A Chapter to a  book

 entitled "Collaborative Geographic Information Systems"

edited by Shivanand Balram and Suzana Dragicevic, Simon Fraser University, Canada

---------------------------------------------------------------------------------------------

A Collaborative Virtual Geographic Environment: Design and Development

Jianhua GONG¹, Hui LIN²

¹State Key Laboratory of Remote Sensing Science

Institute of Remote Sensing Applications

Chinese Academy of Sciences

Datun Road 3,Chaoyang District

Beijing 100101,P.R. China

Phone : +86-10-64849299 (O)

Email: jhgong@irsa.ac.cn

http://www.vgelab.org/

²Joint Laboratory for Geoinformation Science

The Chinese University of  Hong Kong

Shatin, Hong Kong

Tel: (852)-2609-6528  Fax: (852)-2603-5006

Email: huilin @ cuhk.edu.hk 

http://www.jlgis.edu.cuhk.hk

Abstract: A collaborative virtual geographic environment (CVGE) is a 3D, distributed, and graphical world representing and simulating geographic phenomena and processes to enable geographically distributed users to explore geo-problems and theories and generate hypotheses, and to support geo-model building and validation and collaborative ecological planning. This chapter reports an approach to establishing a CVGE across the Internet and its application to the collaborative planning of silt dam systems in watersheds through the integration of distributed virtual environments, Geographical Information Systems (GIS), planning models of dam systems and geo-collaboration. The chapter addresses the conceptual and system frameworks of the distributed CVGE, and the 3-D modeling of virtual geographic environments and virtual collaborative studios, in addition to the mediated tools for collaboration, such as streaming media based communication, shared whiteboards for text input and graphics drawing, and text-based dialogue. In a case study of the Qiu-Yuan-Gou watershed, Suide County, Shanxi Province, China, a prototype system is designed and developed with Java, Java3D, and VRML. The complete dam systems in the Qiu-Yuan-Gou watershed represent a typical example model of a massive silt dam construction project on the Loess Plateau. The study employs the example model of the watershed to explore the methodologies of collaborative spatial planning of silt dam systems. Using the prototype system, participants can implement communication with each other via media tools, mainly in the virtual collaborative studio, and 3-D editing of shared dams, calculation of topographic properties and ideal spatial distribution of dam systems in virtual geographic environment. 

Keywords: Virtual Geographic Environments, Geo-Collaboration, Spatial Planning, the Qiu-Yuan-Gou Watershed, Soil-Water Conservation, Silt Dam Systems, the Loess Plateau

Introduction

Geographic environments are open, huge complex systems in which most complicated geo-problems, such as ecologic planning, sustainable urban development, evaluation of large geographic projects, disaster forecasting and early warning, emergence response and process, and ecologic security need to be collaboratively explored and solved by a group of people. Meanwhile, the rapid development of information and communication technologies facilitates the potential to invent many tools to support collaboration, with computer supported cooperative work (CSCW) becoming an important research field (Mandviwalla and Khan,1999; MacEachren, 2001). In the GIScience community, the limitations of current geographic information systems only designed for individuals and the resultant increase in interest in geo-collaboration is evidenced by the growing body of work on group decision support systems, public participation GIS, collaborative GIS, and collaborative geo-visualization  (Densham et al., 1995; Batty et al., 1998; Jankowski, and Nyerges, 2001; Craig et al., 2002; Cheng et al., 2003; Benko et al., 2003;MacEachren and Brewer, 2004). This chapter will focus on the design and implementation of technologies for geo-collaboration from the perspective of distributed virtual geographic environments.      

The rest of the chapter is organized as follows. In section 2, work related to geo-collaboration, with a special emphasis on distributed virtual environments, is presented through a discussion of relationships with the online community, networked visualization and Internet/virtual GIS. Section 3 elaborates the design of the conceptual and system framework of collaborative virtual geographic environments (CVGE). Section 4 presents a prototype system of CVGE and a case study of the dam systems planning in the Jiu-Yuan-Gou watershed. The last two sections conclude with a discussion of future research trends.

Background and related works

Geo-collaboration can be defined as a group of people working together in both the same or differing geographical location and time, to accomplish geo-tasks or to solve complex geo-problems. The study of geo-collaboration involves diverse aspects ranging from participants and organization to mediated tools, geo-problem contexts and supportive environments. From the viewpoint of geo-collaboration supportive mediation technologies, this chapter highlights distributed virtual environments supporting different place/same time geo-collaboration. Distributed virtual environment technology works to establish distributed, 3-D environments allowing geographic distributed users to meet and interact virtually with objects and processes and collaborate with other users in 3-D worlds on the Web (Normand, 1999; Gong and Lin, 2000). In recent years, distributed virtual environments have drawn increasing interest in academic and industrial communities (Hibbard, 1998; Dykes et al., 1999; Singhal and Zyda, 1999; Oliveira  et al., 2000; Blaxxun, 2005).

