This article was downloaded by: [Colorado College]On: 28 October 2014, At: 17:17Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Spatial SciencePublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/tjss20

The korrong project – semi‐submersibleimaging for environmental mapping inshallow waterJ.H.J. Leach aa Department of Geomatics , University of Melbourne , Parkville,Victoria, 3052 E-mail:Published online: 13 Aug 2010.

To cite this article: J.H.J. Leach (2006) The korrong project – semi‐submersible imaging forenvironmental mapping in shallow water, Journal of Spatial Science, 51:1, 133-142, DOI:10.1080/14498596.2006.9635069

To link to this article: http://dx.doi.org/10.1080/14498596.2006.9635069

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

SPATIAL SCIENCE Vol. 51, No. 1, June 2006133

The Korrong Project - Semi-submersible Imaging forEnvironmental Mapping inShallow Water

J. H.J. Leach

In 1996 a unique vessel with an imaging pod thatdescends beneath the surface was developed to imagethe marine environment. The system is used to takephotographs, make video transects, and to allow humanobservation. Water clarity normally limits the vessel’soperation to less than six metres depth. Video transectsare the most common usage of the vessel. Using thevessel, long distances are covered by video transectrapidly and cost effectively. Frames from thesetransects can be used to create mosaics or stereo pairs.Designed as a research tool, the vessel has proven itselfa valuable operational asset in monitoring the shallowmarine environment and demonstrated the technique ofsemi-submersible imagery.

J.H.J. LeachDepartment of GeomaticsUniversity of MelbourneParkvilleVictoria 3052leach@unimelb.edu.au

INTRODUCTION

On land, optical remote sensing is an importantoperational tool in the monitoring of agricultural andother environmental resources. However, because ofsurface reflection effects and light absorption by thewater column, remote sensing systems suffer from aninherent ambiguity when interpreting the signalreturned from the sea floor. In addition, beyond a depthof 5m, there is a significant drop off in signal strength,although in exceptionally clear water, data can bereturned from depths of up to 20m (Jupp, 1988). Theaccess difficulties experienced by most marine studies,in terms of both cost and safety, mean that remotesensing technology has the potential to be a valuabletool in shallow water benthic mapping. Yet the task ofmonitoring our benthic marine resources, at least inthe coastal zone, is currently carried out either by insitu sampling (Skewes and Long, 1994) or by sonarmethods. In situ sampling is expensive, dangerous, andprone to statistical aberrations, while sonar methodsare difficult to use in shallow water and can also sufferfrom serious ambiguity problems in mapping biologicalresources. Routine and predictable sea floor imagingwould be a powerful tool in the inventory for mappingand monitoring benthic biological resources. Semi-submersible imaging offers one solution in theimmediate subtidal zone. The Korrong Projectdesigned a vessel (Figure 1) that would enable suchimagery to be acquired over selected sites.

Dow

nloa

ded

by [

Col

orad

o C

olle

ge]

at 1

7:17

28

Oct

ober

201

4

Vol. 51, No. 1, June 2006 SPATIAL SCIENCE134

Figure 1. The Korrong, a semi-submersible imaging vessel with observation pod in the raisedposition forward of the helm.

OTHER DATA SOURCES

Hand held photographs taken by divers during samplingexcursions are normally high detail obliques whichare useful for illustration and sample documentationbut which lack a standard geometry and are, as aconsequence, difficult to place in their spatial context.Even when there has been an attempt to create an areacoverage using hand held diver photography (eg.Staniforth and Vickery, 1984), the lack of precisionin image location, spacing and geometry means that thevertical, overlapping coverage familiar in terrestrialapplications is not available. Also, diver based mappingtends to suffer from selective observation. Interestingareas of rocky reef are often investigated in detail whileadjacent areas of sand covered flat, which may be justas significant ecologically, may be given only cursorycoverage. The use of randomly placed sample quadratsor defined diver transects can overcome this problemto some extent, but such procedures still do not give aclear picture of the spatial relationships andinteractions between the various ecological units. Thismeans that, while diver based surveys are still theonly effective way to do detailed ecological studies in aparticular area, they suffer from serious limitations

when involved in resource mapping and monitoring,particularly if large areas are involved.

