VISUALISING COASTAL SEABED CHARACTERISTICS: USING VRML MODELS TO PRESENT THREE DIMENSIONAL SPATIAL DATA VIA THE WEB JAMES COFFEY1, 2, DAVID A. RYAN1 & DAVID J. BEARD1 1
Geoscience Australia, GPO Box 378, Canberra, ACT 2601, Australia Defence Imagery and Geospatial Organisation, R4 Russell Offices, Canberra, ACT 2600, Australia
2
Tel. +61 2 6127 7230 Email.
[email protected] ABSTRACT: Geoscience Australia (GA) has recently produced a number of free Webviewable 3D models for coastal data. These models are good tools for sharing information with Coastal CRC partners and stakeholders. The models integrate a diverse range of spatial data within an easy to use interface. Datasets include multi-beam bathymetry and backscatter, sediment characteristics, benthic habitats and satellite imagery. The models use the open source and ISO standard Virtual Reality Modelling Language (VRML) file format. The models are best demonstrated by the Keppel Bay and Fitzroy River area 3D Model. These coastal 3D VRML models contain developments that have progressed GA’s 3D VRML models. VIRTUAL REALITY MODELLING LANGUAGE, VRML, COASTAL ZONE MANAGEMENT, 3D BATHYMETRY, GEOMORPHOLOGY
INTRODUCTION The development of open source three dimensional (3D) models has been driven by an increased need to share complex spatial information with broad audiences. The primary reasons for the success of generic 3D modelling include its capacity to communicate large volumes of complex data to users of varying technical skill levels. More recently, utilisation of the Internet for the delivery of 3D models has added value to existing techniques by removing many of the technological resource requirements needed to access the information (Beard et al., 2005, Watford et al., 2005 and Beard, 2006). Geoscience Australia (GA) has used the Virtual Reality Modelling Language (VRML) to produce interactive Web-deliverable 3D models of complex geological structures since 2001 (Beard et al., 2005). As part of our involvement with the Coastal CRC, we have recently applied this technology to coastal datasets. Models have been created for Woody Island and Cockburn Sound (in Western Australia), Sydney Harbour (New South Wales) and the Keppel Bay and Fitzroy River area (Queensland). The models are available from http://www.ozestuaries.org/projects/3D_index.jsp. The adaptation of our 3D modelling capability to include coastal data has driven considerable change in the way we process and deliver our models. In particular, the increasing use of multibeam (or ‘swath’) echosounders for the detailed bathymetric mapping of the seabed has placed unique demands on techniques for the delivery and representation of very large datasets (Hughes Clarke et al., 1996). These developments are best demonstrated by the Keppel Bay and Fitzroy River area 3D model. McCann (2002 and 2004) and Watford et al. (2005) have also used VRML to view marine zone data.
BACKGROUND The Coastal CRC Fitzroy Project The Fitzroy River Contaminants study is a collaborative project under the Cooperative Research Centre for Coastal Zone, Estuary and Waterway Management (the Coastal CRC, www.coastal.crc.org.au). The project aims to model the transport of catchment derived nutrients, pesticides, and sediments through the Fitzroy estuary, and determine their ultimate fate once they reach the marine environment. Of particular concern for environmental managers is the yield of terrestrial sediment from coastal catchments in Queensland that reaches the Great Barrier Reef, which has implications for coral health. The Fitzroy River has the second largest catchment in Australia (following the Murray-Darling Basin), encompassing 144,000 km2 in dry-tropical central Queensland. Land use in the extensively cleared catchment is dominated by agriculture and coal mining. The Fitzroy River drains into the southern Great Barrier Reef through Keppel Bay, a turbid, shallow, macro-tidal embayment. The gently sloping continental shelf in the region is approximately 100 km wide, and the shelf edge is lined by numerous platform reefs of the Capricorn-Bunker Group. Geoscience Australia’s role in the Coastal CRC Fitzroy Project was to determine processes of sediment transport, erosion and deposition in the Fitzroy River-Keppel Bay system, and develop a quantitative model of the volume of sediments trapped in, and escaping from the estuarine system. This was achieved using detailed field data, as well as assessment of seabed geomorphology using various benthic mapping techniques. These data types lend themselves to display in a 3D environment, however also represent unique challenges due to the diversity of data types and formats, and the large size of some files acquired. Advantages of displaying spatial data in 3D Displaying spatial data in 3D, rather than 2D or plan view, provides a more natural view of the data. This real-world resemblance allows audiences to quickly and easily gain a clear understanding of the physical environment represented by the data, and the relationships between datasets. This is especially the case for audiences not familiar with interpreting spatial data. For further detailed discussion in relation to 3D marine data see Watford et al. (2005). What is VRML? VRML is an open source ISO standard file format for 3D graphics, allowing 3D content to be freely distributed over the Internet. To view VRML files users require a VRML viewer plug-in for their web browser, easily downloaded from the Internet. VRML was developed by the VRML Consortium and formalised into an international standard (ISO/IEC 14772) in 1997 by the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC). The VRML specification is available from the following Web site: http://www.web3d.org/x3d/specifications/vrml (ISO/IEC 1997). For more information on the history of VRML see Carey and Bell (1997). Much information on VRML, including tutorials, is available from the Web3D Consortium Website: http://www.web3d.org/x3d/vrml/index.html (Web3D Consortium, 2006a). Also, a
