The Development of a Towed Vehicle for Optical Mapping in

Shallow Water.

Joseph H.J. Leach Department of Geomatics University of Melbourne Parkville, Victoria, 3052 Australia

Abstract - Environmental mapping in shallow water is normally carried out by divers. However, divers make poor mapping platforms. In order to meet these needs, a semi-submersible imaging vessel was developed. However, its operation has demonstrated some important limitations. The first was that the system could operate to only about 6m depth. The second limitation was that, as the observation pod was hard mounted into a work boat, it was difficult to transport long distances. In order to counter these difficulties, a small towed body was developed. The vehicle is designed to operate from the surface to 70m depth and to provide vertical, stereo video imagery of the sea floor. The whole system is self contained and can operate ofrvery small vessels. The vehicle is easily transportable which will allow us to do international comparisons of functional ecological groups and their relationship to depth and geology.

I Introduction

Environmental mapping in shallow water is normally carried out by divers (Vogt 1995). However, divers make poor mapping platforms. Areas perceived as boring will often receive limited attention while small but visually prominent areas will be covered in detail. The images produced will normally be high detail obliques which can not show the spatial relationships necessary for environmental planning and assessment. Even when this sort of coverage is attempted, there are sever difficulties with stability and navigation.

Routine video transects are a way of overcoming the bias of selective observation since they record a complete record of the transect area. However, such transects need to be precisely positioned and conducted with a known, preferably vertical, image geometry. The navigation and platform stability required to generate the sort of routine, repeatable image coverage required for resource and environmental assessment is probably beyond the capability of diver based systems.

I1 Semi-submersible Imagery

In order to meet these needs, a semi-submersible imaging vessel, the Korrong, was developed (Leach 1995). This is a 5.2m monohull with a manned observation pod which can be lowered to a depth of l m while still remaining attached to the hull. This separates the cameras from surface effects (see Figure 1).

Imaging ports mounted in the floor of the pod allow the use of a wide variety of cameras and, using the fore and lateral ports, the observer in the pod can view ahead and to the side to plan further coverage. Imagery from the camera ports in the pod has a known, normally vertical, geometry and a precise position is available via the DGPS on board the boat. Continuos video transects are the most popular imaging technique used in the pod, although film cameras are carried in applications which need the higher resolution, The video transects can later be compiled to form terrain maps of the sea floor. The observer in the pod can also enter data directly into a GIs.

Figure 1. The Korrong, with pod extended for imaging operations.

0-7803-5045-6/98/$10.00 0 1 998 IEEE 1455

The vessel has been used to map shallow reefs, the extent and type of seagrass cover, the area of infestation of Japanese kelp, and shallow historic ship wrecks. However, its operation has demonstrated some important limitations. The first was that, given the clarity of coastal water, the system could routinely operate to only about 6m depth. This left a gap between the imagery of the semi-submersible and the deeper shelf imaging systems which would typically begin operation at depths greater than 50m (eg Wadley and Barker 1996). Attempts to fill this gap with ROV transects were only moderately successful because of the limited speed of the ROV and the difficulties with stability and control in areas of high current. The small ROV used also had the same imaging limitations as divers. The imagery was normally oblique and the platform stability was poor. The ROV did overcome the health and safety issues involved with using divers but the need to operate relatively high power electrical equipment from open boats in a salt water environment raises other safety concerns.

The second limitation was that, as the observation pod was hard mounted into a 5.2 meter aluminium work boat, it was difficult and expensive to transport long distances. Even though the whole system is road trailerable, this effectively confined the vessel to the inshore waters of South Eastern Australia.

I11 CELPIE Project

In order to counter these difficulties, a small towed body was developed. This was called the CELPIE (Coastal Exploration by Line-towed Platform Imagery Evaluation). This is also the name of an Australian breed of working dog and of the “water horse”, a mythical Celtic monster.

The vehicle is designed to operate from the surface to 70m depth, a range which overlaps both the semi-submersible systems at one end and the deeper shelf imaging systems at the other. The whole system is designed to be as self contained as possible, with each instrument having its own battery power. This means that the CELPIE can operate off very small vessels with virtually no on board infrastructure, although since the submerged vehicle is positively buoyant and uses the speed of the towing vessel to control its altitude, reasonable engine power is required. The small size of the vehicle means that it is relatively easily

transportable. Together these features will allow us to do international comparisons of functional ecological groups and their relationship to depth and geology using only normally available seaborne facilities.

A. Vehicle Description

The vehicle is composed of two tubular sections two meters long and 300” in diameter. These are joined by a curving ramp slope upwards, fore to aft (see Figure 2). The tubular sections contain all the instrumentation while the flat ramp section provides a downward hydrodynamic force and serves as the site of the tow harness attachments. Two fins are mounted bottom aft, one on each of the tubular sections. The system is a bit over a meter wide. This width was necessary for the operation of the camera system (see below) and the length was a consequence of the need,for stability.

Figure Two. Image showing the general layout of the CELPIE.

B. Imaging System

The vehicle is designed to operate from the surface to 70m depth and ‘to provide vertical, stereo video imagery of the sea floor. It has two identical, commercial Hi8 camcorders which are synchronised using their internal time signal. Each of the cameras is mounted in housings in either the left or the right tubular sections. The camera housings are identical. They are made of polycarbonate and each has a flat view port.

The images are recorded on board the vehicle for later viewing. Video tape provides a very data dense storage medium. After viewing, selected frame grabs are made from the tape for further

1456

study. These can be mosaicked to provide a larger area image or analysed to provide 3D information.

