UNIT 25 - TRENDS IN GIS

• Fast geoprocessing
• Parallel Processing
• Memory
• Workstations
• Networks
• Hardware for specialized processing functions
• Operating systems
• Peripheral devices
• Specialized workstations

• Database management systems
• Relational DBMSs
• DBMS versus Fourth Generation Languages
• GIS system integration
• Display products
• Interfaces to other technologies
• User interfaces

• Modeling and decision support
• Sciences and mathematics

• Remote sensing
• Error/uncertainty
• Data sharing

This unit reviews some of the current trends in GIS. This unit, in particular, will date very quickly. You will need to read the current trade magazines to see what the newest trends are.

UNIT 25 - TRENDS IN GIS

Compiled with assistance from Jack Dangermond, ESRI

A. INTRODUCTION

• this unit discusses
  • trends in computer hardware and software for GISs
  • new applications of GIS technology
  • new sources of data

Fast geoprocessing

• computing power is often measured in MIPS
  • Million Instructions Per Second

• MIPS measures oversimplify the measurement of computing power but are nevertheless useful as broad bases for comparison
  • arithmetic calculations can require execution of as little as two and as many as 100 instructions
  • arithmetic on real numbers is better measured in MFLOPS ("megaflops") or millions of floating point (i.e. decimal) operations per second
  • MIPS="meaningless information processing statistics"

• current personal computers and workstations used for GIS range from 1 to 5 MIPS

• within the next five years, 20 to 30 MIPS workstations are likely at roughly similar prices
  • the power of personal workstations is currently increasing by close to a factor of 2 per year

• workstations will likely be coupled to 1,000 MIPS file servers which can extract software and data at high speed for analysis in personal workstations

• over the next decade, 1,000 MIPS machines will become common in large organizations

• to this point, advances in computing power have always found new areas of application. What new applications of GIS will take advantage of higher speeds?
  • larger data sets, higher levels of spatial resolution
  • more complex models
  • more complex analysis for decision-making
  • better methods of display and visualization

• trend toward different computer architectures
  • away from single processors operating on data in sequence
  • parallel processors can perform tasks on several different processors simultaneously within the same computer

• what GIS processes will be suitable for parallel processing?
  • analyses which require repeating the same steps everywhere on the map
  • easier to see applications for raster data than for vector since each pixel is independent
  • e.g. finding route for a vehicle across a rugged terrain
  • e.g. image processing applications such as image classification, visualization, scene generation

• trend is toward lower costs for ever larger computer memories
  • cost of storing large GIS datasets will come down
  • more data can be placed "on-line" for faster access

• the nature of geographical data (high volume, infrequent update) is suitable for optical storage
  • once written, cannot be changed
  • CD-ROM - 5 1/4 inch disks with 250 MBytes, enough to store all streets in Los Angeles
  • once a "master" has been created, the unit cost of CD-ROM data is only about $10
  • optical WORM - 12 inch disks with 2 GBytes, enough for the contents of 100 topographic maps

• erasable optical disk is available (e.g. NEXT computer)
  • very high density of data

• "dumb" terminals connected to a central processor are gradually being eliminated in favor of desk top computers

• especially popular are "workstations", which have excellent graphic performance and sophisticated user interfaces
  • the exact distinction between personal computers and workstations is unclear
  • workstations are generally more powerful
  • workstations are generally more expensive ($10,000 vs. $4,000)
  • workstations generally use UNIX operating system rather than DOS
  • workstations have more powerful graphics capabilities (1280x1024 rather than 640x480)
  • originally, workstations developed for the scientific and engineering market

• workstations function effectively as nodes on a network devoted to GIS processing
  • a large CPU (a fileserver) for database management and centralized processing may still be required as part of the network, while most processing is done at individual workstations
  • data can be distributed around the disks on the network

• the dominant hardware system architecture of recent years, the multiuser host, is giving way to multiuser network architectures
  • the network integrates compute servers, file servers, workstations and shared peripherals
  • any user can access data, peripherals across the network
  • the network will likely be linked to other networks through "gateways"

• this architecture requires fast and economical data transfer and the availability of powerful workstations
  • data transfer rates of MBytes/second are common at low cost
  • only hundreds of bytes per second were possible 15 years ago
  • this change of cost has had enormous impact on the ways people organize computing

• hardware manufacturers are beginning to offer network architectures, and networking among hardware of different vendors is likely to become more available over time
  • requires interchangeability of parts; standards for communication, data and software; common operating systems

• this will lead to reduction in the power of data processing centers in favor of "information management" systems based on control of transactions
  • the role of the "computer center" based on a large mainframe is changing rapidly

• compute servers, file servers, sort servers (e.g. TRW's Fast Data Finder) and search servers (e.g. Excel's Sorting Engine) are now being developed for networks
  • these are specialized computers attached to networks for specific functions
  • map overlay using hardware intersecting tools will be developed in the future, perhaps other GIS functions as well

• this trend will continue since such hardware tools can provide enhanced system performance for the entire network

• continued diversity is likely for the immediate future

• UNIX is making gains, perhaps especially in inexpensive machines, and in scientific and engineering applications, but is it likely to become universally supported?

