the direction of the wind due to the Coriolis deflection.  The speed and direction of the wind current will depend on the latitude, the depth at which the current is measured, and the wind speed.  The sailing rule of thumb (2% of the wind speed after 12 hours of steady wind, about 15º to the right of the wind in shallow coastal waters in the northern hemisphere) is harder to apply because the wind was changing during the time period of the Bonhomme Richard motion.

Because we have no direct information about the speed and heading of the Bonhomme Richard, we have not tried to separate the wind-induced current from whatever motion they might have gotten as they tried to sail the ship.  We will cover the battle and its aftermath later in the course, but for now you should realize that masts, sails, and rigging had all suffered extensive damage, and both ships had significant damage to the hulls which caused the Bonhomme Richard to sink.  We enter values for the model (Table 2) with a starting location and a speed and direction for each hour, and like a spreadsheet, the GIS computes the other values in the table and plots the positions on a map (Figure 3).

Table 2.  Drift  models (Sym_1).  Bold values in red are inputs to the model, and the other values are computed.

54.1666667 0.0833333 Fri 1500 90 2 155 1.76 3.17 120.2
54.1401168 0.1609517 Fri 1600 90 2 159 1.63 3 120.4
54.1148335 0.2342313 Fri 1700 90 2 159 1.19 2.68 114.46
54.0963752 0.3031991 Fri 1800 90 2 150 0.57 2.34 102.29
54.0880849 0.3678099 Fri 1900 90 2 32 0.23 2.13 84.78
54.09129 0.4278226 Fri 2000 90 2 356 0.78 2.09 68.1
54.1042621 0.4827875 Fri 2100 90 2 348 1.17 2.1 57.03
54.1232388 0.5326226 Fri 2200 90 2 342 1.28 2.01 52.76
54.143465 0.5779576 Fri 2300 90 2 335 1.12 1.83 56.2
54.1603576 0.6209806 Fri 2400 90 2 323 0.69 1.67 70.74
54.1695345 0.6657916 Sat 0100 90 2 255 0.24 1.77 91.98
54.1685083 0.7159007 Sat 0200 90 2 176 0.67 2.16 108.11
54.1573462 0.7739765 Sat 0300 90 2 163 1.14 2.57 114.99
54.1392417 0.8400513 Sat 0400 90 2 156 1.28 2.78 114.87
54.119777 0.9114672 Sat 0500 90 2 147 1.16 2.81 110.39
54.1034845 0.9859583 Sat 0600 90 2 133 0.89 2.73 102.93
54.0933156 1.0611604 Sat 0700 90 2 112 0.62 2.59 95.27
54.0893387 1.134101 Sat 0800 90 2 105 0.5 2.49 92.97
54.0871652 1.2045646 Sat 0900 90 2 120 0.59 2.52 96.75
54.0822102 1.2755041 Sat 1000 90 2 135 0.98 2.78 104.27
54.0521511 1.4337701 Sat 1100 90 2 141 1.43 3.11 111.08


We have done 5 simulations (Table 3) which we think represent end members, with maximum and minimum probable velocities.  Taken together they define a maximum likely search region.

Table 3.  Five simulations presented at the 2009 USNA Naval History Symposium

Sym_1 Bonhomme Richard sails due east (90˚ true) with a speed of 2 knots.  Given the wind conditions, this is probably the extreme heading they could have achieved, and the fastest speed given the condition of the ships and the need to transfer men and supplies among ships in the squadron.  This is the speed and heading of the Serapis when the log resumes at 1300 Saturday, adjusted for the difference between magnetic north and true north, and reflects that fact the ship actually did sail generally eastward to safety.
Sym_2 Bonhomme Richard sails due northeasterly with a heading of 30˚ true with a speed of 2 knots.  This assumes the validity of comments from British observers, attributing the ships as heading toward Scandinavia.
Sym_3 Bonhomme Richard  had no headway and drifted with the tides.  Because of the time period involved, this simulation sinks only about 1.7 nautical miles from the starting location, although it had drifted about 5.5 nautical miles.  Given the assumed starting location, this would be the closest the ship could be to land.
Sym_4 Bonhomme Richard's heading is 50˚true, intermediate between Sym_1 and Sym_2.  It sails at 1 knot on Friday, and 2 knots Saturday.  This might be appropriate if damage control done Friday allowed John Paul Jones to increase speed on Saturday.
Sym_5 Like Sym_4,Bonhomme Richard's heading is 50˚true.  It sails at 2 knots on Friday, and 1 knot Saturday.  This might be appropriate if John Paul Jones recognized the inevitable and slowed down to transfer as many men and supplies as possible from the Bonhomme Richard before it sank.


Besides our use in looking for the Bonhomme Richard, similar models are used to model the dispersion of oil slicks, or to guide search and rescue efforts for lost boaters.  One of the most "interesting" studies of drifting with the currents was conducted by oceanographer Curtis Ebbesmeyer who studied the differences between rubber ducky bath toys and brand-name running shoes.  These behave quite differently--the toys stick up and catch the wind, and move with the very surface layer of the water, while the neutrally buoyant shoes move with a deeper water layer, and have a greater Coriolis deflection from the wind direction.  Besides being a fun story, this emphasizes that the drift of the Serapis mast, probably not sticking much above the water line or extending very far below, might be very different than the ship, whether or not the crew was actively trying to sail.



GIS is a powerful tool, both for research and the dissemination of results.  GIS does for mapping what the spreadsheets did for computation--lets you rapidly and easily change parameters, and graphically see the results.  Our models give us insight into the main questions which we need to take back to the historical record, of which I will mention probably the two most important:

  1. When did Jones stop sailing northward (or northeastward) toward Scandinavia, as reported by most of the British observers on land, and start sailing eastward, as reported in the log of Serapis starting at 1300 Saturday?
  2. How fast was Jones able to get the two damaged ships going, and how did he manage the transfer of men and material off the Bonhomme Richard?

The answers to these questions in large part determine which of our simulations is most nearly correct, and where the search effort should next focus.


Lesson summary--Click for WMV video clip.

Return to course syllabus


Last revision 3/1/2012