more detailed explanation of VRML can be found in Beard et al. (2005) and Beard (2006). Geoscience Australia’s use of VRML GA has used VRML to share 3D geoscience data via the Web since 2001. The main achievement of GA has been the development of VRML-HTML-JavaScript interfaces that provide users with a high degree of interaction with the 3D data. Datasets can be added or removed from the display, moved within 3D space, interrogated for attribute information and scaled vertically. Differing data types and large data volumes can be combined via the interfaces (Beard et al., 2005, and Beard, 2006). Figure 1 shows this interface for the Keppel Bay and Fitzroy River area 3D model. GA’s 3D VRML models use the Blaxxun Contact and BS Contact VRML plug-ins. See Blaxxun Technologies (2006) and Bitmanagement Software GmbH (2006) for more information. The models are constrained to these plug-ins, rather than other VRML plug-ins, because they user HTML and JavaScript to interact with the VRML content and each VRML plug-in performs this interaction in a different way. More information on GA’s 3D VRML models is provided in Hay (2003), Beard et al. (2005), Watford et al. (2005) and Beard (2006). DATA COLLECTION, PROCESSING AND VISUALISATION Bathymetry Two resolutions of bathymetric data are contained in the model, including the regional 250 m bathymetric grid (Webster et al., 2005), and high resolution multibeam sonar acquired by the Coastal CRC. Regional bathymetry GA, in collaboration with the Australian Hydrographic Service (AHS), compiled archived single-beam bathymetric surveys to develop a base bathymetry surface (Webster et al., 2005). Archived paper records from the AHS were digitised, compiled, and all data were interpolated onto a regular grid using ER Mapper 7 software, to produce a broad-scale bathymetric representation of the Fitzroy River mouth, Keppel Bay, and the continental shelf of the southern Great Barrier Reef. For the area of interest, the average density of soundings ranged from 10-100 soundings/km2. High resolution multibeam sonar bathymetry A relatively high resolution (centimetre vertical, decimetre horizontal) ResonTM Seabat 8125 multibeam sonar system was used to delineate physical features of the substrate for targeted areas. The multibeam system emits acoustic energy at a frequency of 455 kHz, and resolves signals returning from the seabed into 240 focussed beams in an arc 120° across track, and 1° along track, providing thousands of soundings per square kilometre. Bathymetry data were acquired for areas of interest that were identified in sub-bottom profiles, such as Centre Banks. Multibeam sonar data were processed using Caris HIPS/SIPS 6 software, and soundings were originally decimated into 0.25 m grid cells. For efficient delivery and visualisation of the above bathymetric data over the Internet the source point data were converted to triangulated surfaces with irregular triangle
sizes (small triangles in areas of complex bathymetry and large triangles in areas of simple bathymetry). This processing was performed in GOCAD software. For the regional and acoustic sub-bottom profiling bathymetry, low, medium and high detail surfaces were created (Figure 2). The model interface allows users to switch between levels of detail. This allows users with low bandwidth Internet connections to view lower detail data and users with high speed connections to view higher detail data. This capability is enhanced by the inclusion of the high resolution multibeam sonar data for small regions. The model interface allows users to turn these high resolution areas on and off (Figure 3). The borders of the high resolution surfaces mesh well with the larger bathymetry surfaces. However, the integration of bathymetric datasets of varying quality and resolution introduced data artefacts into the bathymetry surfaces. These flaws are manifested by sudden changes in height and detail that can be seen between datasets. Removing these errors poses significant problems, as they are a result of direct discrepancies between the two datasets rather than processing or representation errors. Acoustic sub-bottom profiles To obtain an overview of the sub-surface stratigraphy of the bay, a sub-bottom profile survey was undertaken. A Datasonics CAP-6600 Chirp II acoustic profiling system was used, and set to sweep through a frequency range of 1-10 kHz, providing a range resolution of approximately 7.5 cm. The location of this data is represented as a polyline which is draped on the bathymetry surface. In other