SO482A.  Historic shipwrecks: Science, History, and Engineering

Marine Geophysics--Search Tools

About the Presenter:

Professor Peter Guth (PhD, MIT) teaches oceanography, including geology and GIS, at the United States Naval Academy in Annapolis.  This is a long way from his initial undergraduate training for two years at Deep Springs College in eastern California (mail came through Dyer, Nevada), and then four years at the Military Academy.


Lesson Objectives:


Lesson Introduction--Click for WMV video clip.

Oceanographers often say we know the surface of Mars and Venus better than we know earth, because the 70% of our planet covered by water cannot easily be seen.  On Mars we can put up a satellite, and over time it can image the entire planet at whatever resolution we want.  Venus was a bit trickier because of the complete cloud cover, but by switching to microwave/radar imagery the Magellan mission mapped all of the planet.  For the ocean portion of our planet, we have to use sound instead of electromagnetic energy, and we cannot use a satellite or aircraft.  This cranks up the cost (research vessels can cost $50K/day, whether doing survey work or transit to the search area), and increases the time it will take to cover the entire planet.  Underwater autonomous vehicles (UAVs) might provide faster coverage at lower cost, but it will still be a long process.

Searching for lost shipwrecks uses the same search tools used by geologists studying the seafloor, or by engineers planning or inspecting pipelines and other underwater structures.  We will first discuss the detection of a feature by measuring an anomaly, and then go through the various instruments that could be used to search for the Bonhomme Richard or any other wreck.

Marine Instruments

Marine instruments can be either mounted on a survey ship, or towed behind the ship on a "fish".  Towing isolates the sensor from the ship, such as a magnetometer not being influenced by the ship's magnetic field, and can also isolate the sensor from the pitching and rolling of the ship.  Towed instruments will often give the best results if towed close to the bottom, but must be carefully monitored to insure they do not plow into the bottom.  The depth of a towed instruments depends on towing speed, the hydrodynamic characteristics of the fish, and the length of towing cable let out.  The fish can have a pressure gauge to measure its depth, or an altimeter to measure the height above the bottom.  Since the underwater fish cannot have a GPS sensor, the survey ship must also track the location of the fish relative the ship's GPS receiver so that survey results can be accurately plotted on a map.

Marine surveys can be done in broad terms to determine general characteristics of a large area, and then in more detail for interesting regions discovered in the first phase.  The type of survey frequently determines the line spacing between ship tracks, and thus the time required to survey a particular area, and the choice of instrument.


In January 2005 the USS San Francisco encountered a seamount south of Guam, causing substantial damage and killing one crew member.  In the southern ocean, shipping lanes are far apart and detailed mapping very sparse, and our best knowledge of the bathymetry generally comes from satellites measuring variation in the the earth's gravity field and the resulting changes in the sea surface.  The earth's gravity field is normally given by the equation


which works well when we can consider M to be the mass of the earth, and m the object being attracted, and view both as point masses.  If we want to consider the small differences in gravity that will result from variations in the distribution of mass within the earth, we might have to use calculus:

M, instead of simply being the mass of the earth, will be the mass distribution in three dimensional space, and we have to consider both the density of anomalous material, and the distance away from our gravity meter.  In Figure 1, we have three ways of covering the ocean with the gravimeter to measure the value of gravity: towing it in mid-ocean depths (A1 to A3), putting it on a ship near the surface (B1 to B3), and flying it on a plane or satellite (C1 to C3).  The seamount will have a specific gravity ρ=3.0, and the seawater will have ρ=1.0.  Near the seamount, we will have rock replacing water, so we have more mass, and because it is close, the inverse square fall-off will be much less and we will see a substantially larger value of gravity.  The difference will be greatest for the deep towed instrument, because as we go from A1 to A3, the relative distances change significantly.  At the sea surface, B1 to B3, the differences in relative distance to the mass anomaly will be less drastic, and we will have a harder time picking out the actual size and shape of the seamount.  If we have to fly the sensor (C1 to C3), our resolution becomes even less, and a satellite will be hundreds of kilometers up.  We will still see a slightly higher value for gravity at C1 compared to C3, but in nothing like the detail for the deep towed instrument.

Figure 1.  Seamount, with nine locations at which we might measure the earth's gravity field.