Acoustic methods are the traditional way of acquiringdata (including imagery) over large areas underwater.Acoustic systems can operate to the greatest depths ofthe ocean basins and provide continuous or nearcontinuous cover over a large area. However, since itsmedium is sound rather than light, it will giveinformation on the physical state of the sea flooraveraged over a fairly large footprint and often,depending on frequency, to some depth in the sedimentcolumn. It does not provide clear, unambiguousinformation on the nature and health of biologicalmaterials. Even those systems designed to classify thesea floor either do so on the basis of the assumptionsmade during their calibration (Roxann) or they producean unsupervised classification which must be groundtruthed by other methods (Questor Tangent). In eithersystem there are ecologically significant differencesin bottom type which may produce identical acousticsignals (eg. bare rock/rock with thin sand cover/rockwith encrusting algae). These ambiguities will varydepending on the frequency of the system, with higherfrequency and multi-frequency systems performing

Dow

nloa

ded

by [

Col

orad

o C

olle

ge]

at 1

7:17

28

Oct

ober

201

4

SPATIAL SCIENCE Vol. 51, No. 1, June 2006135

better. However, some level of ambiguity, particularlyin the mapping of biological resources, is probablyinherent in the use of sound as a sensing medium.

Satellite and airborne imaging systems can providegood imaging results, particularly in the first fivemetres (Honey and Byrne, 1977). However, they arerestricted by the absorption of light in the water column.This absorption is wavelength dependent and, sincecolour is used in feature classification, the ability todetermine the nature of the sea floor drops off evenmore sharply than the light intensity. To be useful inmapping, these images also need extensive groundtruthing. Currently, this is most commonly providedby divers. Satellite and airborne imaging systems alsoloose much of their energy from surface reflection(Gross, 1977) and are not usually optimised for theblue/green window frequencies which allow greatestwater penetration. They have a valuable role inproviding broad area coverage but they are not able toprovide precise, high resolution imagery optimised formarine applications.

SHALLOW MARINE IMAGING

The target of optical undersea imaging is light reflectedfrom the substrate. In order for a sensor to receivethis light, the light must twice pass through the watercolumn and cross the air/sea interface. During thispassage the light signal is attenuated by reflection,scattering and absorption. Semi-submersible imagingcan overcome the signal loss from surface reflectionbut it is still limited by the absorption and scatteringin the water column.

The signal loss due to the incident light path beingpartially reflected at the air/sea interface can beconsiderable. The degree of this reflection depends onthe sun angle and the sea state (Gross, 1977). It mayvary from as low as 2 percent, on a calm day at noon, toalmost total reflection in the late afternoon or earlymorning. A typical loss figure in the mid-latitudes wouldbe around 30 percent. A similar reflection will takeplace on the underside of the air/sea interface duringthe return path. Sea state increases the degree ofreflection by locally increasing and decreasing the sunangle in a semi-random manner, and by the albedo effect

of breaking waves. This effect can increase surfacereflection by as much as 50 percent (Lauscher, 1944).This reflection not only causes a loss of incident radiationand radiation reflected from the target, but morenotably, the reflected light acts as signal noise in theimage. The image noise caused by surface reflection hasproved a major barrier to airborne underseaphotogrammetry (Tewenikel, 1963) and while it canbe compensated for by digital image processing (Jupp,1988) this compensation causes a further loss of signal.Semi-submersible systems can be adjusted to operatein the lower ambient light conditions caused by thissurface reflection, although this may significantlyreduce operating hours in the winter months, and withthe cameras placed below the water surface, thereflected light does not itself interfere with the imagery.

Scattering from within the water column has asimilar effect on the image signal to the interfacereflection. The extent of this scattering varies widelydepending on water condition, particularly on theamount of suspended matter and on its particle size.The water column itself absorbs light preferentially atthe longer wavelengths. Clear oceanic water starts toabsorb strongly at wavelengths greater than 550 nm(Neumann and Pierson, 1966). In coastal waters thesituation changes. Coastal water is more likely to havehigh concentrations of photosynthetic plankton. Theresulting chlorophyll concentration absorbs the blueportion of the spectrum with the result that maximumlight penetration is shifted towards the green and reducedin value. Inshore coastal waters may also contain aconsiderable amount of suspended terrigenous material.This material absorbs light in both the blue and greenportions of the spectrum. In such turbid environments,imaging by any means other than sonar is likely to beineffective. The preferential absorption of the longerwavelengths means that real colour imagery is notavailable to an ambient light system beyond the firstfew metres and that all such imagery rapidly becomeseffectively monochromatic with depth.