VRML models (e.g. Cockburn Sound), techniques have been developed to ‘hang’ the 2D sub-bottom profile section below the bathymetry surface, providing a realistic impression of stratigraphic layers. Sediment sampling Twenty vibracores were collected in Keppel Bay during September 2004, using a shallow-water vibracoring system. Vibracore sites were selected to ground-truth the uppermost reflectors in the sub-bottom profiles and provide a wide spatial coverage of Keppel Bay. A number of cores were acquired from the Fitzroy estuary and billabongs, using a hand operated push corer developed by Geoscience Australia, which is capable of retrieving cores up to two metres long in soft sediments. A series of samples were also collected from the floodplain using a geoprobe percussion corer, operated by the Queensland Department of Natural Resources and Mines. A number of hand auger sediment samples were collected from beach ridge units, for the purpose of optical stimulated luminescence (OSL) dating. These core samples are presented in the VRML as point data, in their correct 3D locations, allowing users to determine where samples were taken, and request full datasets from Geoscience Australia if required. Sediment samples for the inner shelf were acquired using a 0.5 litre grab sampler. For each sample, detailed grain size analyses, CaCO3 proportions, and mineralogy were determined. To visualise some of this sediment data in the VRML model, 3D pie charts were created using GAWK scripts to convert tabular data to VRML. (GAWK is a pattern scanning and processing language good for reformatting digital text files (Open Source Technology Group, 2006).) These pie charts represent the sample content visually and are placed in their correct 3D location. They are also ‘hot-linked’
to attribute information, including percent gravel, sand and mud; percent CaCO3; and percent feldspar (accessed by clicking on a pie chart) (Figure 4). Other GIS layers A range of other GIS data sources are included in the Keppel Bay and Fitzroy River model, to provide context for the primary data. Great Barrier Reef Marine Park Zones This data represents the various management zones of the Great Barrier Reef Marine Park and is included as transparent triangulated surfaces, visible above the bathymetry. The polygons are again hotlinked to attribute information (Figure 5). The triangulated surfaces were created in GOCAD from shape file polygons. Land surface terrain model Shuttle Radar Topographic Mapping (SRTM) data was used to create a triangulated surface for the land component of the study area. As with the bathymetry, the gridded data was processed in ER Mapper and then converted to a surface of irregularly sized triangles in GOCAD. Satellite imagery Several mosaiced Landsat TM 7 satellite images are included, to provide users with a real-world view of the study area. For the land area these images are draped over the terrain model, whereas over-water imagery is shown as a flat surface at 0 m elevation. Cultural data Road, rail and airfield data are displayed as vector objects. Also, labels for many places of interest are provided. Even more data types are included in the other 3D models available on the OzEstuaries Website, such as acoustic backscatter and benthic habitat data (Woody Island) and sub-bottom profile section images that ‘hang’ below the bathymetry surface (Cockburn Sound). File sizes Some of the datasets in the model are very large. To Web-deliver these datasets a number of techniques were employed. Gridded data were converted to surfaces of irregularly sized triangles, greatly reducing file sizes, since small triangles are used in complex areas and large triangles in less complex areas. Also, the ASCII (text file) VRML files were compressed to binary format gzip files, using Chisel, a free VRML optimisation tool (Institutt for energiteknikk 2006). Compression is possible because VRML viewers are able to uncompress gzipped files on-the-fly. File sizes for the source datasets are compared to the VRML file sizes in Table 1. Of particular note is the large size reduction for the bathymetry data. For some of the other data types the VRML files are larger than the source data, but the file sizes are still small. DISCUSSION Evaluation In the Coastal CRC projects, one of GA’s primary roles was to acquire and house large volumes of geoscientific data relevant to natural resource management in the project areas. Although all material was published in reports and scientific journals, it
was recognised that online access to data is needed in order to provide stakeholders with easy access to digital datasets for further analysis and consideration. The OzEstuaries website (www.ozestuaries.org) was developed in part to assist the dissemination of relatively straightforward and simple coastal datasets, however the large data volumes generated during bathymetric surveys requires better visualisation tools. The VRML technique allows users to gain an appreciation of the extent of the data and the seabed features therein. Overlaying other information such as sediment samples and satellite imagery further aids interpretation and understanding of processes. In this way, the project has demonstrated how large 3D datasets may be delivered over the Web.