If there are different values of gravity at B1 to B3, the sea surface will respond.  Sea level is a gravitational equipotential surface, with the water at no place having a tendency to flow downhill as it would at B3 if the sea surface were truly flat.  B1 will have a higher sea surface than B3, by about 1/1000th of the height of the seamount above the ocean floor.  The seamount thus might have sea level a few meters above normal, but spread over a few tens of kilometers horizontally, so the slope will be very small (tangent about 3/30,000, so angle essentially 0, since for a small angle, the tangent equals the angle).  Nevertheless, we can measure these changes in the sea surface from space, and the "best" maps of the entire ocean floor have been measured using radar altimeters from space, and then work back to likely mass distributions on the ocean floor that can translate into ocean depths.  The resolution is not great, and while we have detected a large number of seamounts, we had not previously detected the seamount encountered by the San Francisco.

Interpreting the results can be complicated, because in addition to the ρ=3.0 seamount, we might have ρ=3.3 upper mantle involved, the shape of the seamount might not be a cone, and there is likely to be a "moat" around the seamount where the crust subsided due to the weight.  Despite the shortcomings, measuring anomalies (changes for the expected values) in gravity and the sea surface are powerful tools for geologists, but we must always consider that there could be multiple ways to get the observed patterns we see.

Gravity has been used to search for seamounts, and was in fact the first use of scientific geophysics used to look for oil.  Many oil fields in the Gulf Coast region are associated with salt domes (ρ=2.2) which will be substantially less dense than the rocks around them (ρ=2.7), and the use of gravimeters led to the discovery of many oil fields and convinced the oil company managers that employing scientists made sense.  Changes in gravity can only be detected for very large objects, and shipwrecks will not have a noticeable effect on the gravity field.

Anomaly Patterns

The anomaly that we detect will depend both on the properties of the material, and where our sensor flies:


If we replace the seamount in Figure 1 with a large mass of iron ballast, we might be able to detect the shipwreck based on the impact on the earth's magnetic field.  The iron will change the earth's field, generally nowhere near enough to affect a compass, but enough to measure.  Like measuring the seamount, we will be best able to detect the anomaly if it is concentrated (Figure 2, shipwreck A) instead of being dispersed (shipwreck B, subjected to centuries of trawling by huge nets) or buried (shipwreck C).    Like gravity, the magnetic field drops off with distance, so we will have the greatest chance of detection  with a deep towed instrument compared to a ship-mounted or aircraft-flown one.  The size of buried mass, and hence the anomaly it creates, will determine how close to it we must be in order to detect it.

A magnetic mass will have two types of magnetization, permanent (or remnant), and induced.  The permanent magnetization has the same orientation as the earth's field at the time and location where the material cooled below its Curie temperature (770°C for iron).  Below the Curie temperature, only a small portion of the iron atoms can realign their dipoles to align with the earth's field, which leads to an induced field in the same orientation as the earth's field.  For single objects, the permanent  magnetization generally dominates, and if the object can move (like a cannon), can have any orientation.  For a large number of objects, like the debris field of a ship, the permanent magnetizations will have random orientations and cancel each other out, leaving the induced magnetization in the direction of the earth's field to dominate.

Figure 2.  Iron ballast mounds of three wrecks.  Even if all three have the same amount of iron, the chance of detection will be different for all three since each will have a different anomaly.

As the ship tows the magnetometer (Figure 3), the results appear on a strip showing the values being recorded.  Anomalies (Figure 4) appear as positive or negative departures from a generally linear background pattern.  The size and shape of the anomalies depends both on the magnetic feature causing it, and the height of the fish above the feature.  Magnetometer data must be corrected by removing the diurnal variation and any geologic anomalies in the region, so final results are usually compiled at the end of each day's survey operation.  They can then be plotted on a map, and color coded and contoured (Figure 5).  Looking at both the track line records, and the contoured data, an analyst determines the locations of anomalies (Figure 6).  In addition to wrecks, magnetic debris will also create anomalies, as will manmade features like oil wells and pipelines (Figure 6).  Because surveys can show many anomalies, they must be categorized on how likely they are to represent features likes wrecks compared to debris and other random anomalies.  Magnetometers rarely provide clear evidence on their own, and the anomalies must be checked with other geophysical tools and eventually divers or ROV/AUV photography.