THE KORRONG PROJECT

The aim of the Korrong Project was to design an imagingplatform which could provide the same quality ofmapping imagery in shallow water as that available on

Dow

nloa

ded

by [

Col

orad

o C

olle

ge]

at 1

7:17

28

Oct

ober

201

4

Vol. 51, No. 1, June 2006 SPATIAL SCIENCE136

land. The project needed to design a platform that issufficiently flexible to allow for a variety of camerasto be used in different mapping applications. The centralKorrong platform is a 5.2m long aluminium monohullcustom built as a semi-submersible vehicle forsubmarine imaging. The vessel can be considered asconsisting of a fore and aft bay, separated by the helmand console. The aft bay contains the navigation andcontrol positions while the forward bay contains theobservation pod, its lifting frame and the moon pool.The vessel is equipped with a Differential GlobalPositioning System (DGPS) and a single frequencyecho sounder.

The most important component of the Korrongsystem is the retractable observation pod which islowered into position through a moon pool (Figure 2a).When fully extended, the floor of the pod is one metrebelow the surface. The observation pod is designed tocarry an observer and is equipped with three viewports to provide forward, port and starboard viewing.There are four camera ports in the floor of the pod(Figure 2b), each of which has a generic camera mountassociated with it. The system was designed to beflexible enough to take any reasonable imaging systemwith only minimal modifications.

The vessel’s imaging ports were originally of a domedesign to minimise refraction effects and preserve thearea coverage of the lens system. Using dome portspreserves the angular coverage of the lens system by

Figure 2a. The Korrong with observation pod inthe raised position

Figure 2b. The Korrong with observation pod extendedfor imaging operations. (Photo courtesy of R. Roob,

MAFRI)

eliminating the refraction at the air/water interface.A major drawback of the dome port is that theconsequent plane of focus is curved. This makes focusdifficult and visual observation through the portsvirtually impossible. In effect, the dome port becomesa part of the lens system of the camera and the cameramust be precisely positioned to correct for this. Thisis difficult to achieve with different cameras being usedfor different projects. As a result, the front cameraports were enlarged and replaced by flat glass whichallows greater operational flexibility.

The trade-off in using a boat mounted imagingsystem, where the target will certainly be less than20m away, comes in area coverage. At a depth of 20m,a standard 35mm imaging system viewing through

Dow

nloa

ded

by [

Col

orad

o C

olle

ge]

at 1

7:17

28

Oct

ober

201

4

SPATIAL SCIENCE Vol. 51, No. 1, June 2006137

one of the flat ports with a 24mm lens generates acoverage of about 26m square. This relatively smallarea coverage is a major limitation of the systemalthough its effects can be reduced by using videorather than film based cameras. The reduced coveragemeans that the Korrong is best used for high resolutionsurveys of small areas (such as shipwreck sites), as aground truthing tool for satellite or airbornesurveys, or in detailed mapping of representativebottom quadrates or transects. It is not suited toproviding complete bottom coverage for large areaimage maps. The benefit of the system is that itgenerates extremely detailed and precisely locatedimagery of the sea floor. The imagery can be sufficientlydetailed for taxonomic identification at the generic, ifnot species level. For any given lens system, areacoverage and image detail have an inverserelationship with each other, the higher the detail thesmaller the area covered. The trade-off between areacoverage and image detail is shown in Figure 3. In asemi-submersible system which will always operateat a specific depth from the surface, this trade off isdetermined by the depth of the water. The deeper thewater the larger the area covered and the lower theresolution. The scale of the resultant imagerychanges continuously with depth across a video transect.In the Korrong a timed echo sounder log, with thetransducer head located at the base of the pod in closeproximity to the cameras, was used to determine thecamera altitude, and hence the scale, for any givenimage.

IMAGING OPERATIONS

There are three distinct ways in which the Korrong hasbeen used to image the sea floor: still imagery at selectedlocations, video transects, and transects with anobserver as recorder. Over some high value, low areatargets such as shallow historic shipwrecks, it may bepossible to obtain complete coverage of the sea floor.However, this has not been attempted and would bedifficult to accomplish.

The original mode of operation from the Korrong wasas a ground truthing tool for satellite imagery. Theapproach was to image a number of randomly selectedtest sites selected in specific sea floor classes usingsmall or medium format film cameras. The classes werederived from an unsupervised classification of LandsatThematic Mapper (TM) imagery. The Korrong images,together with airborne imagery if appropriate, couldthen be used to identify that class (Leach, 1997). Thelarger format cameras were used when higherresolution was required although film format was nevera limiting factor, and even the 35mm cameras producedimagery with sufficient detail to show such features asthe growth zonation within encrusting algae. Thismethod of using the Korrong is similar in design andapproach to conventional diver or sample basedinvestigations. In 2003 a similar approach was usedwhen the Korrong was deployed to produce in situdocumentation of sediment samples taken during asurvey of Port Curtis, Queensland, by the Universityof Central Queensland.

