Alan Drayson and Vincent May

Introduction

For most people, the world beneath the surface of the sea is inaccessible both because we are not naturally sea creatures and because the science is often difficult to understand. The sea has always been regarded as a place of mystery and danger, as well as a place rich in resources. For most of human history, little was known about what lay beneath the sea’s surface. Information about the oceans was mostly confined to their shores and until the medieval period there were few documents which described the oceans. Exploration of the seas was accompanied by the development of charts (See Theme 1 Topic 4 Dorset’s Underwater World).

Most of us have little idea of what the seabed looks like, apart perhaps from images of brilliantly coloured coral reefs. However, technology allows us to visualise a world in which sound is the main means of communication and investigation. This requires several stages including seabed survey , post-survey processing of very large amounts of data, constructing an integrated picture of the whole area surveyed ( mosaicing ), describing and classifying the features, constructing a 3-dimensional model ( visualising ) and finally interpreting the story which the images tell.

The earliest charts were drawn to help navigation along the coast. Sailors needed to find anchorages and avoid reefs. The earliest charts have very few records of depths. Navigators used charts and other information, known as “Pilots”, to find their way along coasts. They often provided drawings of the coastal landscape from the sea. This unusual perspective of the landscape was, of course, the one which mattered to the navigator, for it was by the landmarks that he marked his progress and avoided the hazards of the rocky or shallow shore.

As maritime trade expanded, identification of these hazards became more important, especially around ports. Information about water depths became more important as ships with greater draughts entered ports that had been used for centuries by shallow draught vessels. Hydrographic charts provided more detailed bathymetric (depth) information. Depths were measured using a leadline lowered to the seabed whilst the position of the ship was fixed using a sextant or astrolabe. As the oceans were explored, individual soundings revealed their great depths, but little was known about their detailed topography. The ocean floor was often regarded as a flat plain.

Today, echo sounders and satellites provide electronic depth and position information. Provided the extent of potential error is accepted, they provide remarkably consistent means for navigators to know where their ship is and how much water is under its keel. Ships still run aground, however, even when submerged rocks are well known. However, because traditional charts were based on spot or transect measurements, not all features are shown on charts. In shallow waters, this can mean that rock pinnacles exist which have never been mapped – usually they are only recorded when a boat hits them or nets are snagged by them. Modern technology using sonar allows a more comprehensive picture of the seabed to be produced, but such images are not widely available.

Chart of the coast between Portland and St Alban’s Head 1937 (DCM)

This chart of Weymouth Bay dates from 1937 and shows very well the sparse information about the seabed compared to the detailed depiction of the land area. The topography of the sea bed is shown simply by the bathymetric contours of depths beneath chart datum. Chart datum is usually the height of the lowest astronomical tide, namely the level when there is least water available beneath a ship. Individual depths are shown wherever a sounding had been taken. In addition, there is usually a letter code alongside the depth to show if the bottom was hard ground or softer sediments in which an anchor might hold. Where the water becomes shallower, soundings are closer together and so banks (such as the Shambles or Lulworth Banks) and the nearshore seabed stand out simply because the points are closer. Away from the shore, in deeper water the soundings are fewer and the variability of the seabed less fully described.

As submarine warfare developed and exploration of the continental shelf for oil and gravel increased, it became essential to know what the shape of the seabed was. Submarines could only navigate at depth if their navigators knew with confidence the shape of the seabed over which they were sailing. This requires three essential pieces of information:

· the position of the ship (latitude, longitude and depth),
· the depth of the seabed beneath the ship and
· the shape of the seabed.

If the navigator has these, the ship can be “flown” over the submarine landscape in exactly the same way as the pilot of an aircraft can navigate safely through a mountain range. For the identification and assessment of submarine resources such as aggregates or oil and natural gas, it is also necessary to know what lies beneath the seabed. As a result, seismic and sub-bottom profiling methods have been used to investigate both shallow sediments and the deep submarine geology. However, more recent technology allows much more detailed description of the sea bed so that seabed hazards such as landslides or the seabed habitats can be described in similar detail to features on land.

We are able to detect many quite subtle features of our terrestrial landscape because they are outlined by shadows. So in the colour image below, we can see a set of sand ripples, each of which is visible mainly because of the contrasts between lit and dark areas with a very low angle of sun lighting them.

Sand ripples on the shore, lit from the right (Drayson 2004)

On the seabed, we can light-up similar features by using sound to produce patterns of lit surfaces and shadow. The second image is also of ripples, but this time on the seabed in about 10 metres of water. The seabed has been ensonified rather than photographed. Unlike true pictures which rely on reflected and retransmitted light, these “pictures” rely on sound reflections from the sea bed.