From the perspective of online communities, originating from the online text-based or voice-based chat rooms such as the famous Muds (Multi-User Dungeons) and Tencent OICQ in China, distributed virtual environments are now used to create online, 3D communities for conducting a variety of activities such as chatting, game playing, forming clubs, virtual house building, and shopping in the virtual mall (Cooper, 2000; OICQ, 2005). Fig. 1 illustrates a snapshot of Cybertown, a well-known 3-D Internet community (Cybertown, 2005). In Fig. 1, online users are represented as 3-D avatars who can navigate in 3D worlds and chat with other users. From the perspective of networked computer graphics integrated with CAD and visualization technology, distributed virtual environments are applied to the building and distribution of CAD or visualization environments for collaborative designing or visual data interpretation. Fig. 2 shows a distributed CAD design environment, called DMUConference (Tecoplan, 2001). In the DMUConference environment, geographically distributed designers can meet virtually and discuss the design of cars.

Fig. 1. Cybertown, an online 3D community

Fig. 2. DMUConfernce, a distributed CAD design environment

Applications and importance of virtual reality/virtual environment technology have increasingly drawn attention in the GIScience, modern Cartography, and Geographic Sciences communities (Faust, 1995; Batty et al. 1998; Gore 1998; Verbree et al., 1999; Gong and Lin 2000; Van Maren and Germs, 2000). Much effort has been directed towards the integration of the traditional GIS, geographical application models, AutoCAD, and distributed tools such as ActiveX, VRML, Java, and Java3D to construct distributed virtual geo-environments (GeoVE) on the Internet. Using Java and Java3D, Hibbard et al. (2005) designed and developed a VisAD system to create applications that enable many users to implement the visualization of a shared set of numerical geo-data and geo-computations. Based on Geometrek’s DeepMatrix 1.1, an existing multi-user virtual environment, Manoharan et al. (2002) established a collaborative urban planning prototype system to assist shared analysis of urban planning proposals by visualization and interaction with spatial data. MacEachren et al. (1999, 2005) address geospatial virtual environments (GeoVE) and dialogue-assisted visual environments focusing on visual representation, exploratory interaction, visually-enabled dialogue, and team/group collaboration in the context of geovisualization. From the perspective of Geography, Gong and Lin (2000) present the concept of virtual geographic environments, and define a virtual geographic environment as a human-centered environment that represents and simulates geographic environments (physical and human environments), and allows distributed multi-users to implement exploratory geospatial analysis, geocomputation and geovisualization, and to conduct collaborative work for supporting design and decision. The building of virtual geographic environments needs to integrate such technologis as GIS, distributed virtual environments and CSCW. The major aim of the virtual geographic environment is to allow traditional geographers to carry out their research work on comprehensive and complex geo-problems in an efficient and innovative way on a data and graphics driven integrated platform.

In this chapter, we employ theories and technologies of collaboration and distributed virtual geographic environments to first explore the features and framework of CVGE and then carry out the design and development of the CVGE system.

Framework and system design

Conceptual Framework

What are the specific characters of geo-collaboration, and the differences from general collaboration? The framework for geo-collaboration is an important and difficult issue. In view of landscape modeling, Fall et al. (2001) address a collaborative framework that divides participants into three overlapping groups: all participants including decision makers and stakeholders, domain experts, and a core team of 3-5 people. MacEachren et al. (2001, 2004) contend that visual representations plays a significant role in geo-collaborative activities, and the development of a conceptual framework for visually-enabled geo-collaboration, which comprises six dimensions: problem context, collaborative tasks, perspective commonality, spatial and temporal context, interaction characteristics, and tools to mediate group work. Mandviwalla and Khan (1999) argue that there are many invented collaborative support features and that the challenges of usage centre upon how to integrate these features into a single usable tool. From the viewpoint of technologies for geo-collaboration, we first develop a conceptual framework of CVGE followed by the system framework and design.

We argue that geo-collaboration is involved in four components: geographic environment, geo-tasks, task-related geo-problems, and multi-participants. Multi-participants mediated by collaborative tools act in unison to accomplish geo-tasks and solve geo-problems contained within a geographic environment. From the implementation level, a framework of CVGE may then be designed by including virtual geographic environments and virtual collaborative studios and avatars expressing participants who can carry out communication and collaboration within virtual the geographic environment or virtual collaborative studio.