Figure 3. Comparison of area coverage and detail between a image captured at 2m depth (left, 1.2m across)and an image captured at 1m depth (right, 0.6m across). Both images were captured offshore of

Mt. Martha on the eastern side of Port Phillip Bay in Southern Australia.

Dow

nloa

ded

by [

Col

orad

o C

olle

ge]

at 1

7:17

28

Oct

ober

201

4

Vol. 51, No. 1, June 2006 SPATIAL SCIENCE138

While this approach had some success in verifyingsatellite imagery based classifications, it becameevident that important information was being discarded.The images of areas selected within any given TM classshowed both variation between images and spatialcomplexity within images. As well, visual observationsof the sea floor made from the pod whilst transitingfrom one image site to another, indicated there was adegree of spatial variation, including completelydifferent bottom types, which was not representedeither in the Landsat imagery or in the scatteredKorrong image frames. This use of the Korrong, beingsimilar to a diver mapping randomly selected quadrats,suffers from the same drawbacks: it is prone tostatistical aberration and can not represent the spatialrelationships between bottom types. Also, it was clearthat far more information about the sea floor could begained from the Korrong than from satelliteobservation. This caused a shift in emphasis in theKorrong operations. Instead of the Korrong being aground truthing tool supporting satellite observations,the satellite data became a way to provide broad areaobservations to support the Korrong data.

The most effective mode of operation for the Korrongis the collection of video transects (Figure 4) withcomments by the observer in the pod recorded on thesound track. While still only a sampling technique,these transects provide as complete a coverage of thesea floor as is possible given the imaging platformcharacteristics. Transect locations are normallyconducted at right angles to the shore. The length of thetransect are normally from the position where the wateris too deep for imaging the sea floor (optically deepwater - dependent on water condition but normallyless than 8m) to water too shallow to operate (about1m). However, in some areas of shallow, sandy seafloor this results in transects of undue length andshorter transect lengths are used. The transect sitesare selected at a regular spacing with special transectsadded to cover sites of particular interest, such aschanges in geology or aspect, which may be missed bya regular spacing. DGPS data are used to record theactual orientation while making way and position ofthe transects since tidal currents and surface conditionsoften cause variations from the selected course.During the 2003 Port Curtis surveys, video transects

Figure 4. Image sequence over an algae coveredrocky reef. Images are about 0.85m across and werecaptured at approximately 1.4m depth offshore of MtMartha on the eastern shore of Port Phillip Bay in

Southern Australia.

A

B

C

Dow

nloa

ded

by [

Col

orad

o C

olle

ge]

at 1

7:17

28

Oct

ober

201

4

SPATIAL SCIENCE Vol. 51, No. 1, June 2006139

from the Korrong were used to calibrate acousticsounder data taken from the Korrong at the same time.This allowed a very rigorous investigation of the typesof habitat changes that the acoustic system coulddistinguish.

The images in Figure 4 were acquired from a singlevideo transect. They do not, however, form a straightline. Each succeeding image is offset to the right of theprevious frame. The offset of the imagery can be causedby drift or by the boat crabbing into the current tomaintain course. The DGPS log can distinguishbetween the two causes. In the case of Figure 4, theoffset was due to drift caused by a strong tidal currentaround a rocky headland.

The transect imagery can be viewed and aclassification of meaningful and mappable bottom typesdetermined. These map units will be a combination ofsubstrate, topography and macro benthic biology. Theyare termed sub-tidal terrain units because of thesimilarity with terrain units or land system mappingon land (eg. Purdie, 1984; Wirth et al., 1996). Thesecan then be related to airborne or satellite imagery toprovide a broad area coverage linking the transect lines.Representative frames, or sequences of frames, can beacquired for detailed analysis and as a basis forcomparison across transect lines. It would be possibleto repeat these transect lines and produce repeatimagery of the same areas of the sea floor. In theory,this allows for a direct comparison and an estimation ofchange over time. It would, however, involve betterline steerage to maintain the same transect, but at thisstage has not been attempted.

