Dr. Winnie S. M. Tang
Managing Director, ESRI Hong Kong Limited., Level 10, Cyberport 2, 100 Cyberport Road, Hong Kong,
Tel: (852) 27306883, Fax: (852) 27303772, E-mail: [email protected]
Jan Robert Selwood
Project Manager, ESRI Hong Kong Limited.,
Tel./Fax: 81 5617 2 8887, E-mail: [email protected]
Introduction
This year is the 10th anniversary of the first Web based Geographic Information Service. Xerox® Palo Alto Research Centre (PARC) launched the Map Viewer service in June 1993. Celebrations in June 2003 are likely to be muted as PARC has moved onto other research initiatives and the Map Viewer was discontinued in 1997 (PARC Company, 2002). Its legacy is however impressive, as can be seen by the explosion of web based mapping that has occurred since.
It is notoriously difficult to gauge this ‘explosion’ by numbers of users. van Elzakker (2001:42) for example, makes reference to a survey in which 90% of respondents claimed to have used some form of web map; statistics provided by Media Metrix which indicated that map based weather and route finding sites appear within the top 40 most frequently accessed web sites in the US; and figures provided by MapQuest claiming production of 75.4 million maps for over 16.6 million user sessions in November 1999. These figures are now out-of-date. MapQuest alone claims 400 million maps requested by 60 million user sessions as its average monthly rate for 2002 (MapQuest, 2002); ESRI’s Geography Network shows peak traffic of up to 100,000 maps per hour (ESRI, 2002). These examples have been joined by an expanding list of public access web sites that provide map visualisation or spatial query/search functionality. Added to these are a host of specialist web servers providing dedicated mapping services to banking, tourist, routing, mobile telecommunications and utilities industries. Whilst the examples used by van Elzakker may be out-of-date they remind us of two important facts:
- firstly, comparing like with like (van Elzakker’s MapQuest statistics with their current (November 2002) rates) it can be seen that there are roughly 3.6 times as many users producing 5 times as many maps as there were 3 years ago; and
- secondly, such figures become out-of-date as fast as they are published, and more often than not reflect only a fraction of the services provided, and as a result cannot necessarily be taken as a true guide to the impact of a particular technological trend.
This paper first looks at the development of web-based Geographic Information Services since the PARC Map Viewer, and explores their impact on geospatial computing rather than attempting to measure impact by total numbers of users. It then briefly examines technologies that are only beginning to take advantage of web based computing – web based data networks and data editing functionality, which, it is suggested will sustain and expand the impact of the web on Geographical Information Systems (GIS) for the foreseeable future.
Evolution and Impacts of Geographic Web Services
The PARC Map Viewer permitted zoom and pan functionality and enabled searching for predefined geographic locations. This introduced the concept of presenting geographic data on a web page in a format that can be interrogated; in other words in a format other than a static picture. The Internet or Intranet could be used to support web-based client access to, and interrogation of, centralized geographic information – this was of major significance for the GIS.
From its inception GIS has struggling with a dilemma inherent within geographic data and analysis. On the one hand, geographic information and reasoning is deeply integrated with decision making at all levels – whether in connection with our daily lives, our businesses, administration and government, homeland security or emergency response. In addition, often the same data and the same services are used by widely differing organisations for widely differing purposes – the same street plan, or boundary outlines for example, are used by schools, the police, the local authorities, government and businesses alike. On the other hand, storage and analysis of geographic data is both complex and resource intensive. As a result, geographic or ‘spatial’ data has traditionally been very poorly integrated with other information delivery techniques – an observation applicable both to traditional hardcopy maps, but particularly so to computerised information systems. Not only has it been difficult to integrate GIS with other technologies, dissemination of GIS data and functionality has also been severely restricted as analytical complexity, the size and format of data sets involved and the limitations of network bandwidth have conspired to restrict GIS largely to a high end workstation application. Distributed access has been both limited and relatively expensive. GIS therefore, provided the tools to meet the widespread demand for geographic analysis in decision-making, but lacked a way of disseminating these.