A virtual geographic environment is virtual representation of a physical geographic environment. It is also a virtual environment in which geo-spatial and social events occur (i.e. task-related geo-problems exist) and geo-spatial experimental results can be visually observed. A virtual collaborative studio is similar to traditional laboratories or physical discussion rooms, enabling distributed participants to interact for hypothesis shaping, dialogue implementation, control and justification of experimental schema and experimental parameters and decision making supported by a variety of virtual experimental tools, such as a virtual parameter justification device, shared text editors, shared graphics/maps/images editors, and a profile analysis device.

We consider, thus, the virtual geographic space with its geo-problems and the virtual studio space with its avatars and collaborative tools are the defining features that separate geo-collaboration from general collaboration. The space of the virtual collaborative studio may bear no relation to geographic space. In practice, in a virtual collaborative studio, it is easy for avatar-based participants to see and meet with each other. However, in a virtual geographic environment, the scaled space is often very large and at a different spatial level where participants easily disappear and lose visual connections with others.

System Framework

According to the conceptual framework noted above, and in the context of the application of CVGE for dam system planning in watersheds, a framework for establishing a CVGE system can be designed in five levels, namely network level, data level, modeling level, graphics level, and user level (Fig. 3). At the network level, distributed network communication structure such as client-server and peer-to-peer decides the capabilities of CVGE (Singhal and Zyda, 1999). The data level deals with geographic data base, remotely sensed image base, application model base (e.g., ideal dam systems planning model), user data base, streaming media data base, and collaborative data base. These may be both the data sources for establishing the virtual collaborative environment and the data produced within the collaboration process. At the modeling level, comprehensive geographic modeling includes 3-D modeling of the geographic environment and the representation of task-related geo-problems (e.g., different types of dams, and possible flooded regions and area after dam building). Avatar modeling is involved in the 3-D graphic representation of users’ identities and behavior and expression. To specifically support communication and interaction among multi-users, mediated tools may include text-based dialogue, text whiteboard, graphics whiteboard, and streaming media. Collaborative process modeling is often concerned with task dispatching, operation collaboration, roles, etc. At the graphics level, 3-D graphic worlds including virtual geographic space and the virtual studio, as well as the graphic representation of task-related geo-problems are taken into account. Finally, the user level considers the user’s knowledge base, technical capability, role playing, and ages. Also considered at this level are the user’s computer and network conditions.

Fig. 3.  A five-level framework for establishing a CVGE system

Distributed System Architecture

The aim of the CVGE system in our study is to be able to support collaborative dam system planning across the Internet. Considering the complicated Internet environment, in particular, the varied data-handling and 3-D graphics-rendering capability of users’ computers and the current limited Internet bandwidth and data-transfer speed, this chapter adopts a balanced client-server approach to the design of the CVGE architecture.

A balanced client/server structure signifies that some tasks are handled on the server side while others are handled on the client side (Fig. 4). The working mechanism is as follows: a client on the client machine sends a request, such as a change of VRML world or spatial query on the CVGE user interface via Internet to the Http server on the host machine; the Http sever passes the request to an application sever; the application server then connects with a data base server and retrieves data from the data base if it is needed; the application server then prepares a response (image/graphics, text, 3-D data or other data/information) and sends this back to the client; a VRML browser or Java applet/application on the client machine then displays or processes the response data on the client side.

Fig. 4. Distributed System Architecture of CVGE

Server-side applications, as shown in Fig.4, include the VGE servers, the virtual collaborative studio servers and the data base server. The VGE servers include the VRML geo-world generating server the dam management server, the dam system planning server and data query and analysis server; The virtual collaborative studio servers include the VRML studio management server, the multi-user management server, the text and graphics whiteboard management server, the text-based dialogue management server, the streaming media management server and the collaborative process management server. The data base server manages all geo-data concerning geographic environments and task-related geo-problems, and new data produced in the collaboration.

When a client-side user navigates a large geographical space, the VRML geo-world generating server will constantly receive users’ requests and retrieve corresponding data from a file processing server. The server will then produce a new VRML world and transfer it to the client-side; The dam management server is designed to manage 3-D dams that have been added, edited, or deleted by users in collaboration, while the dam system planning server deals with computation of topographic properties such as channel network and sub-catchment boundaries, quantitative relation of dam height and silt land area of reservoir; The data query and analysis server handles geographical data queries by connecting with the database server ( e.g., dam number and attributes), and spatial data analyses such as sub-catchment area computation.

The VRML studio management server is responsible for processing the 3-D room and 3-D objects such as chairs, desks, and pointing rod in the room. The multi-user management server retrieves and forwards messages and information about the user's name, position and orientation from and to online users. The text and graphics whiteboard management server is used for editing text and drawing lines through collaborative whiteboards for the free discussion and the representation of new ideas. The text-based dialogue management server effects the message communication between users via text, while the streaming media management server provides real and natural communication with the online users via real-time audio and video media. The collaborative process management server deals with task dispatching and management of roles and operations.