• this will make software development and networking more difficult, and puts a premium on GIS software, database management systems, and applications which can work on different vendors' machines

• excellent raster devices are now available (e.g., electrostatic and laser printer/plotters) for graphic/cartographic output
  • costs of these systems are directly related to the size of the product

• scanning has not taken over from digitizing in GIS applications, and is unlikely to do so until some means, perhaps artificial intelligence, can be applied to separating extraneous information

• a data entry device is needed which will allow correction of data as they are acquired
  • will probably have a large flat display, multiple graphic memory planes, and interactive data capture capabilities
  • the workstation should be able to "check-out" and "check-in" work areas (e.g. mapsheets) from the larger database maintained by a server on the network

• an "electronic sandbox" will be useful for interactive, GIS-based analysis/modeling and land use planning
  • the design of this workstation will require some very creative thinking

• workstations specialized for particular uses (e.g., land planning, water resources, forestry) are likely to be developed as the number of users increases in such specialized fields
  • analyzing data on the globe (e.g. oceans, atmosphere) will require a specialized workstation which can display data on the globe's curved surface
  • e.g. the globe could be "browsed" using a track-ball to rotate the image

• as GIS becomes a standard decision support technology, entire conference rooms will be devoted to its use
  • containing specialized GIS workstations, large GIS display devices, and GIS planning/conference tables

Database management systems

• while present DBMS's are effective for managing tabular data, they are not effective for the "long transactions" required when cartographically referenced and topologically related data are altered
  • such "long transactions" occur because simple changes in cartographically referenced data may require changes in topologically related items, and because the whole process of updating geographic information differs from updating tabular data

• transactions on geographical data could be confined to a single specialized workstation

• queries and analysis typically do not modify the data, so these are easier to execute across the network, e.g. between a workstation and a central fileserver

• trend toward using relational DBMSs (often with SQL style user interfaces), because of their "open architecture"
  • it is becoming easier to exchange one DBMS for another within a GIS

• object-oriented database structures are being proposed for GIS use, but they present some problems
  • e.g. they are not well adapted to storing natural feature information
  • not well suited to complex spatial analysis
  • often have embedded proprietary data structures

• the DBMS approach often involves highly structured application programming, often at the expense of ad hoc query capabilities
  • user must learn complex rules of syntax
  • may be a valid approach for static databases which are only used for simple, repetitive queries

• the trend in GISs is toward the use of Fourth Generation Languages (4GLs) which provide commands, tools, procedures, and report writers to permit easy ad hoc querying of a database
  • these provide intelligent interfaces close to natural language, however, definitions become vague, less rigorous
  • use of 4GL may detract from the GISs ability to perform complex analysis

• the marketplace increasingly demands compatibility between diverse hardware and GIS software

• at the same time, GIS software needs to interface to an increasing diversity of DBMSs, because different applications often require different DBMSs
  • in many applications, records are already stored in a DBMS
  • when the GIS capability is added to allow geographical access to these records, it must interface with the existing DBMS

• improved cartographic products are certain, since there is both intense user demand and the technology required to support such improvements
  • map output will continue to be judged against hand- made products

• 3D displays, overlaid with both cartographic data and representations of the built environment, are likely early developments in GIS technology
  • but major questions remain about how to gather, compile, model and structure 3D data

  • because of these problems, it is not clear what kinds of analysis are needed or appropriate for 3D data

• interfaces between GIS, CADD, remote sensing, image processing, architectural graphics, and other technologies are going to be increasingly easy to create

• the differing data types produced by these technologies will be more frequently combined in shared databases

• more sophisticated, flexible and well managed graphic user interfaces are inevitable

• users are becoming increasingly impatient with software which requires any training or support from the vendor
  • training and support add high and probably continuing costs to GIS acquisitions

• because GIS technology is becoming more affordable, more reliable, more widely used and better known, new applications of GIS technology are likely to rapidly increase, just as the applications of computer graphics have increased

• Geographic Information Modeling System (GIMS) technology will be developed and used in providing decision support in a growing number of fields

• e.g. current interest in GIS applications in transportation planning, requires some modifications to standard GIS models, addition of new functions
  • e.g. modifications to allow lines which cross but do not intersect
  • e.g. functions to measure distances between objects via the network
  • e.g. functions to solve standard problems in transportation, such as predicting traffic flows

• can applications in areas as different as transportation planning and forestry be served by the same GIS software?
  • will there be fragmentation of the GIS field by area of application?