The third way in which the Korrong has been used isdirect mapping by the human observer in the pod. Thistechnique relies on having, or being interested in, onlya small number of known bottom types that can berepresented as hot keys on a laptop computer. In thisprocedure the observer enters a change in bottom typedirectly into a database with the DGPS positionalinformation. These data can then be loaded directly intoa GIS to map the occurrence of the specified bottom typealong the transect path. The technique has mainly beenused by the Victorian Marine and Freshwater ResearchInstitute (MAFRI), an arm of the Victorian state

government charged with operational research into themarine environment. This procedure was developed tomap the area of infestation by an exotic species ofJapanese Kelp, where the bottom types considered wereeither the presence or absence of the kelp. MAFRIfurther developed the technique in mapping seagrasscommunities in Corner Inlet (Roob et al., 1998) andthe Gippsland Lakes in South Eastern Australia (Rooband Ball, 1997). Here the bottom types consideredwere the species of seagrass (2 species) and four levelsof cover ranging from absent to complete cover.Obviously this technique can not be used in an unknownenvironment or where the sea floor terrain is complex,and it helps to have video and other imagery as a checkon the observer’s categories. However, the approachhas been used to map all of the significant bays andinlets along the Victorian coast, including Port Phillipand Westernport Bay, and has been successfully appliedto the mapping of rocky reef areas. Airborne imagerywas used to provide the coverage between the transects.

IMAGE MANIPULATION

Once the imagery has been acquired as digital frames,it can be analysed and manipulated like any other set ofdigital imagery. Two techniques in particular have beenfound to be useful in the analysis of Korrong semi-submersible imagery. These are the construction ofimage mosaics and stereo pairs (Figures 5 and 6).Mosaics can be used to increase the area coveragebeyond what is possible with a single frame. Thisovercomes, to some extent, the major limitation of boatmounted imagery: the lack of area coverage. It alsoallows image measurements, such as percentage coverand the spacing between organisms, to be made over alarger sample size. It does have the limitation that themosaic can only be constructed along the transect. Thiscan rapidly lead to a mosaic image which has a veryunwieldy aspect ratio and, in practise, only shortsections of high interest are used to construct mosaicimages.

The stereo pairs are constructed from a single videosequence using frames from different points along theboat track: along track stereo. Along track stereo ispossible because of the over sampling that occurs withvideo imagery. Each area of the sea floor will be imaged

Dow

nloa

ded

by [

Col

orad

o C

olle

ge]

at 1

7:17

28

Oct

ober

201

4

Vol. 51, No. 1, June 2006 SPATIAL SCIENCE140

Figure 5. Mosaic constructed from the sequential video frames at Figure 4

Figure 6. Stereo pair constructed from frames B and C at Figure 4

Dow

nloa

ded

by [

Col

orad

o C

olle

ge]

at 1

7:17

28

Oct

ober

201

4

SPATIAL SCIENCE Vol. 51, No. 1, June 2006141

several times from different positions as the boat movesalong its track. Selected frames can then be isolated anda stereo pair constructed. Since the Korrong camerasand ports have not been calibrated, the resultant stereocan only be used as an aid in qualitative interpretation.They do, however, give the investigator a very clearindication of the three dimensional structure of thebenthic community. It would be possible to mount twocameras in the pod and take conventional stereo imagery.Since the stereo pair would be taken at the same time,this would eliminate the sea floor motion problemdiscussed below. The increased operational complexity,however, will only make this worthwhile if stereoimagery is of special importance to the particularinvestigation.

Both of these techniques suffer from the sameproblems in the shallow marine environment: they aretime consuming and there is often considerable relativemovement of the biological substrate between one frameand the next. This is particularly true in shallow areaswith a high vegetation cover. During mosaicconstruction, wave or current induced motion in thevegetation makes it very difficult to match frames in avideo sequence and the same motion makes it hard toconstruct a stereo model from along track information.There are, however, many frames to choose from in avideo sequence and it is a simple, if time consuming,matter to select frames with similar movement andthat can be matched.

CONCLUSION

The Korrong project has demonstrated the usefulnessof semi-submersible systems in monitoring the marineenvironment. It has been a major research toolinvestigating the relationship between the sea floorterrain and the physical environment, and it has alsobeen used as an operational tool, mapping the spread ofmarine pests and sea grass distribution and cover. Ithas proven very cost effective both in initial capitalcosts and in operation. The construction costs were onlya little higher than those for a standard work boat, andthe operational costs are about the same. The exactoperational capacity of the vessel depends onenvironmental conditions, such as light intensity andwater quality, and the mode of operation, but tens of

kilometres of video transect are easily accomplished ina day.