Geographic Data Model

The 3-D representation of geographic environments and task-related geo-problems is a key factor in the capabilities and efficiency of the CVGE system. In this study, we adopt the OO idea to design our data model. As a natural way of observing and constructing worlds, the object-oriented (OO) methodology is now preferred for dealing with data modeling or system design and development.

As for objects in landscape and ecological systems, there are two typical models: field based and entity based. Field based models include the digital elevation model (DEM), aspect model, slope model, soil distribution model, and land use model. Entity based models include individual, discrete entities such as trees, buildings, roads and dams. In terms of CVGE, real-time navigation and interaction and realistic 3-D graphics are key factors in enabling users to feel immersed and present. However, since geographical data are always large and complicated, geo-entity modeling requires special modeling methods. Two such models are discussed in this paper, i.e. topographical landscape models and 3-D entity models.

Topographic landscape models

Topographic landscape models can be defined as the integration of geometric and non-spatial attribute distribution models such as DEM, slope, aspect, land use and soil type. Because of the large data volume of DEM and its geometric and thematic complexity, as a foundational 3D spatial framework, topographic landscape models are the keys to realistic, real-time handling of 3-D scenes. This chapter addresses multi-block and multi-level schemes to model complicated topographical landscape models. In general, when a handled region is too large to carry out data transference across the Internet and geo-computing and graphics rendering in real-time, we need to divide this region into several blocks. In Fig. 5, we design R´C blocks to represent a region. Each block may comprise many levels, describing different geometric and thematic details. Fig. 5 demonstrates that the block S₂₂ is represented by T₁, T₂,..., T_i,..., T_n in terms of spatial resolutions, and by A₁, A₂,...,A_i,...,A_m in terms of  thematic resolutions. Meanwhile, we can hypothesize that resolutions increase concomitantly with an increase of i.

  Fig. 5. Multi-block and multi-level representation 

In practice, the number of (R´C) of blocks is determined by the real-time handling capacity of client computers and the highest spatial (T_n) and thematic attribute (A_m) resolutions that depend upon the data sources. The maximum observing range of users decides the lowest spatial (T₁) and thematic attribute (A₁) resolutions. When application systems are executed, users can choose the most appropriate spatial and thematic attribute levels for their own requirements, Internet capability and computer performance. Given a specific computer environment, the number of triangles handled for real-time 3-D graphics rendering should be fixed, and is presumed to be NR. When the region defined as “R” needs viewing and is selected, users should choose a spatial level that allows the number of triangles within the region R to be equal or similar to NR. Thus, it is possible for users to implement real-time visualization and analysis.

l        3-D geo-entity models

In traditional 2-D GISs, geographic entities such as buildings, trees, and dams are usually represented by color or 2-D graphic symbols. General users are not familiar with abstract, 2-D symbols and worlds for representing real, 3-D, geographic worlds; a conversion from abstract, 2-D symbols to 3-D geo-entities may place a heavy load on information cognition and processing.

In this chapter, we design a 3-D graphic object base to simulate 3-D geo-entities so that users can directly identify geo-entities and feel immersed in their virtual environment. The building process of 3-D geo-entities begins with the use of 3-D application systems such as AutoCAD and 3D Studio Max to create 3-D geometric objects. Together with the geographic co-ordinates, size, and orientation of entities obtained from the ground survey (or the dam planning), the 3-D geometric objects can be further processed into 3-D objects for constructing CVGE. These objects are stored and managed in a data base server.

A case study: prototype system and rudimentary application

Application Scenarios

The Loess Plateau in China is one of the areas of the world that has suffered the most severe damage from soil erosion. To help save the fragile eco-system of the Loess Plateau, a total of 163,000 silt dams are scheduled to be constructed, with a tentative budget of US$ 10.027 billion by 2020. Silt dams- or so called silt-trappers – are constructed across gullies to trap silt that erodes during heavy downpours from watersheds or through small river systems. Meantime, silt dams form a natural underground reservoir, which benefits the local water cycle and vegetation. When the dams have silted up, the increased supply of water ad nutrients they provide increases the land’s suitability for farming.

In the case study of the Qiu-Yuan-Gou watershed, with a total area of 70.7 km² and a length of 18 km for the major channel, there are 202 silt dams, of which 17 are the backbone dams (big dams) and 185 are the ordinary dams (middle and small dams). The complete dam system in the Qiu-Yuan-Gou watershed is the typical example model dam system for the massive silt dam construction project on the Loess Plateau. This study employs the watershed to explore the methodologies of building a CVGE for supporting spatial planning of silt dams across the Internet.

Prototype System

3-D world construction

As previously mentioned, there are two types of models in terms of ecological landscape systems. One type is the field based model type, such as the DEM and land use model; the other is entity based, i.e. the model uses entities, such as dams, trees, and roads.