Sidescan sonar image of seabed sand ripples in Weymouth Bay (Drayson 2004)

The development of echo-sounders, sonar, aqualungs and more recently underwater cameras and remotely-operated vehicles (ROVs) have brought about a rapid and large increase in information, most of which reaches the public as still or moving pictures. As a result, our view of the ocean depths is predominantly visual. However, most of our knowledge about the shape of the seabed is produced using sound and converted to visual images. Seabed mapping depends on sound. But why use sound? Although humans have very limited use of their senses underwater - both visibility and hearing are very poor - we can use electronically generated sounds to describe the sea bed very well, provided they are converted into pictures which we can interpret and compare (See Theme 1 Topic 4 Dorset’s Underwater World)

Sonar is widely used for mapping the seabed. Sound travels very efficiently through water, although its density, temperature, pressure and chemical composition affect the transmission of an acoustical pulse through water. The general speed of sound through water is 1495 metres per second (ms-1), but it can be as low as 1410 ms-1 in Arctic waters and as high as 1540 ms-1 in warm equatorial waters. Sounds occur naturally through mechanical disturbances of water such as waves or cetacean sonic pulses. The submarine natural world generates many different sounds, some vital for communication between marine mammals. Discrimination between these natural background sounds and deliberately transmitted sounds is essential. The use of acoustic energy for surveying is structured and follows distinctive patterns. Sonar makes use of repeated and predictable signals. The frequency of the pulse is defined as the number of times a signal completes a full sinusoidal wave in one second, measured in Hertz (Hz). Design of surveys depends on a structured choice of frequency which balances the conflicts between attenuation (energy loss) and the requirements of range and resolution.

Sound emitted by a transducer can determine the range of objects (typically the depth of the seabed). Knowing the speed of sound through water, we can measure the time taken for the echo to return to the transducer, and so estimate the distance from the transducer to the seabed. Sidescan transducers produce a cone of sound directed towards the seabed, but the shape of the cone differs between different sonar systems. A broad cone will result in a large area being ensonified and the resulting image will have a low resolution. Narrower beams will result in a greater spatial resolution and so provide more detail.

An analysis of the echo pattern may also be used to gain some information about the nature of the seabed. The power and frequency of the sound will also determine the extent to which the signal penetrates the surface sediments and reflects from deeper buried features. As a result, sonar systems can produce images of the seafloor based upon the reflective characteristics of the surface and the equipment used. It is possible to distinguish, for example, between sand ripples and boulder fields. Resolution can be good enough to detect the line between lobster pots. Some images, such as ripples, are easily recognised because they are familiar. Others are much more difficult and require specialised training and experience.

Until the later twentieth century, information about seabed topography and sediments was largely dependent on slow and expensive sampling by divers or grabs and corers. Since the 1940s, rapid advances have been made in the use of sound to map the seabed. Acoustic remote sensing provides a cost-effective approach to submarine mapping. Rapid, remote acoustic survey techniques are increasingly being used to study marine environments and have many applications for hydrographic and marine geological mapping.

Echo sounders and side-scan sonar differ in the data they provide. Side-scan sonar does not provide depth information and is commonly used in conjunction with a single beam Acoustic Ground Discrimination System (AGDS). AGDS is able to discriminate between different sediments and bathymetry, at a low resolution. Conversely, side-scan sonar offers high-resolution images of the seabed, with a much better coverage, but is largely confined to topographic features with limited measurable point data on sediment characteristics. AGDS and side-scan sonar are complementary.

AGDSs use an echo sounder to generate a signal and then analyse the returned echo. They measure water depth under the vessel and the reflective properties of the seabed, which indicates material type. As there is a strong correlation between the strength of acoustic return and the type of material, the strength of acoustic signature can be used to classify the bottom type. A number of methods have been developed to infer bed form roughness based on the backscatter energy from the echo. Burns et al (1985) and Chivers et al (1990) devised a system which based the classification on the first and second echo. The RoxAnnTM system is an

The pattern of sonar echoes from the seabed (Drayson 2004)

extension of this technique using digital technology to assess the first returned echo and also the second return.