Figure 3.  Deploying Fluxgate magnetometer on EXPLORER.
Image ID: theb2720, 1960 , from NOAA Photo Library


Figure 4.  Track line anomalies recorded on ship.
Figure  5.  Contoured magnetic anomalies, overlaid on side scan sonar imagery.

from Neyland 2007, Naval Historical Center, Underwater Archaeology Branch report. 



Figure 6.  Magnetic anomalies in the North Sea, color coded by the size of the anomaly.  Because of the tides, the ship tracks ran NW to SE.  The continuous line of anomalies corresponds with an underwater oil pipeline.  Some of the others correspond with known wrecks or oil wells, but most are of small size and not related to any known features.

Unlike most of the other instruments available for survey work, a magnetometer can detect a buried wreck.  Because of the rapid drop in the intensity of anomalies, the survey track lines in a magnetometer survey must be close together.


Sonar Surveys

Sound travels about 1500 m/sec in the ocean, much faster than in air, and unlike light and other forms of electromagnetic radiation, sound propagates well in water.  Low frequencies have the least attenuation, but also provide the least detail, so sonar instruments frequently offer a choice of frequencies and the users can select either long range or high resolution of the results.

Sonar instruments usually record in two-way travel time (TWTT), the time it takes for the sound pulse to travel from the towfish to bottom, and then return.  If the sound stays in water, TWTT can be easily converted to depth using 1500 m/sec as the sound speed.  This will be a very good estimate because sound speed only varies by a few percent from 1500 m/sec depending on temperature, salinity, and pressure, and if you need extreme precision you can adjust for the actual sound speed.  Problems in converting TWTT to depth occur with subbottom profilers when the sound spends some of the time in the sediment or rock of the seafloor, in which the sound velocity can be much faster than 1500 m/sec (up to 6000 m/sec or more).  For shallow sediments and surveys for wrecks, a 1500 m/sec assumption will not be far off, but for deeper work geologists typically show results in TWTT instead of depth, and then show an interpretation that assigns different velocities to each layer and migrates the reflections to the correct depth if the assigned velocities are correct.

Some sonars aim directly under the ship, and collect data only underneath the vessel; examples include the single beam echo sounder and subbottom profilers.  The single beam echo sounder is no longer a sophisticated survey instrument, but every ship will have one, and the subbottom profiler is not a good general survey instrument because the ship must pass directly over a feature to image it.  Other instruments, like the side scan sonar and multibeam bathymetry, collect a swath, and make much better survey instruments because they can cover a larger area.  The width of the swath varies with the height above the bottom at which the instrument files because the sound beams usually make a fixed angular pattern (Figure 7).  Multibeam bathymetry systems are usually hull mounted, and swath width increases as water depth increases because they have a fixed angular field of view.  Side scan sonars are usually towed behind the survey vessel, and their swath depends on the height above the seafloor.  The swath systems sends a "ping" to collect a series of readings perpendicular to the track of the survey ship, and repeat as the ship moves to build up a series of pings that completely fills in the swath.

Figure 7.  bottom coverage obtained with both multi-beam sounding data and sidescan sonar data. These datasets can be combined in many ways to derive much information concerning the nature of the seafloor.

Image ID: theb4125, (NOAA Photo Library)

Echo Sounders

Single beam echosounders first came into use in the 1920's (Figure 8), and represented a tremendous advance over the lead line.  In World War 2 and immediately after, the use of echosounders on U.S. Navy ships led to the recognition of the major features of the sea floor basins, and led Harry Hess, a professor at Princeton and an officer in the Naval Reserve, to propose the concept of sea floor spreading.  While all ships used in survey work will have an echosounder, it will not be used for the survey work other than to keep the bridge team aware of the current water depth.

Figure 8.  First published diagram of an echo sounder.  In Die Meteor-Fahrt Forschungen, The Meteor Expedition (1928) by F. Speiss. Fahrt Forschungen und Erlebnisse der Deutschen Atlantischen Expedition, 1925-1927. D. Reimer, Berlin. (NOAA Photo Library.)