Table 1. VRML file sizes compared to source data file sizes
Dataset High resolution multibeam bathymetry Landsat TM images Shuttle Radar Topographic Mapping Regional bathymetry Cultural data Great Barrier Reef Marine Park Zones Grab samples and core locations
Data Type Point/ Triangulated surface Raster Raster/ Triangulated surface Point/ Triangulated surface Line Polygon/ Triangulated surface Point
Source Size (Mb) ~10,000
VRML Size (Mb) 0.75-3.9
400
0.56~9.5
30
0.75~1.6
25
0.15~0.4
0.6
1.2
0.5
1.1
0.3
0.75
Comment High resolution surfaces
Image resampled & compressed Surface provided in low, medium and high resolutions Surface provided in low, medium and high resolutions
VRML includes description of visual properties
New VRML model developments in the Keppel Bay and Fitzroy River model The Keppel Bay and Fitzroy River area 3D VRML contains important developments for GA’s VRML models. It set a new benchmark in the way that it allows higher resolution data to be disseminated, giving users the ability to view low, medium and high resolution imagery, and by including very high detailed data for small regions within the study area. This provides high resolution data to users with high speed connections, without compromising the ability of users with low speed connections to view the data. The 3D pie charts were a new development that allows users to quickly visualise the distribution of materials within sediment samples, as well as obtain the exact attribute figures for the sample contents. Shortcomings of VRML GA’s 3D VRML models are an excellent tool for sharing information with a wide range of web users. However, at present the VRML system only operates with web browsers that have the Blaxxun Contact 5 or BS Contact 6 plugins installed. VRMLS are also limited by Internet connection speeds, and may not meet the interactivity needs of all users as further data can not easily be added by web users. These shortcomings are discussed in more detail by Beard et al. (2005) and Beard (2006). Further developments for the Keppel Bay and Fitzroy River model could include an automated flythrough, allowing easier navigation by novice users. Furthermore, future integration of vertical, 2D sub-bottom profile images and images for core samples will further improve data availability and visualisation.