When a sound pulse is transmitted from an echo sounder or sonar, it is reflected from the seabed, usually with two separate signals (‘echos’). The diagram above shows the signal from a single line of side-scan data. It shows the transmitted pulse followed by two separate pulses, usually identified as the first return (or Echo 1) and the second return (Echo 2). Both are lower in amplitude than the transmitted pulse because there is some loss of energy during travel through the water and because of absorption and scattering by the reflecting surface. The first return is the reflected signal from the seabed, the second is the signal reflected from the seabed to the sea surface and back again to the ship’s receiver.

In the diagram above, Fig. a i represents the initial reading for the interference beneath the transducer. Fig. a ii is the first echo from the seabed also represented in Fig. b. This initial reflection is generally composed of three sections:

1. the initial reflection from the seabed directly beneath the transducer
2. backscatter from the area surrounding the point of first contact
3. reflections from the sub-bottom

One widely used AGDS system, RoxAnnTM, employs signal processing software to select two elements from the echos and measure the value of signal strength and time. RoxAnnTM uses these values to classify the seabed. The first selected section of the echo (Fig. a ii) is the decaying echo after the initial peak in strength. This measure of the time/strength of the decaying echo is termed ‘Echo 1’ or E1 and is taken as a measure of roughness of the ground. The beam width of the sounder is important for E1 since a wide beam will give greater scope for measuring signal decay than a narrow beam. For this reason, the manufacturers of RoxAnnTM recommend that the AGDS be operated with a sounder of moderate beam width (15-25 degrees) (Rukavina, 1997).

The second segment registered by RoxAnnTM is the whole of the first multiple echo, (Figs a iii & c). Multiple echoes are produced when the first echo is sufficiently enough to rebound off the water/air interface and reflect for a second time off the seabed. The characteristics of the second echo are slightly more complicated. The dominant ray paths of the second echoes undergo two reflections at the seabed and a single reflection at the sea surface. The amount of reflection is related to the acoustic differences between the seawater and seabed. Consequently, the harder the seabed, the more acoustic energy is reflected to the surface and back to the transducer. The reflected acoustic energy is a more sensitive value of ground hardness than the first return and is termed ‘Echo 2’ or E2.

In summary:
· The acoustic reflections/backscatter from the first echo is a diagnostic of the seabed roughness (E1).
· Reflection information from the second echo determines seabed hardness (E2).
· Together roughness and hardness provide a reliable indicator for seabed classification

AGDS Survey Design

The image shown below was captured using a RoxAnnTM signal processor, operating at 200kHz, and a Koden CVS-8112 echo sounder with a beam of 18o. Post processing was completed with the services of SeaMap from Newcastle University. The computer also used Microplot navigation software to log the ship’s position.

An example of an active screen during AGDS data collection. The tracks of the survey vessel are shown, as is an initial classification of the sediment type (indicated by the different colours (Drayson 2004)

The location of the AGDS data is collected using the survey vessel coordinates (taken from the dGPS). The combined continual changes of GPS, footprint size and time gap between saved datasets determine the maximum resolution of the AGDS. Thus, for a vessel operating in 10 metres of water, at 10km/h, with a beam angle of 18o and a dGPS error of 10 m, a set of E1, E2 and depth values saved every 2 seconds would be expected to represent an area of approximately 13 metres wide by 20 metres long. Any increase in depth, speed or save rate would reduce the resolution.

Acoustic data from the AGDS is built up as the survey vessel tracks back and forth across the survey area. All AGDSs have the disadvantage of limited coverage because the data is essentially point information along tracks with no data collected from the spaces between the tracks. Thus, for most AGDS surveys utilising an echo sounder, there is significant averaging between tracks. However, because the echo sounder only produces readings for targets directly beneath the sonar, extrapolation between survey lines is used to construct area coverage maps. Track spacing is chosen to provide the best resolution possible within the constraints of the survey timetable. The Weymouth Bay images are based mainly on North-South tracks in order to record changes in bathymetry and potentially in sediment type. East-West tracks were used for quality assurance, using intersected data. The same pattern was used for the side-scan survey, with some N-S tracks in order to reach closer inshore and also to protect the “fish” from any potential damage due to rapidly shallowing depths in the E-W direction.

AGDS coverage over the survey area. This shows each track across the survey area and also indicates where it was not possible to obtain acoustic data, because the areas were too shallow for the survey vessel (Drayson 2004)

Ground Truthing

Just as an aerial photograph can be checked by visiting the site on the ground , so submarine surveys can also be checked by “ground truthing”. Ground truthing helps, for example, classification of the acoustic AGDS data.