Multibeam Bathymetry

For each ping the multibeam bathymetry system collects a number of returns.  The time the ping spent in the water determines the distance to the bottom, and the geometry determines where the ping intersected the bottom.  The points can be displayed as a point cloud (Figure 9) or turned into a grid (Figure 10) such as we saw with the SRTM data on land.  Much of the area in the North Sea where the Bonhomme Richard might lay has large sand waves such as those in Figure 9; these migrate over time, and might now cover the wreck, and characteristic scour marks that usually occur around wrecks with a substantial tidal current.  Depending on the resolution of the system, the multibeam might reveal only a generalized mound, or it might show fine details on the wreck.  Computer display of multibeam bathymetry using color codes the depths by color, with one common scheme using blue or violet for deep, and red for shallow (the same convention used by many GIS data sets).

Figure 9.  3-D rendition of  BOW MARINER  from processed multi-beam soundings.

Image ID: expl4051 (NOAA Photo Library.)

Figure 10.  Wrecked tug with current scour marks and large sand waves displayed on multi-beam image.

Image ID: expl4070 (NOAA Photo Library.)


Side Scan Sonar

The side scan sonar creates an image of the bottom using sound waves.  While this can look like a picture, you must understand that the image depends on the interaction of the sound waves with the bottom.  The system uses the time of the return to compute a distance, and then displays the intensity of the return in a shade of gray.  The intensity of the return depends upon:

Strong returns are now generally shown in white, and no returns in black, denoting the sound shadow.  This was not always the case with the first systems which only used a paper recorder, so you should always verify the color convention used in the imagery you are looking at.  Colors are sometimes used as well to highlight very strong returns, and keep a watch stander alert as they come in.

Figure 11 shows two views of Submarine S5.  The image on the left is smaller, indicating the sonar was operating with a longer range.  This provides less detail, but covers a large area, and is generally how sonars are used in searching.  Side scan sonars are designed to view the seafloor from the side, and provide very poor geometry directly under the towfish (Figure 12).  In both Figure 11 and 12 the track of the towfish is shown by the large pixels.  For the image of the S5, once the NOAA survey ship located the wreck, they switched to a short range on the sidescan to collect a better image, and they returned for a second pass with the ship track oriented in the same direction as the wreck.  In addition, they insured the wreck was in the middle of one channel, and not under the fish.  If the survey that acquired Figure 12 was interested in details of the wreck, they would have taken a second pass and insured they passed to the side of the wreck.  In addition to not wanting to pass over the wreck, the towfish needs to be close to the bottom to enhance the shadows.  Note that in Figure 11 the shadows provide more information than the actual imaged portion of the wreck.

Figure 11.  Submarine S5 (sank in 1920 40 nm off the Delaware coast) sidescan sonar images. Image on the right obtained by knowing the orientation of the submarine and maneuvering the sidescan such that its track ran parallel to the vessel.

The illuminated area is the reflection and dark is the shadow.

Click for higher resolution image.

Image ID: expl4060, expl4061, (NOAA Photo Library.)

Figure 12.  Sidescan sonar record of shipwreck. The track of the towfish went directly over the wreck.

Click for higher resolution image.

Image ID: theb4045 (NOAA Photo Library.)


Survey vessels can acquire multibeam bathymetry and side scan sonar imagery at the same time (Figure 7), and combining this results greatly increases what an analyst can see in the data (Figure 13).

Figure 13.  Comparison of older side scan and multi-beam systems on ship wreck. The newer higher-resolution systems are to the right of the image.

Image ID: cgs00879,  (NOAA Photo Library.)


Subbottom Profilers

Subbottom or chirp or reflection profilers use sound to look into the bottom in a narrow swath directly under the towfish.  The sound pulse, unlike those used for bathymetry or side scan sonar, is designed to penetrate into the seafloor.  Some energy will be reflected with each change in impedance in the rock or sediment; impedance is the product of sound velocity and density.  The images show sediment layering, scour structures, and the presence of gas hydrates in the sediment.  Because the subbottom profiler shows only a very narrow strip directly under the ship, it is not likely to see a wreck in the search phase of an operation, but it can provide useful information in a detailed survey after the wreck was located.  Figure 14 shows results below the Severn River near Annapolis; changes in color and texture indicate the presence of layers with different characteristics.  Subbottom profiles should be as close to the seafloor as possible, to minimize attenuation in the water so that maximum energy enters the sediment.

Figure 14.  Subbottom profiles in the Severn River  Changes in color indicate layers in the sediment and old channels that have been scoured out and refilled. 

Image: P.Guth and A. Terwey, with data from YP686.