The Web changed this. It brought renewed focus on data standards, transfer formats, and an awareness of the benefits of open storage and development environments. Within GIS technology this change is apparent in the introduction of spatial objects to Object Relational Databases (such as ESRI’s ArcSDE and Oracle Spatial), and the move towards using open format data structures such as ESRI’s Shape file. It also led to the development of data transfer formats and applications designed specifically for delivery and interrogation of information across thin networks. The move away from proprietary development languages (e.g. MapBasic, MDL, AML) to standards such as COM and Java have also introduced an object oriented focus on encapsulation, componentisation and polymorphism which has seen very large complex applications (unwieldy and difficult to maintain, upgrade and integrate) being broken into lightweight, loosely coupled code. This permitted many geospatial services to be rendered as small applets and plug-ins that could operate across thin Internet or Intranet connections. Ultimately this lead to the increased sophistication of international standards such as Open GIS Consortium’s (OGC) Geographic Markup Language (GML) and other open variants of the Extensible Markup Language (XML) language such as ArcXML which facilitate transfer and exchange of intelligent geographic data between different applications.
The significance of these developments is that they address the issues of dissemination and integration. Initially web-based GIS primarily permitted visualisation and simple query of spatial information, however increasingly; dedicated spatial web servers can perform complex spatial operations such as geocoding and route networking. Since the beginning of 2002 these are being provided as web services delivered in various forms of XML with their own Simple Object Access Protocol (SOAP) wrapping and Web Service Description Language (WSDL) descriptions, permitting complete integration of spatial functionality into other applications.
Not only is web based GIS resolving problems with dissemination of geographic data and analyses – it is making these cheaper by dramatically altering the cost:benefit ratios of GIS implementation when assessed against user accessibility. This is shown schematically in Figure 1.
Figure 1: GIS Implementation Cost:User Ratio
However, as Figure 1 shows, though overall cost of GIS ownership may have fallen, costs of data ownership remain significant. Such costs include collection or purchase of original data as well as data management and continued maintenance and update. To date, apart from permitting increased view access to datasets, web technology has had relatively little impact on the way data is actually managed. Future significant reductions in the cost of GIS implementation will need to come from addressing the collection, maintenance and management of data. The remainder of this paper considers two web-based initiatives that may go some way to addressing this: web data portals and web-based editing.
Web Based Data networks
Difficulties in disseminating data have meant that more often than not the same data is duplicated by each organisation using it – a complete copy of the same base map can be at each organisation access it. It is not uncommon for the same dataset to be duplicated even between different offices and departments within the same organisation. This is not only wasteful in storage; it incurs a significant overhead in data management and update, and inevitably leads to inconsistency between the datasets used by different organisations (or parts of organisations).
Web based data networks seek to promote data sharing between organisations and permit multiple organisations to locate and utilise a single copy of a dataset without the need to duplicate it. This initiative differs from data warehouses that sought to concentrate data management of all required data in a single massive database. Where data warehouses fail is in the understanding that often data is best maintained and stored at source – with the organisation that is responsible for collection and update – rather than in massive centralised systems.
The data network is designed around the web services concept of Publish, Discover, and Use. Data service providers are responsible for maintaining their data and for making it accessible to other users on the web. Such access may be permitted free of charge (as in the case of some Government’s or international organisation’s datasets) or on a subscription or one-off-charge basis. The provider is also responsible for providing a basic description of the service that can be posted to a central registry. Documentation includes metadata (such as the Federal Geographic Data Committee (FGDC) or ISO standards), as well as a service description increasingly provided using the Universal Data, Discovery and Integration (UDDI) and WSDL protocols.
At the heart of the network is a central web site that acts as registry or ‘Yellow Pages’ to which data providers can register their service and through which users can discover data offered. If the user decides to use a service, the register will pass them to the respective data provider’s site from which data can be drawn directly. The adoption of open Internet transfer standards (such as GML or ArcXML) means that an increasing number of desktop GIS applications can directly read and integrate such data drawn from the web. Using a package like ArcView or ArcInfo, for example, a user can now integrate and work with data from a local hard-drive, network and multiple web sites. The result is an open, multi-participant, community based network that encourages data sharing and communication.