In our case study, we utilize Arc/View and our own data preprocessing system to build 3-D field-based objects such as DEM. The building process begins with the ArcView GIS 3D Analyst, which is used to produce a triangular irregular networks (TIN) model of DEM from the contour theme. We then employ our own data preprocessing system，which is developed with VTK API ( VTK,2005), to import the TIN model and build multi-level DEM objects. In terms of the thematic aspect of topographic landscape object, Arc/View is applied to producing multi-level, raster-based models for thematic attributes. When 3-D graphics rendering is implemented, these thematic, raster-based models are taken as texture images mapped onto the surface of DEM. Fig. 6 shows the 3-D, virtual watersheds with a sub-regional DEM in the Jiu-Yuan-Gou watershed, overlaid by thematic maps of a SPOT 5 remote sensing image and of a channel network, while the DEM and the thematic attributes are both represented in the highest spatial resolution. The 3-D field-based objects are stored in files on the server computer and can be accessed through file processing servers.

Fig. 6.  3-D images of a sub-regional DEM in the Jiu-Yuan-Gou watershed covered by a SPOT-5 remote sensing image (left) and a channel network (right)

With regards to the 3-D virtual studio (including desk, chairs, avatars, etc.) and 3-D individual entities such as dams in the Jiu-Yuan-Gou watershed, the 3D Studio Max is firstly utilized to create their 3-D models in VRML format. Then our own data preprocessing system imports and converts these models into a specific 3-D CVGE system processing format. All the 3-D objects are stored and managed by the Access database server. In the database, exempt from the essential spatial and thematic attributes, every 3-D entity has more attributes (fields) such as ID number, object name, metadata, center coordinates, 3-D boundary box, information related users, and general description.

System development

Using Java, Java3D and VRML technologies, a prototype CVGE system for dam systems planning in watersheds is developed.

The 3D browser for the virtual studio is developed with Java and Java3D supporting collaboration by providing mediated tools, while the VRML geographic world browser for virtual geographic environment is freely available on the Internet. In Fig. 7, the BS_Contact_VRML system was used for browsing the VRML geographic world (Bitmanagement, 2005). As we do not need to develop VRML browsers (plug-ins) ourselves, we are able to focus on the development of client-side Java applications to communicate and  interact with VRML geographic  worlds, or application servers on the server-side.

In an addition, avatars, 3-D graphics representations of human bodies and body behavior in 3-D virtual space, are often used for social interaction among online users. They can move in both the virtual studio and the virtual geographic environment. The locations and orientations of avatars change depending on viewpoint positions and view directions. VRML is often used to model avatars. Users can freely select their avatars, with their choices leading to different perception, cognition, and social behavior in each user.

Fig. 7 is a snapshot demonstrating the client-side interfaces of the virtual collaborative studio (Fig. 7-a) and virtual geographic environment (Fig. 7-b). The detailed functions are described in the next section.

a: Virtual Collaborative Studio        b: Virtual Geographic Environment

Fig. 7.  A Snapshot of the Client-Side Interface of a CVGE System

Rudimentary Application  

Collaboration implementation in the virtual collaborative studio

Multi-participants can implement collaboration in the virtual collaborative studio through many channels. If participants wish to publish their opinions ad ideas, each individual can start a text or graphics editor to write texts or draw a graphics ( Fig. 8 ). When finishing editing, the participant presses OK and the text or graphics appears on text and graphics whiteboards in the 3-D virtual studio on all on-line client computers via the server (see Part A and Part C in Fig. 7-a). Participants can also view others’ dynamic video images in real time. The image of Part B in Fig. 7-a is used for video-based natural communication. Part D is an avatar describing a participant’s identity. Users employ avatars to feel others’ existence, activity, and spatial location and orientation. Part E is used for text-based talking with online participants.

              Text  editor                      Graphics editor

Fig. 8.  Text and Graphics Editors

Dam planning implementation in the virtual geographic environment

In Fig. 7-b, the interface of the virtual geographic environment comprises three major windows. The upper window in Fig. 7-b is a 3-D world viewing window; the bottom left hand window is used for choosing parameters of spatial and thematic levels and thematic type. It is used for data query and object operation (addition, removal, rotation, translation, and scaling) and for geo-computation of topographic properties of a DEM. In addition, it is used for distributing dams by participants and ideal dam systems computation. The bottom-right window is used for displaying a 2-D map of the Jiu-Yuan-Gou watershed, which closely connects with the 3-D world viewing window.