The flat view port means that the focal length of the cameras is increased by about 30% (Scoones 1985) and these two cameras operate at an effective focal length equivalent to 67” in a 35” camera. This means that each camera will have an image area of 0.519 times the vehicle altitude. So that at two meters the image area will be about one meter square. At ten meters it would be about five meters square, although it would be rare to be able to operate at that altitude in coastal water. The normal operating altitude is about four meters where each camera images an area of about two meters square.

The cameras field of view overlaps at altitudes greater than 1.7 meters. This overlap can be used to calculate the altitude of the vehicle and is important in working out the scale and area coverage of each image (see Table One). It also provides useful stereo coverage at altitudes of greater than about three meters. While this is only useful for vikual inspection and qualitative description at the moment, calibration of both the relative and internal orientation and radial distortion is planned for the very near future. The procedure to be followed will be that described by Fryer and Fraser (1986). This will ehable detailed photogrammetric studies of the sea floor.

Altitude %overlap Image size(meters square)

4 . 7 m 0 variable 2m 16 1.2 3m 41 1.8 4m 57 2.4 5m 66 2.9 6m 71 3.6 7m 76 4.2

Table One. The imaged area and overlap of the Celpie’s cameras as a function of the vehicle altitude.

At altitudes greater the 1.7 meters, the normal minimum operating altitude, the two cameras also provide a greater area coverage than one camera alone. This effect is most marked at about two meters altitude where the total cross-track image swath is two meters. An increase of about 80% over the swath of a single camera. As the altitude, and area of overlap, increases, the percentage increase in effective swath is reduced, although the total swath increases. At the normal operating altitude of four meters the total image swath is 3.5 meters but

this is only a 63% increase over one camera alone. Along track, any number of frames can be grabbed and mosaicked to provide an image of the desired aspect ratio, although crabbing, or side drift, and the tendency of plants to wave in the current, can sometimes make this difficult (see Figure 3).

Figure Three. Image sequence over an algae covered rocky reef.

1457

C. Vehicle Control

The vehicle is inherently buoyant and hydrodynamic pressure is used to cause it to descend. The rate of descent is controlled by the speed of the towing vessel. The faster the towing vehicle travels the faster the Celpie will descend. If the towing vessel slows or stops, the inherent buoyancy of the vehicle will cause it to rise. Real time control is provided by a video feed from one of the cameras to a head mounted display at the surface. This is used to estimate the altitude of the vehicle and to ensure that the vehicle is towed at a height which is both safe and provides good image quality. A small pressure sensor which can be seen from the navigation camera is used provide depth information. This is automatically recorded on the video tape for later analysis. A combination of depth and altitude data can give a profile of the sea floor along the transect.

The trim of the vessel is controlled by varying the attachment point of the towing harness and the length of the tow line. There are three possible attachment point for the harness, a forward point for deep operations, a mid point, and a rear point for shallow or surface operations. Different tow line length can be used at each of these attachment points.

The position of the vehicle can be calculated using the DGPS position of the towing vessel, the length of tow line, and the depth of the vehicle itself. The accuracy of this positioning is inversely proportional to the length of tow line since the curve and stretch of the tow line are the main sources of error. Using an average DGPS error of five meters and with 100 meters of toe line deployed, it is estimated that the position can be calculated to within 20 meters.

IV Conclusion

Video imagery is a powerful tool in mapping and monitoring the sea floor environment. It is a data dense medium and provides for reliable identification of macro epibenthos. These can be then used as indicator species for the micro and infaunal communities. However, in order to be used quantitatively, the imagery must have a known geometry and scale. Both sonar and leaser methods have been used to achieve this in the past (Davis 1997). These methods are expensive and do not work well in shallow, sunlit waters.

The Celpie project attempted to provide a cheap but effective means of imaging the shallow sea floor. Costs were reduced by insisting that the vehicle be self sufficient in power and data recording. This reduced the cost of both the cable and the surface infrastructure required.

Using two video cameras, as well as providing stereo imagery and the possibility of detailed photogrammetric studies, also eliminated the need for expensive altitude sensors with the degree of overlap between the two cameras providing a very sensitive measure of vehicle altitude.

These features of the Celpie may be particularly attractive to developing countries, where the severity of the marine environmental pressures is often matched by the financial constraints placed on marine monitoring programs. Even in industrialised countries, it provides a tool for mapping and monitoring the sea floor environment across a depth range that is often difficult to study.

References

Davis, D. 1997. h e r Meusure, Users Guide, Version 1.0. Monterey bay Aquarium Research Institute, Moss Landing, Califomia.

Fryer, J.G. and Fraser, C.S. 1986. On the Calibration of Underwater Cameras. PholcJg"?tric Record, 12(67): 73-85.

Leach, J.H.J. 1995. Development of a Semi-submerged Survey Vessel for Environmental Resource Mapping and Monitoring in Port Phillip Bay. Proceedings of thel2rh Austrulusiun Conference on Criustul und Ocean Engineering. Australian Institute of Engineers, Barton A.C.T., pp 305-310.

Scoones, P.J. 1985. Designing Underwater Cameras. In Underwater Photogruphy und Television f o r Scientists, Underwater Association Special Volume No.2. George, J.D., Lythgoe, GI., and Lythgoe, J.N. Eds. Pp 82-87. Clarendon Press, Oxford.

Vogt, H. 1995. Video Image Analysis of Coral Reefs in Saudi Arabia: a comparison of methods. Beitr. Puluont., 20:99- 105, Wien.

Wadley, V. and Barker, B. 1996. Video and Still Photography of the Sea Floor Habitat of the South East Fishery. Abstract Vol. AMSA 96. Australian Marine Science Association, Hobart.

1458