This has a number of implications:
- Clients do not need to maintain their own copies of standard datasets – they can be accessed directly from the supplier;
- Suppliers can provide more flexible data supply contracts based, for example on: intensity of usage and/or areal coverage, which can be tailored to meet the specific needs of a client;
- The introduction of economies of scale can permit levels of investment in data update and maintenance and the provision of server security and backup facilities, that are well beyond the means of most organisations, providing not only better quality data, but also real 24:7 reliability of service;
- Competition brought about by publishing the availability of data will ensure that data supply costs are kept to a minimum.
The Geography Network launched by ESRI in June 2000 pioneered the concept of spatial data networks. The Geography Network was established as a registry and discovery engine for spatial data and application services offered by a large number of public, private and commercial organisations. Highly scalable, the Geography Network model has been duplicated around the world (e.g. in New Zealand, UK, Holland, Hong Kong), with each new node linking into the wider network framework. The Hong Kong G.Net (www.gis-webservices.com) is illustrative of this trend.
The Hong Kong G.Net (HK-G.Net) offers an expanding range of datasets covering HK SAR including:
- general base map information;
- demographic and census data;
- point of interest (PoI) at various levels (public services, commercial, tourist etc.);
- Yellow Pages location directories;
- a complete navigable road network (including turn directions);
- Orthophotography; and
- Satellite imagery.
These datasets are derived from a range of data suppliers both commercial and public, and are either hosted locally by the HK-G.Net itself, or by the suppliers themselves. The G.Net site permits data to be viewed, searched and interrogated. As data is served in ArcXML, live data can be served to any ArcXML compliant application. Some of the datasets (for example imagery and census data) require payment before they are made available to the user, either on a one-off purchase or subscription basis. The registry handles payment through direct linkage to 3rd party payment gateways.
The HK-G.Net is not only permitting greater access to, and integration of, a wide range of datasets, it provides location search and mapping functions in the form of web service components. The Simple Open Access Protocol (SOAP) wrapped location and visualisation functions permit organisations to integrate maps and spatial search functionality directly into their local applications/site. Thus for examples, banks and property agents that do not, themselves have any GIS or mapping software, are using the service to integrate maps generated by the Hong Kong G.Net into their own websites and internal systems. They do not have the additional overhead of purchase and management of separate software and datasets, and are assured up-to-date, accurate mapping provided in a format that can easily be integrated.
As the Hong Kong G.Net is part of the wider Geography Network framework, it exposes local datasets to the global community – helping to provide awareness of the highly detailed data resources in one area of the world and to make these readily available for potential users. As more organisations establish web based data networks the volume of data that can be easily searched and accessed is increasing dramatically.
Web Based Maintenance of Data Assets
For organisations that still need to develop and maintain their own datasets, data maintenance and update remains a significant cost. To date, update of spatial data has been largely restricted to dedicated workstations running relatively expensive software and located within the same LAN as the databases being edited. Despite the fact that frequently data collection/ update is being undertaken in the field or remote offices, facilities for remote update – across either fixed WAN or Internet connections – have been limited.
This is beginning to change. Based on web services technology and standards, web based spatial edit applications are appearing. These rely on the ability to write, transmit data in standard formats (XML or Vector Markup Language (VML)) across the web, and convert these to standard GIS formats such as Shape, ArcSDE or Oracle Spatial.
Web editing tools offer organisations maintaining data a cost effective approach to disseminated editing functionality across either intranets or the Internet. This is of significance as it permits the update and maintenance of data within central data stores to be carried out remotely – allowing data to be updated from the field, site office or remote station. Data update is no longer restricted to LAN on which the central database is stored, it is accessible to WAN and Internet connections.
One example of such initiatives is WebEdit (we) from TIG Centre, which appeared in July 2002 (www.tigcentre.com). Based at present on ESRI technology, it operates within a standard ESRI ArcIMS environment and focuses on editing ESRI suite of data formats (ArcSDE and Shape). It is briefly described below as an illustration of the functionality and approach taken by this kind of technology.