Participants can navigate in the 3-D geographic space and meet together virtually via avatars, communicating with each other using collaborative tools provided by the virtual collaborative studio. Depending on the performance of users’ specific computers and the data transfer speed of the Internet, users can set up appropriate spatial and thematic levels for a rapid 3-D graphics rendering and interaction in (near) real time. For example, the 3-D virtual world in Fig. 6 describes the terrain of the Jiu-Yuan-Gou watershed with the spatial level set at 1. Also, different thematic models can be added in 3-D, virtual space. In Fig. 6, a channel network theme is selected and is viewable in the 3-D world viewing window.

In the virtual geographic environment, coordinate measurement and data query can be conducted. In the 3-D world, users may use the mouse to measure the co-ordinates of any point. The bottom output window shows the co-ordinates (x, y, z) of selected points. In data query view, the virtual geographic environment provides users with two methods: one is spatial, and allows users to select objects in the 3-D world for querying the object information from the data base server; the other method employs keywords to obtain information about keyword-related objects.

The virtual geographic environment can allow users to add, remove, and edit 3-D objects directly in the 3-D scene. Before adding an object, a specific object type and position for adding the object should be selected. For object removal, users select an object in the 3-D world, and press the Remove button to complete the object removal operation. Meanwhile, the selected object disappears into the 3-D world. Added/removed objects are all managed by the data base server. Furthermore, the added objects can be shared, i.e., when a user edits an object in the 3-D world, other online users can also see it. As shown in Fig.7-b, some 3-D dams were added to the virtual world. This functionality of the virtual geographic environment, together with avatar-based and text-based communication among online users, allows remote collaborative work in the virtual country park to be easily and effectively carried out.

The virtual geographic environment provides functionalities for dam systems planning. At the time of writing, such models as topographic properties calculation and ideal distribution of dams had been accomplished. In Fig. 7-b, participants can click the Topographic Properties button and several data files such as channel network and sub-catchment boundaries will be calculated. With regards to the spatial distribution of dam systems, users can first set up some dams based on their own knowledge. Fig. 9-a shows a 2-D distribution of dam systems set up by one user, which can be modified by other online users. In Fig. 9-a, there are three types of silt dams: backbone, middle and small. According to the practical production, the backbone silt dam should be responsible for an area of 3-10 square kilometers with the middle and the small silt dams for less than 3 square kilometers. After the initial spatial setting of dams, the system will then automatically compute the ideal distribution by a rule such as the shortest time of silt deposited land, see Fig. 9-b. Based on the ideal distribution of dam systems, future dam spatial planning model work will be needed for the consideration of constraints of flood detention in safety, social (e.g. population distribution) and economic (e.g. investment) elements for a practical application of the dam systems planning system.

  a: Dams set up by users       b: Ideal distribution of dams

Fig. 9. Initial setting of dams and auto-computation of ideal dam systems

Discussion and future trends

CVGE integrates the technologies of GIS, remote sensing, distributed virtual environments, multi-media, and CSCW. The successful development of CVGE for practical dam systems planning in watersheds continues to face many challenges. Deficiencies in the knowledge base hamper the process of securing optimal benefit from the virtual collaborative studio, as there are insufficient data for recording the viewpoints and ideas and for supporting the collaborative process in the context of dam planning in watersheds. For the 3-D viewing of virtual studio and virtual geographic environment in this study, we used the Java3D viewer, due to the need to develop a range of collaborative media tools to optimize our usage of the VRML browser. A geographic world, constructed with a large amount of data, may be very large, making it difficult for the VRML based geo-world browser to facilitate efficient processing and real-time navigation and collaborative interaction across the network. Thus, a unified, 3-D display and powerful browser for CVGE should be designed. Considering its application, spatial silt dam systems planning in watersheds on the Loess Plateau in China is a new area of research and practice that has not yet a mature theory and methodology. This study has completed some functions, such as the computation of topographic properties and ideal spatial distribution of dam systems. At a practical level, however, there remains a need for further development of soil erosion models, forecasting models of flood detention in safety, optimization models of dam systems planning with a number of constraint factors, estimation models of investment, evaluation models of economic and ecologic benefits and comparison model of different dam systems planning schemas.

As an integrated technology, CVGE is involved in heterogeneous data and application models of complicated geographic and ecologic environments, the complex problems, and dynamic change factors related to human collaboration. From a practical viewpoint, the three-tier client/server architecture used in this study is limited in its capacity to handle integration and sharing of diverse resources, and to assure security. However, the Grid technology promises to support the vision to process the above issues. The Grid architecture enables flexible, secure, coordinated resource sharing among dynamic collections of individuals, institutions and resources (Hertzberger, 2003). In addition, resources in this context include computational systems and data storage and specialized experimental facilities. Thus, we argue that the Grid based VGE for geo-collaboration should be given a priority in future research. If the Grid based VGE is realized, it could become a part of e-Science that will change the dynamic way science is conducted (Taylor, 2001; RCECP, 2003). 