• GIS technology will be applied widely in the sciences in the near and middle term future
  • e.g. 3D capabilities for geology, geophysics, hydrology, mining
  • GIS modeling in landscape ecology

• in the longer term, applications for GIS technology may develop in areas of image processing, e.g. X-rays, other types of medical imaging, where superimposition of data, analysis may have similar importance

• global issues - tropical deforestation, acid rain, greenhouse effects, endangered and threatened species, and similar problems are likely to be analyzed using GIS technology in the 1990s
  • GIS networks, similar to the global weather monitoring and prediction network, may evolve
  • they will probably make use of super computers, parallel processing, and artificial intelligence to cope with the massive databases and the complex models involved
  • such models are currently in very rudimentary form, e.g. global climate models used to predict greenhouse warming effects use very large cell sizes (5 degrees lat/long)
  • GISs will play a role in managing the global environment, perhaps used to identify the world's most sensitive habitats; then a country's agreement to conserve these habitats may be exchanged for forgiveness of international debts

Remote sensing

• following Landsat, ERTS, Thematic Mapper and now EOS, and competing with them to supply the demand for satellite- borne imaging systems are the SPOT system and systems which the Japanese, Russians, and others may bring on line in the 1990s

• as data resolution increases, costs fall, service becomes more reliable, and user demand increases, satellite remote sensing will become more important in supplying data for GIS use
  • already the supply of data vastly exceeds our ability to analyze it
  • better methods of scanning, archiving large amounts of data will be needed

• remote sensing cannot provide the information required for many kinds of analyses
  • in the longer term (and with such new technologies as the fabrication of ultraminiature sensing devices on silicon chips), the trend will be toward "an instrumented universe"
  • instruments to monitor the globe will be broadcast over the earth, probably telemetering their information to networks of users

• as more data and more forms of data are gathered, and there is increased pressure to combine these data for analyses, increasing attention will have to be paid to data error and uncertainty

• essential to reduce costs, solve widespread problems, and fully utilize available technology

• security considerations, political divisions, and other factors will continue to inhibit sharing

• because problems will increasingly be recognized as crossing international boundaries and having global implications, and because the necessary technology is now becoming available, more global databases will be built

• in some situations, private businesses will compete with government agencies in supplying data to both government and the public
  • questions will increase about how to best serve the public interest in such cases
  • data gathering by government will always be under fire because of its cost

• the immediate future will be a time of explosive growth in the development and use of GIS technology
  • the signs are that current growth rates will continue

• it is very likely that these predictions about the period will prove to have been too conservative
  • predictions about the value of technological change often prove to be totally wrong
  • e.g. the development of microchip technology which was the key to cheap computers was driven first by the need to save space and weight in space vehicles

Dangermond, Jack and Morehouse, Scott. 1987. "Trends in Hardware for Geographic Information Systems," Proceedings AUTOCARTO 8, ASPRS/ACSM, Falls Church, VA. (Also available from ESRI)

Dangermond, Jack, 1987. "Trends in Geographic Information Systems Software," Proceedings IGIS: The Research Agenda, NASA, Washington, DC. (Also available from ESRI)

U.K. Department of the Environment, 1987. Handling Geographic Information, Report of the Committee of Enquiry chaired by Lord Chorley (the "Chorley Report"), Her Majesty's Stationery Office, London.

DISCUSSION/EXAMINATION QUESTIONS

1. Imagine a few important future developments which you believe ought to occur in GIS technology. Do they depend chiefly on hardware or software, or both? Explain.

2. Select a particular discipline, field, or specialty to which you think GIS technology might be applied. Discuss some specific future applications and their implications for professional practice in that field. What specialized GIS features would these applications require?

3. What are the characteristics of an "ideal" source of data for use in a GIS? List them and discuss each one. Now, are there any potential sources of GIS data which have not yet been utilized but which meet most of these criteria?

4. Design a GIS workstation for the support of global science (i.e. the analysis and modeling of data for the entire globe). What operations, database models and user interface features would it have?

Back to Geography 470 Home Page

Back to GIS & Cartography Course Information Home Page