LIDAR, short for Light Detection and Ranging, can provide 3D point clouds much like multibeam bathymetry, and the data can be gridded.  LIDARS typically fly on small planes or helicopters.  Because of the poor transmission of light in water, LIDAR works much better on land than in the ocean, but it can penetrate clear coastal water.  The Shoals LIDAR operated by the Army Corps of Engineers collected data in Hawaii down to 30 m depths in clear tropical waters (Figure 15).  More typical results would have much shallower penetration, and in turbid waters like the Chesapeake Bay, LIDAR would not work for bathymetry.

Figure 15. Point cloud of SHOALS data in Hanauma Bay, Oahu.  

Image: P.Guth, with USACE data

Archaeologists have used terrestrial LIDAR, with a scanning instrument mounted on a tripod and collecting a dense, 3D point cloud of an archaeological site or artefact.  If a shipwreck were recovered and brought to a museum, the complete geometry of the vessel could be rapidly captured with a LIDAR system (Figure 16).  In addition to measuring a large ship, the terrestrial LIDAR could scan individual small artefacts to reconstruct their precise shapes.

Figure 16. Point cloud of Swedish warship Vasa in the Vasa Museum, Stockholm.

Image: P.Guth and E. Ziel, with Vasa Museum data


Color photographs of shipwrecks fire our imaginations, but we often do  not consider that it is impossible to actually see everything in the image at one time.  Because of the attenuation of light in water, seeing a complete wreck requires creating a mosaic with a number of individual photographs (Figure 17).  Geometric distortions may cause overlap and underlap where images must be merged, and color balance can also vary.  Despite these limitations, underwater still and video photography provides an outstanding way to document wreck sites.  Because of the limited field of view, sonars generally remain better search tools when we want to cover a larger area.

Figure 17.  Photo mosaic of 7,000-foot deep wreck in the Gulf of Mexico. Images collected in June 2009 with Woods Hole Oceanographic Institute’s Sentry AUV.

Image courtesy of Lophelia II Team 2009, NOAA-OER.


Integrating and Interpreting Results

The  Oceanography  Department has a sidescan sonar (Figure 18) and subbottom  profiler (Figure 19).  Figure 20 shows a small wreck imaged by the sidescan, with the wreck sitting in the middle of a large area with very low returns shown in black.  The wreck is surrounded by oyster bars with very strong returns shown in white, and the strong returns result from the oyster shells being very hard, having a very rough surface, and being elevated above the bottom.  All of these characteristics can be verified by looking at the subbottom profile (Figure 21).  The GPS data stream from the two instruments allows us to match the corresponding data from the two instruments.  Records 3500 to 3600 on the subbottom correspond to one of the oyster bars, which is in about 6 meters of water, and has a very strong return from the surface and then no reflections at depth because little energy penetrates through the oyster shells.  The muddy bottom areas from records 3700 on about 1-2 meters deeper, and a layer of soft mud (weak or no returns) overlies a much different layer about 1 meter deep.  The subbottom, which only looks below the vessel, did not image the wreck itself because the ship's track was chosen to get a good image with the side scan system.  The use of both sonars greatly increases the knowledge that can be obtained about the wreck and its surroundings.


Figure 18.  Side scan sonar towfish.  This is a shallow water system, and an easy one person carry.
Figure 19.  Subbottom profiler towfish.  This is a shallow water system, and an awkward one person carry because of the weight.


Figure 20.  Side scan sonar image of a small wreck in the Severn River, surrounded by oyster bars.

Image from Smart and Bushong, Fall 2006 midshipman research project.
Figure 21.  Subbottom profiler sonar image in the neighborhood of the wreck.  The Y axis shows depth, assuming that the sound velocity is 1500 m/sec, and the x axis shows the pings which correspond with distance along the ship's track.

Image from Smart and Bushong, Fall 2006 midshipman research project.


Underwater searches, whether looking for shipwrecks, geological features, or engineering site surveys, use the same tools.  Most of the tools, especially those trying to survey a large area rapidly, use sound waves because of the low attenuation of sound in the ocean.  Designing a search requires understanding the characteristics of the potential sensors, and then interpreting the results requires skill and practice in determining what features on or under the seafloor would cause the observed patterns.


   Lesson summary--Click for WMV video clip.

Discussion forum:

Return to course syllabus


Last revision 3/22/2011