Conclusion

Technologies and mediated tools are key to the implementation of geo-collaboration. To date, there are no commercial GIS systems available for collaborative work. This paper reports the design and development of the CVGE system by integrating GIS, distributed virtual environments and collaboration. Geo-collaboration varies from general collaboration in the context of complicated geographic environments into which complicated geo-problems are implanted. Meantime, two spaces regarding the virtual geographic environment and virtual collaborative studio are defined for their different functions. The virtual studio space is a 3-D space for collaborative work, which has no necessary association with the real world, while the space of the virtual geographic environment is used for representing and simulating the real, physical geographic environment. Under these conceptual models, the system framework, data model, distributed architecture, and interfaces of the CVGE system are presented. In the case study of dam systems planning in the Jiu-Yuan-Gou watershed, a prototype the CVGE system is developed with Java, Java3D, and VRML. Participants can implement 3-D spatial navigation in the virtual studio or the virtual watershed, and communicate with each other via collaborative tools such as text and graphics whiteboards, 3-D avatars, streaming video media, and text-based dialogue windows, and carry out the editing of shared dams, calculation of topographic properties and ideal spatial distribution of silt dam systems. 

Acknowledgement

This research is partially supported by National Natural Science Foundation Project No. 40341011, the Innovative Project of the Institute of Remote Sensing Applications, Chinese Academy of Sciences No. CX020021, and National High Technology Plan (863) Project No. 2001AA135130 and No. 2002AA135230. We would like to thank other members of this project for their assistance, in particular Ms. ZHOU Jieping, Mr. LI Wenhang, Mr. CHEN Zhen, and Ms. DU Wei.

References

Batty, M., Didge, M. , Doyle S., & Smith A. (1998). Modelling Virtual Environments, GeoComputation: A  Primer ( P.A. Longley, S. M. Brooks, R. Mcdonnell, and B. Macmillan, editors ).  Jone Wiley & Sons, New York, pp. 139-161.

Benko H., Ishak, E., & Feiner, S. (2003). Collaborative Visualization of an Archaeological Excavation. NSF Lake Tahoe Workshop on Collaborative Virtual Reality and Visualization ( CVRV 2003), October 26-28,2003. http://www1.cs.columbia.edu/graphics/projects/ArcheoVis/

Bitmanagement (2005). BS_Contact_VRML, http://www.bitmanagement.de/

Blaxxun (2005). URL: http://www.blaxxun.com/.

Carey, R., & G. Bell (1997). The Annotated VRML 2.0 Reference Manual. Addison-Wesley Developers Press, 501p.

Cheng, C.Q., Hu X.L., & Ma A.N. (2003). Concept and Framework of Collaborative GIS. Geographic Information Worlds, 1(1):23-29. (in Chinese)

Craig W J, Harris T M, & Weiner D, eds (2002). Community participation and geographic information systems. London: Taylor and Francis，2002. 1－5.

Cooper, W. (2000). MUDs, Metaphysics, and Virtual Reality, The Journal of Virtual Environments ( Electronic Journal ), 5(11), January, URL: http://www.brandeis.edu/pubs/jove/HTML/v5/cooper.htm.

Cybertown (2005). URL: http://www.blaxxun.com/.

Densham, P.J., Armstrong, M.P., &Kemp, K.K. (1995). Collaborative Spatial Decision-Making Scientific Report for the Initiative 17 Specialist Meeting, 16^th-19^th September, 1995, Santa Barbara, California. National Center for Geographic Information and Analysis. http://www.ncgia.ucsb.edu/Publications/Tech_Reports/95/95-14.pdf

Doyle, S., & Dodge, M. (1998). Toward virtual london: developing a virtual internet GIS. Proceedings of  International Conference on Modeling Geographical and Environmental Systems with Geographical Information  Systems, 22-25 June, 1998, Hong Kong, pp. 624-629.

Dykes, J., Moore, K., & Wood, J. (1999). Virtual environments for student fieldwork using network components, Int. J. Geographical Information Science, 13(4): 397-416.

Fall,A., Daust D., &Morgan, D.G. (2001). A Framework and Software Tool to Support Collaborative Landscape Analysis: Fitting Square Pegs into Square Holes. Transactions in GIS, 5(1):67-86.

Faust, N. L. (1995). The Virtual Reality of GIS. Environment and Planning B: Planning and Design, 22:257-268.

Gong，J.H., & Lin, H. (2000). Virtual Geographical Environments and Virtual Geography. In: Forer P, Yeh A G O, He J, eds. Proceedings 9th International Symposium on Spatial Data Handling, Beijing, China, 10-12 August 2000. 28～39.

Gore，A . (1998). The Digital Earth: Understanding our planet in the 21st Century，  given at the California Science Center，Los Angeles，California，on January 31，1998，http://www.digitalearth.gov/VP19980131.html.