Video camera and mounting used to ground truth side-scan sonar and AGDS images (Drayson 2004)

For the Weymouth Bay survey, video recordings of between 2 and 5 minutes, depending on the extent of passive drift, covered between 50 and 150 metres of the seabed. The position of each track was registered and plotted. An operator then viewed the footage and assigned a classification to be associated with that specific acoustic signature. The video sampling points were evenly distributed within the survey area. In addition to the camera data, free-swimming divers were used to obtain a substrate description of the seabed. The location of the diver entering and leaving the water was noted and information fed into the GIS database.

This image shows an area of boulders scattered across the seabed as seen by side-scan sonar (Drayson 2004)

This image, in contrast to the one above shows boulders as seen by the ground-truthing video and free-swimming divers (Drayson 2004)

Sidescan Sonar

Side scan sonar usually uses equipment towed behind a survey vessel. Sound emitted by transducers on the towed equipment and reflected back from the seabed is received by the “fish” and transferred to signal processing equipment aboard the survey vessel. The sidescan images of Weymouth Bay were captured by an Ultra WideScan system linked to an Octopus Euterpe 460 digital acquisition system, which provided full geo-referenced data capture.

Example of the imagery seen during a side-scan sonar track (Drayson 2004)

Each transducer emits pulses of sound, which are ‘fan-shaped’, being wide in the port and starboard plane but very narrow fore/aft. A single pulse results in an echo from the seabed, which becomes more distant as the slant angle (the angle at which the sound meets the seabed) increases. The boat’s surface unit displays the vessel position with the changing recorded echo strengths presented on each side of the computer monitor. With each new pulse, an image of the sea floor is built up line by line. The intensity of the echo depends on reflectance, which in turn relates to the topography. Surfaces angled towards the “fish” produce a strong echo whilst surfaces hidden from the ‘line of sight’ of the fish result in a shadow or area of weak backscatter

The typical image comprises three spatial elements:
· the side-scan image to port (left) of the boat
· the side-scan image of starboard (right ) of the boat, and
· a central area in which the water column beneath the boat is recorded. This is an echosound trace and shows the variations in elevation of the seabed beneath the “fish”. In the image above the seabed rises towards the centre of the image.

Because of the way in which shadowing occurs, shadowed areas will be to the left of the features on the port side and to the right of features on the starboard side.
For the side-scan survey, the vessel followed predetermined tracks, with a lane spacing (distance between two successive tracks) chosen to obtain a complete data set or overall image of the seabed, with no data gaps. With a side-scan swath of approximately 200 metres (depending on water depth), tracks 150m apart create a 50m or 25% overlap of the previous swath, enabling complete coverage. A lane spacing of 75% of the swath width results in a seabed coverage of 150% on all but the first and last lanes.

Post-processing analyses and corrects potential errors contained within the collected acoustical data. It also enables data sets to be combined or mosaiced into a single dataset. Survey depths were corrected to chart datum by applying corrections calculated from a tidal prediction program using the simplified harmonic method produced by the UK Hydrographic Office (Anon, 1991). The corrections are applied at time intervals of 10 minutes. The resulting depths may differ slightly from the soundings on the charts, because atmospheric conditions affect tides.
Corrections are also necessary for
· Positional uncertainty,
· Erratic changes in depth, and
· Erratic changes in RoxAnnTM E1 and E2 (roughness and hardness).

Octopus side-scan-editing suite software was used to edit and analyse the side-scan data. Processing includes:-
· Variations in Time-varied Gains which are used to increase or decrease the selected areas of the returned acoustic reflection.

· Integration of laid-back position of the towfish (this is required as the dGPS coordinates relate to the position of the vessel and not the towfish)

· Incorporation of slant range data, which enables the removal of the central track of the side-scan data. This relates to the two way travel time taken for the acoustic pulse to reach the sea floor and return to the transducer.

The side-scan-editing suite tracks the acoustic signature, which represents the seabed by selecting the highest strength of the first return as being the bottom. An “S” within a, b & c demonstrates this. A smaller return was recorded due to interference from the sea surface. This is common in shallow water surveys where the “fish” is relatively close to the surface and is clearly seen on the sonographs from the survey area. This is demonstrated with an “I” within a, b & c. The final processed data is then mosaiced to enable the side-scan data to be visualised, analysed and interpreted in its entirety.

The height, spacing and orientation of the ripples provide information about the direction and strength of the currents which produce them. The image on the left is a sonar image, that on the right is a photograph of sand ripples which provides ground truthing for the sonar image

Constructs an integrated picture of the whole area surveyed.