Future directions GA is in the process of migrating from the VRML file format to X3D, the eXtensible Markup Language (XML) successor to VRML. The benefits of moving to X3D include: improved connection to database information; improved interoperability with other XML data formats; and easier access to the data since X3D models should be viewable on all X3D viewers, unlike VRML. For more information on X3D see the Web3D Consortium Web site: http://www.web3d.org (Web3D Consortium, 2006b). CONCLUSION In this paper we have discussed the benefits of using VRML to share and visualise coastal datasets via the Web. The Keppel Bay and Fitzroy River area 3D model was a successful example of the amalgamation of disparate datasets into a single communication tool for the Coastal CRC. The openness and versatility of VRML makes it a good tool for sharing 3D scientific data via the Web. Through the continuing work at Geoscience Australia on 3D modelling functionality, we can be sure that this capability will continue to evolve to meet the future data delivery demands. ACKNOWLEDGEMENTS 3D modelling with VRML at Geoscience Australia was pioneered by Russel Hay and Malcolm Nicoll. The models currently available on the OzEstuaries Website were created by James Coffey, David Ryan, Dale McNally, Helen Bostock, Benjamin Hardy and David Beard. The authors greatly appreciate John Creasey’s editorial support for this paper. REFERENCES Beard, D. J., Hay, R.J., Nicoll, M. G. and Edge, D. O. (2005), “3D Web Mapping – 3D Geoscience Information Online”, Proceedings of SSC 2005 Spatial Intelligence, Innovation and Praxis: The national biennial Conference of the Spatial Sciences Institute, September, 2005, Melbourne, Spatial Sciences Institute. Beard, D. J. (2006), “Using VRML to Share Large Volumes of Complex 3D Geoscientific Information via the Web”, Web3D 2006 11th International Conference on 3D Web Technology, Columbia Maryland, 18-21 April 2006, ACM. pp. 163-167. Bitmanagement Software GmbH (2006), Bitmanagement Software GmbH Web site, http://www.bitmanagement.com, Web site accessed 20 June 2006. Blaxxun Technologies (2006), Welcome to blaxxun.com, http://www.blaxxun.com, Web site accessed 20 June 2006. Carey, R. and Bell, G. (1997), The Annotated VRML 97 Reference Manual, http://www.cs.vu.nl/~eliens/documents/vrml/reference/BOOK.HTM, Web site accessed 20 June 2006. Also available in hardcopy from A-W Developers Press. Hay, R. J. (2003), “Visualisation and Presentation of Three Dimensional Geoscience Information”, Proceedings of 21st International Cartographic Conference 2003, Durban, International Cartographic Association.
Hughes Clarke, J. E., Mayer, L. E., and Wells, D. E., (1996), “Shallow-water imaging multibeam sonars: a new tool for investigating seafloor processes in the coastal zone and on the continental shelf”, Marine Geophysical Research 18, 607-629. Institutt for Energiteknikk (2006), Chisel 2.1.1 (HVRC Edition) VRML Optimisation Tool, http://www2.hrp.no/vr/tools/chisel/install.htm, Web site accessed 20 June 2006. ISO/IEC (1997), Information Technology - Computer Graphics and Image Processing - The Virtual Reality Modelling Language (VRML) - Part 1: Functional Specification and UTF-8 Encoding, ISO/IEC 14772-1:1997, http://www.web3d.org/x3d/specifications/vrml, Web site accessed 20 June 2006. McCann, M. P. (2002), “Creating 3D Oceanographic Data Visualisations for the Web”, Proceedings of the seventh international conference on 3D web technology, ACM pp 179-184. McCann, M. P. (2004), “Using GeoVRML for 3D oceanographic data visualizations”, Proceedings of the ninth international conference on 3D Web technology, ACM. pp 15-21 Open Source Technology Working Group (2006), Gawk for Windows, http://gnuwin32.sourceforge.net/packages/gawk.htm, Web site accessed 20 June 2006 Watford F. A., Beard, D. J., Hay, R. J., and Ambrose, K. (2005), “Innovative Visualisation for Australia’s Marine Zone”, Proceedings of SSC 2005 Spatial Intelligence, Innovation and Praxis: The national biennial Conference of the Spatial Sciences Institute, September, 2005, Melbourne, Spatial Sciences Institute Webster, M. A., Petkovic, P. (2005), Australian bathymetry and topography grid, June 2005, Geoscience Australia Record, 2005/12, 30 pp Web3D Consortium (2006a), VRML Archives, http://www.web3d.org/x3d/vrml/index.html, Web site accessed 20 June 2006 Web3D Consortium (2006b), Web3D Consortium, http://www.web3d.org, Web site accessed 20 June 2006
Figure 1. Keppel Bay and Fitzroy River 3D VRML model interface, showing bathymetry and terrain, with a 20x vertical exaggeration.
Figure 2. Low, medium and high detail triangulated surfaces for the terrain data. This figure also shows the irregular triangle sizes in the surfaces.
Figure 3. High resolution data for the Jabiru Shoals off (left) and on (right).
Figure 4. Pie charts of percent Calcium Carbonate for sediment samples, over bathymetry. Also shown is the attribute information for one sample.
Figure 5. Marine zone data, with Landsat imagery draped over terrain. The green regions are marine national park.