Hertzberger, L.O.(Bob) (2003). Research Infrastructures & Grid, http://www.surfnet.nl/bijeenkomsten/escience/ hertzberger/hertzberger.ppt.  

Hibbard, W. (1998). VisAD: Connecting People to Computations and People to People. IEEE Computer Graphics,32(3):10-12.

Hibbard,W., Rueden C., Emmerson S., Rink T., Glowacki D., Whittaker T., Murray D.,Fulker D., &Anderson J. (2005). Java distributed components for numerical visualization in VisAD, Communications of the ACM 48, No. 3, 2005, 98-104.

Jankowski, P. & Nyerges, T. (2001). GIS-supported collaborative decision-making: Results of an Experiment. Annals of the Association of American Geographers, 91(1), 48-70.

Li, J. , & Wen, Z. (2001). Adorning VRML Worlds with Environmental  Aspects, IEEE Computer Graphics and Applications, January/Febtuary, 6-9.

Lin, H., & Gong, J.H. (2002). Distributed Virtual Environments for Managing Country Parks in Hong Kong -A Case Study of the Shing Mun Country Park. Photogrammetric Engineering & Remote Sensing. Vol. 68, No. 4, 369-377.

MacEachren, A. M., Edsall, R., Haug, D., Baxter, R., Otto, G., Masters, R., Fuhrmann, S., & Qian, L. (1999). Virtual environments for geographic visualization: Potential and challenges. Proceedings of the ACM Workshop on New Paradigms in Information Visualization and Manipulation, Kansas City, KS, Nov. 6, 1999, pp. 35-40.

MacEachren, A. M. (2001). Cartography and GIS: Extending Collaborative Tools to Support Virtual Teams. Progress in Human Geography, 25(3), 431-444.

MacEachren, A. M., & Brewer I. (2004). Developing a conceptual framework for visually-enabled geocollaboration. International Journal of Geographical Information Science. 34(1): 1 – 34.

MacEachren, A. M. , Cai, G. , Sharma, R., Rauschert, I., Brewer, I. , Bolelli, L. , Shaparenko, B. ,Fuhrmann, S. , & Wang, H.(2005). Enabling collaborative geoinformation access and decision-making through a natural, multimodal interface. International Journal of Geographical Information Science,19(3): 293 – 317.   

Mandviwalla, M., & Khan, S. (1999). Collaborative Object Workspaces (COWS): exploring the integration of collaboration technology. Decision Support Systems, 27(1999),241-254.

Manoharan,T., Taylor, H., & Gardiner, P. (2002). A Collaborative Analysis Tool for Visualization and Interaction with Spatial Data. ACM Web3D’02, February 24-28, 2002, Tempe, Arizona, USA. 75－83.

Normand, V. (1999). The COVEN Project: Exploring Applicative, Technical, and Usage Dimensions of Collaborative Virtual Environments. Presence: Teleoperators & Virtual Environments, 1999, 8(2):218-237.

Oliveira, J. C., Shen, X., & Georganas, N. D. (2000). Collaborative virtual environment for industrial training and e-commerce. In Proc. Workshop on Application of Virtual Reality Technologies for Future Telecommunication Systems, IEEE Globecom’2000 Conference, http://www.mcrlab.uottawa.ca/papers/CVE-WS-VRTS2000.pdf.

OICQ (2005). URL: http://www.tencent.com/.

RCECP (Research Council e-Science Core Programme) (2003). e-Science. http://www.nesc.ac.uk/nesc/define.html

Rhyne, T. M. (1999). A Commentary on GeoVRML: a tool for 3D representation of georeferenced data on the web. Int. J. Geographical Information Science，13 (4): 439-443.

Singhal, S., & Zyda, M. (1999). Networked Virtual Environments: Design and Implementation. ACM Press, SIGGRAPH Series, New York, pp. 1-17.

Taylor, J.(2001). e-Science Core Programme. http://www.rcuk.ac.uk/escience/ .

Tecoplan (2001). DMUConference. URL: http://www.tecoplan.de .

Van Maren, G. & Germs, R. (2000). K2VI: A Virtual Reality Interface for the Spatial Database Engine. URL: http://www.infosystems.co.nz/k2vi/esri/p551.htm.

Verbree, E.，Maren, G. V., Germs，R. , Jansen, F. , & Kraak, M. J. (1999). Interaction in Virtual Worlds Views - linking 3D GIS with VR. Int. J. Geographical Information Science，13 (4): 385-396.

VTK (2005). URL: http://www.kitware.com.

         Wood,J., Wright, H., & Brodlie, K. (1997). Collaborative visualization. Proceedings, IEEE Information Visualization’97, Pheoniz, Oct. 19-24,1997,pp. 253-259.