Each individual track of data only represents a small area in comparison to the entire survey area and so a single mosaic is required for interpretation. The mosaics are generated as geo-referenced images of the seabed, enabling features to be measured and their location determined directly from the image. The mosaics are produced using specifically designed software to view side-scan tracks in unison. The software positions the boat and subsequent seafloor by including a figure for the layback of the “fish” being towed behind the vessel. The production of a complete image is possible because the survey tracks overlap. The inherent accuracy of the GPS results in an error value between 0-15 metres on the position in the “fish” and hence that of any seabed features.
Because the direction of the vessel was recorded during the survey, the direction of the shadow from any seabed features can be allowed for. This is particularly important when joining parallel tracks, as some features may become ensonified along the port side and then the starboard or vice versa. The post-processing software resolves these ambiguities by adding an algorithm that balances any abrupt changes.

In this image, you can see a series of displacements. Just as on land, the orientation of the individual faults can be measured. The image is made of several mosaiced tracks. The blurred areas running horizontally across the image are the area beneath the survey vessel and indicate the survey track.

This image, just off Worbarrow Tout, shows a detail of the mosaic above. The low sloping ridges, steeper on one side than the other, are displaced in the centre of this image. The displacement is about 12 m. The dark horizontal band across the centre of the image is a result of data distortion.

With numerous sources of potential error at different stages of the survey and subsequent analysis, recognising the accuracy and spatial resolution of the data is very important for subsequent interpretation.

Positional Accuracy: The positional error based on the dGPS system used is considered to be between 3 and 15 metres (Gilbert, 1999). The position of the ground-truthing or sampling device deployed from the boat is also prone to uncertainty, due to the movements of the boat and the position of the sampling device relative to the boat. The length of cable used determines the position of the side-scan fish. The uncertainty of the ground-truth device is much greater due to the increased length of umbilical cable, which carries the video feed. The actual position of the underwater camera depends on the speed of the boat and also passive drift levels. When a video camera is lowered, it may be up to half the water depth again away from the boat. Thus, the true position of the camera might lie anywhere within a circle with a radius equivalent to dGPS error and half the water depth (Foster-Smith and Sotheran, 1999). For example, if a video camera is lowered into 20m of water, and the positional accuracy is 5m, the true position of the video might lie anywhere within a 20m radius from the recorded position of the boat.
Resolution: The size of the acoustic footprint on the seafloor and the positional accuracy of the track data limit the spatial resolution of the RoxAnnTM data. dGPS was available for both of the acoustic surveys and with careful post-processing, the positional accuracy of the track data are estimated to be between 3m and 15m. With the average depth between 10-20m, the resolution of data varies between 13 by 20m to 15 by 23m. The use of interpolation also fundamentally incorporates error, as areas with no data are inferred by the utilisation of mathematical algorithms of surrounding data. Nonetheless such procedures are required in order to obtain data over a broad area.

The spatial resolution that can be obtained with side-scan is dependent on the actual system set-up and the towfish stability during data collection. To maximise spatial resolution, a system that could distinguish features as small as 0.2m was chosen. Moreover, fixed procedures were adhered to during data collection. Towfish instability has distinct effects on the sonograph, thus, during the interpretation stage, features which indicate the effects of towfish instability, require further investigation. Any anomalies need to be taken into account during the interpretation and it is the role of the interpreter to recognise such anomalies.

Methodology for Validation / Interpretation

In order to interpret and visualise the data, a GIS database was constructed. As the data collected have a spatial component, GIS allows large datasets to be managed, analysed and viewed. In addition to efficient data storage, GIS and the associated packages used within this project allow for much more complex simulations and visualisation, aiding in the interpretation of the data. The GIS assists in the analytical procedures and image formulation of the survey data.

The images produced and catalogued within a GIS require careful interpretation and analysis. In order to classify the data, a degree of validation is necessary. This is based on a number of techniques and processes, which enable the images to be classified with a degree of confidence, including:
· Repeatability of image recognition,
· Image comparison to coastal form,
· Physics of acoustics,
· Assessing multiple data sets,
· Experience, and
· Ground truthing.

Repeatability of image recognition is an important technique in classifying marine topography. Similar and repeated surveys (e.g. Heeps 1986, 1987; Donovan and Stride 1961; Knebel et al 1999) provide a basis for comparable analysis of similar patterns. Image recognition techniques are also based upon comparable features seen within the intertidal and terrestrial environment. This procedure is similar to the practice of recognising features on other planets based upon recognition of similar patterns on earth. Such validation is, in many cases, the only form of information for broad scale terrain analysis.