However, its operation has demonstrated someimportant limitations. The first is that, given theturbidity of coastal water, the system can routinelyoperate to only about 6m depth. This limitation isinherent in the design concept and should really beconsidered as a characteristic of the system. This is notas serious a limitation as it sounds as much of the workinvolved in monitoring environmental impact isconcentrated in this depth range where most humanimpact on the seafloor environment occurs.

The second limitation is that, as the observation podis hard mounted into a 5.2 metre aluminium work boat,it is difficult and expensive to transport long distances.Even though the whole system is road trailerable, thisnormally confines the vessel to inshore waters. Whenit is transported long distance, it is sent via rail but itcan also be shipped by road freight. This operationalrestriction is a property of this particular vessel designbut, not of the semi-submersible concept. Work iscurrently underway to design a small towed body whichwill trade off some positional precision for ease oftransport, and to design a cheap modification to glassbottomed tourist boats so that they can be adapted to therole of monitoring the coral reefs over which theyoperate.

The capability to image and map our sea floorenvironment is vital if we are to monitor the conditionof our marine resources. This is especially true in theshallow depth range in which semi-submersiblesystems operate, since it is within this range that mostbiological productivity occurs, where many fish speciesfeed and spawn and where human impact may be at itsgreatest. This monitoring needs to be done in a cheapbut effective way that is not only open to developingcountries, where environmental pressures are oftenhigh, but can also be used locally as a routine monitoringtool in developed countries.

ACKNOWLEDGMENTS

Acknowledgment is given to BHP Billiton Pty. Ltd. whofunded this project.

Dow

nloa

ded

by [

Col

orad

o C

olle

ge]

at 1

7:17

28

Oct

ober

201

4

Vol. 51, No. 1, June 2006 SPATIAL SCIENCE142

REFERENCES

Gross, M.G. (1977) Oceanography: A view of theEarth, 2nd Ed. Prentice-Hall, New Jersey.

Honey, F.R. and Byrne, P. (1978) Air Survey andSatellite imagery: tools for shallow waterbathymetry, Proceedings of the 20th SurveyCongress, Darwin.

Jupp, D.L.B. (1988) Background and extensions todepth of penetration (DOP) mapping in shallowcoastal waters, Proceedings of an InternationalSymposium on Remote Sensing of the CoastalZone IV. 2.1-IV.2.19, Gold Coast, Queensland,7-9 September.

Lauscher, F. (1944) Sonnen- und Himmelstrahlungim Meer und in Gerwassern, Handb. d. Geophysik ,Vol. 7, Ch. 12, pp. 724-763, Berlin.

Leach, J.H.J. (1997) The Use of Multi-level Imageryin Marine Environmental Mapping, RecentAdvances in Marine Technology 96, Ed. N. K.Saxena, PACON, Honolulu, Hawaii, pp. 1-9.

Neumann, G., and Pierson, W.J. (1966) Principles ofPhysical Oceanography, Prentice-Hall, NewJersey.

Purdie, R. (1984) Land Systems of the SimpsonDesert Region, CSIRO, Division of Water and Land

Resources, Natural Resources Series No.2,Australia.

Roob, R. and Ball, D. (1997) Victorian Marine HabitatDatabase, Seagrass - Gippsland Lakes, Unpublished,Department of Natural Resources and Environment,Victoria.

Roob, R. Werner, G. and Morris, P. (1998) VictorianMarine Habitat Database, Seagrass - Corner Inletand Nooramunga, Department of Natural Resourcesand Environment, Victoria, Proceedings of theMarine & Coastal Data Management Conference,CSIRO, Hobart.

Skewes, D., and Long, B.G. (1994) Sampling MarineResources, GIS User, No.8, pp. 31-32.

Staniforth, M. and Vickery, L. (1984) The TestExcavation of the William Salthouse Wreck Site,Australian Institute of Maritime ArchaeologySpecial Publication No. 3, WA Maritime Museum,Fremantle, WA.

Tewinkel. G.C. (1963) Water depths from aerialphotographs, Photogrammetric Engineering, Vol.26, No. 6.

Wirth, T., Maus, P., Lachowski, H. and Dallon, D.(1996) Mapping Ecosystems, Earth ObservationMagazine, No. 5, pp. 14-18.

Dow

nloa

ded

by [

Col

orad

o C

olle

ge]

at 1

7:17

28

Oct

ober

201

4