As is shown in Figure 1, in principle we operates in much the same way as a traditional editing application. The we Server is responsible for handling data extraction, transaction management, post back and conflict resolution and edit tracking. The client selects elements to edit, edits these locally and then posts them back to the server. Where we differs is that all client activity is undertaken in a standard Microsoft Internet Explorer web browser.
Figure 1: we Components and Operational Relationship
we is based on four distinct components:
- we client – handles user interface providing view control, query, selection, edit and print functionality;
- we server – handles edit session administration, data extraction and synchronisation, roll back and edit tracking;
- we website – provides access and run-/one-time download JavaScripts; and
- ArcIMS Server – used to provide the basic map service carrying data to be edited.
The we Server operates in conjunction with ArcIMS server. It uses ArcIMS Image services to broadcast map images to clients, and to establish the location of Shape or ArcSDE files with which it then directly communicates to extract or post back features for client based editing sessions. we Server runs continuously monitoring responses from data sources (e.g. database locking or post-back failure signals) and requests from clients. Communication with we clients is established through opening a dedicated port on the we Server computer. When changes are committed, they are passed from the client station to the server that handles update of the main data store and records a complete transaction history.
Figure 2: we Server roll back functionality
The transaction history is stored in a simple database table (we uses Microsoft Access by default), linked to the graphic features changed and can be used to roll back any changes at any time (Figure 2).
The we Client provides a complete editing tool kit with standard map navigation, search, query and printing functionality as well as a comprehensive range of editing tools. The client connects to one or more ArcIMS image services to provide background mapping.
Features selected for editing are passed by the we Server to the Client in either Vector Markup Language (VML) or weXML an XML compliant data transfer standard. Editing tools include copy, paste and rotate functions, adding and editing line, point, polygon and vertex features as well as offset, snap and split operations (Figure 3). Edits can be input through normal cursor operations (the system adopts the coordinate system defined in the ArcIMS image service) or through a COGO entry option permitting surveyed features to be updated. Through access to the open we client API, tools can be split off and integrated within any corporate ArcIMS Internet map site.
The Client is developed in DHTML and JavaScript and operates within a standard Microsoft IE web browser (Figure 3).
Figure 3: we Client interface
The ability to provide simple, easy to use editing functionality in packages such as we go a long way to addressing some of the constraints on current data update and editing operations. They permit central databases to be updated directly from any web-enabled location. Editing functionality can also be distributed extremely cost-effectively across intranet or Internet connections freeing workstation resources from mundane editing jobs, and permitting users directly involved in data collection to update the core dataset directly from their desks.
Conclusions
This paper has considered the development and impact of web based geographic information services. As has been shown the PARC Map Viewer and early web based visualisation tools initiated a trend which is now permitting GIS to become seamlessly integrated with the decision making at every level. Web based visualisation and query tools resulted in widespread access to existing geographic data stores. Recent focus on web services within the GIS is permitting these to be directly integrated with applications that have no spatial functionality in themselves. This has significantly reduced the cost per user ratio. Further reduction of this ratio is likely to come from addressing issues within data management and update which remain. This paper has considered two new initiatives: web based spatial editing and web based data networks that begin to address these issues. Data networks allow data to be managed and maintained by the suppliers and permit users to easily search and make use of data without incurring the overheads involved in management and maintenance. Users gain access to cost effective, up-to-date and well-maintained data services, whist the data providers reap economies of scale that permit them to concentrate on providing extremely robust backup and security services and increasingly flexible pricing schemes. Web editing tools offer organisations maintaining data a cost effective way of disseminated editing functionality across either intranets or the Internet and ensuring that those responsible for updating data can do so wherever and whenever required. Between web based data networks and web based editing, web based technologies will continue to drive GIS development and to ensure its ongoing integration with other technologies.
Bibliography
- ESRI, 2002, Geography Network,
- MapQuest, 2002, About MapQuest¸
- PARC company, 2002, PARC Map Viewer has been deactivated,
- TIG, 2002, we User and Administration Manual, Hong Kong, TIG Centre
- van Elzakker P.J.M., 2001, “Users of Maps on the Internet”, in Kraak M-J and Brown A, Web Cartography, London, Taylor Francis.