Side-scan sonar image of a near-horizontal rock ledge, crossed by several series of discontinuities. Many are linear, but others are more complex in shape. At the top of the image the surface changes from the ledge surface to its edge where there is a scattering of separate blocks and then to an area of sand ripples

This image is of a small part of a similar stratum exposed at the shoreline. From the bottom right-hand corner, one of the larger discontinuities runs up the image to disappear as the ledge is broken into small blocks. On the left-hand side of the image the surface of the ledge is broken by many smaller cracks.

This side-scan sonar image shows a series of strata which have been planed across by the sea. Their height and width varies depending on the materials which form them and their angle of dip

Here a similar set of eroded strata are exposed at the shoreline and extend out under the sea

Constructing a 3-dimensional model (visualising)
Once all this imagery is compiled and analysed, it possible to construct images of the seabed at different scales which allows us to see the landscape beneath the sea

Imagery can be used to create a portal within which complex systems, consisting of vast quantities of spatial data, can be efficiently and effectively presented. The benefits are wide ranging, offering advantages to image interpretation and an unprecedented variety of visualisation. Visualisation and the field of cartography are clearly suited to the benefits of using colours and shapes to represent topographic form. The use of images and science within cartography has long fashioned the nature of spatial information, its creation, its communication (see Cartography within Interactive Maps) and crucially, its interpretation. Modern technological advances, as epitomised by GIS, provide a flexible and powerful platform for scientific investigation, whilst providing an ever-wider range of artistic and intuitive methods of representation. Significantly, however, the evolution of modern cartography continues to be governed by the same fundamentals that helped to shape all preceding cartographic accomplishments, namely the need to communicate vast quantities of information concisely and efficiently.

3D bathymetry of Weymouth Bay, based on digitised 5 metre contours. Vertical Exaggeration x 3
(Drayson 2004)

Combining Coastal and Marine Data Models

By amalgamating multiple separate datasets from two unique environments, which meet at the coast, it has been possible to classify the offshore geomorphology and reveal potential relationships between on-shore and offshore geomorphological features. To do this, several forms of information, including LiDAR, bathymetry, aerial photographs, classification maps, side-scan sonar and a variety of terrestrial maps need to be combined.

3D view of Worbarrow Bay (LiDAR)

This enabled the production of a virtual terrain which advanced computer technologies and 3D software can seamlessly rotate and move through at any pre-described route, distance and height. The production of an advanced virtual “fly through” is now possible. This imagery is unique in its ability to enable further calculations to be performed within the GIS.

The images allow the users (specialist or non specialist), to access large volumes of data in a simplified form. It is hoped that this informative and flexible view will encourage users to have a more integrated and holistic approach to visualising the seabed within the coastal zone.

3D view of Worbarrow Bay (LiDAR, aerial photos and sidescan)

3D view across Worbarrow Bay (LiDAR, aerial photos and sidescan)

Interpreting the story

We are looking across the seabed from just south of Mupe Rocks, east of Lulworth Cove, into Worbarrow Bay. If the sea drained away we would be able to walk easily across most of this landscape. We would have to scramble of a ridge about 9 metres high in the middle of the bay where the strata which form Worbarrow Tout (the hill in the middle distance) cross the bay to Mupe Rocks). The image above shows very well how the landscape, exposed for thousands of years when sea-level fell during the last glacial period, was gradually trimmed by the rising sea level. It is mostly a gentle landscape on which there are low cliffs, rarely more than a few metres high, sandy ripples and boulders. Only as you reach the southern edge of Weymouth Bay do the features become much larger.

The imagery shown here uses an innovative approach to integrating numerous forms of terrestrial and marine spatial information. Using historical, topographical, geological and structural maps and when combined with elevation models, a variety of imagery has been created. The resulting datasets and images can be produced for interpretation purposes, to a range of audiences and for a range of tasks. Innovative images, combining multiple forms of spatial data, have enabled a novel form of data visualisation. The images allow the users (specialist or non specialist) to access large volumes of data in a simplified form. This informative and flexible view provides opportunities for a more integrated and holistic approach to visualising the seabed within the coastal zone. The UN predicts that a staggering two thirds of the world’s population will live within 60km of the coast by 2020 (Monahan et. al., 200l). When these demographic changes are combined with the potential consequences of global warming and risk of both gradual and very sudden changes of sea level, there is an obvious need for accurate and seamless maps of the coastal and inshore zone. The methodology displayed here is a contribution to that demand.