CHAPTER 6 Continued)

6.5.2.6 Other variations in GIS software

These may include:

1. “User-friendliness” - is the command language or menu system readily comprehensible?
2. The ways in which graphics are integrated with attribute features.
3. The structure of the DBMSs.
4. The topological structure used.
5. The quality, quantity and diversity of product promotion.
6. The ability of a software package to service a range of hardware devices, including workstations.

6.5.3 The Capabilities of Some Typical GIS Packages

For this section we have quite deliberately selected a small number of GIS software packages with the object of illustrating some of the tremendous range obtainable.

6.5.3.1 IDRISI

This is an extremely low-priced, raster-based GIS, which is able to provide the range of functions shown in Table 6.4. It was developed at Clark University in the U.S. and was specifically created for microcomputers. It is designed to be an inexpensive research and teaching tool that can become the focus for a collective program of system development and exchange. By using independent modules linked by a simple data structure, the system allows users to develop their own modules in any computer language. Users may also exchange new modules, data sets and experiences through an IDRISI electronic bulletin board.

The IDRISI system currently operates on all Intel 808x microcomputer systems running PC-DOS or MS-DOS 2.11 (or greater), using 5.25" or 3.5" disks. System requirements are few. Minimally, it requires 128 kilobytes of internal memory and two floppy disk drives, though optimally the system should have a hard disk, a colour graphics card and a dot matrix printer. With a typical 640K memory configuration, the system can theoretically process images as large as 10 000 columns by 32,767 rows. Figure 6.19 illustrates output capability using the “Image” module, in this case some Landsat TM data. Up to 32 half-tone grey levels can be printed. The latest version is 3.2 released in late 1990, at three price bands of <$150. Further details can be obtained from Eastman (1989).

6.5.3.2 MundoCart/CD

This is an example of a single function data base designed for use independently or for integration with a GIS. It is of particular value to aquaculture and inland fisheries, as is demonstrated in Table 6.5 which shows MundoCart's feature classifications. This software produces seamless high-quality maps of any part of the world in a range of scales from 1:250 000 to 1:150 000 000. Users can add their own data to base maps produced by the package or extract cartographic data from it for use with other data bases and with GIS, i.e. it eliminates the need to draw or digitize map outlines. The MundoCart map data base is contained on one CD-ROM - associated software enables the information to be retrieved and manipulated. It is designed for PC use and is distributed by Chadwyck-Healey Ltd.

MundCart contains 20 million mapping points defining the earth's surface, including 500,000 separate features and 60 000 names of places and features. A choice of 24 different projections and 26 ellipsoids is available, and in addition to the complete data base, MundoCart can be purchased in eight regional subsets or combinations of subsets. The complete data base costs about $15 000, though a 50% educational discount is available. Regional subsets are priced at about $3 500 for the first and $3 000 for additional ones.

6.5.3.3 GIMMS

Information on this software is obtained mostly from Maguire (1989) though complete details can be had from Carruthers (1985). GIMMS is a high level, command-oriented, computer cartography system that can be used in both batch and interactive modes. It was initiated in 1970 at the University of Edinburgh, and has subsequently been updated on several occasions. It is currently one of the most widely used cartography packages. It is available for a wide range of mainframe and minicomputers, and is about to be released for microcomputers. GIMMS may be used to create two basic types of cartographic image, i.e. statistical graphics such as line graphs, pie charts and histograms, and maps of points, lines, areas and surfaces. GIMMS is organized as a series of modules each of which covers one aspect of the image creation process. Figure 6.20 illustrates an example of GIMMS output.

Figure 6.19 A Portion of a Print Out From the IMAGE Module of IDRISI

The microcomputer version of the full GIMMS software package will be available in 1990 for the IBM PC/PS range (or compatibles) using MS-DOS or OS/2. It will require an 80286 or 80386 processor and at least 3 Mbytes of memory plus a hard disk with at least 5 Mbytes of available space. Peripherals supported include all screens supported by GSS/IBM plus virtually all major plotters. The price of the GIMMS-PC software will be about $1 700 with a 50% discount for academic users.

Figure 6.20 An Exmple of GIMMS Output Showing Some Population Statistics for Jamaica

6.5.3.4 ARC/INFO

This GIS is supplied by Environmental Systems Research Institute (ESRI) from California, and the company claims this product to be the world leader in GIS in sales/installations. ARC/INFO is designed to run on a variety of 32-bit mini and mainframe computers plus IBM PC/AT compatible microcomputers and 32-bit workstations. The system performs a complete range of GIS functions (Table 6.6), and emphasis is placed on an efficient relational and topological data model for entry, analysis and management plus clear user interface incorporating “pull-down” and “pop-up” windows. ARC/INFO uses an integrated Henco INFO RDBMS, but it also allows the user to integrate other RDBMSs such as ORACLE, INGRESS and dBASE III. This integration introduces increased functionality including transactional data management, multi-user access, data dictionaries and the use of screen editors for attribute entry and updates.

ARC/INFO is composed of two tightly integrated software components: ARC, used to manage cartographic data, and INFO, used for managing tabular data describing cartographic attributes. The entire ARC/INFO system operates in five different computing environments: Prime, DEC VAX, Data General, IBM (VM/CMS) and Sun plus the IBM PC/ATs. The software supports numerous drivers for graphics screens and it contains the Tektronix IGL Library which allows users to write their own device drivers for virtually any device. ARC/INFO supports a variety of normal plotters plus high resolution, colour electrostatic plotters. The prices for software on mini and mainframe computers range from $20 000 to $90 000, depending on the size of the machine, number of packages and location. The PC version is sold as a series of individual software modules, each costing $1 000 – $2 500. PC ARC/INFO can be acquired as a complete “turn key” system, including hardware, for under $25 000.

6.6 The Benefits and Problems of Using GIS

Since GISs are so complex, and since the technology and infrastructure behind GIS has emerged so recently and rapidly, then there are certain to be problems which the potential user must face. On the other hand, GIS would not have expanded so quickly unless there were a great number of benefits. We shall consider benefits and problems from the actual and potential users viewpoint - those within the industry might visualize a completely different set of benefits and problems. Since a list of these might be vast we have selected arguably the 10 major ones in each category. In looking particularly at the problems of using GIS, we would bring attention to one paramount concern which has been repeatedly emphasized and is summed up in Dept. of Environment (1987) - “The success or failure of a GIS effort has rarely depended on technical failures, and almost always on institutional or managerial ones.” (p.154). It seems to many that a tremendous degree of effort and speed has been put into GIS technological and theoretical developments - maybe a pause to catch up and reflect is now overdue! Useful further reading on benefits and problems can be had from Burrough (1986), Dickinson (1989), Goodchild and Gopal (1989) and Tomlinson (1989).

6.6.1 The Benefits of GIS

1. GISs help the planner to make trade-off decisions in resource allocation, i.e. using an interactive VDU, varying assumptions and criteria can be instantly simulated and manipulated to generate different final scenarios, or to test varying hypotheses. This allows the planner a range of choices from which a decision can be made.

2. GISs allow for a range of data, perhaps from widely differing spatial sources, to be entered into an analysis scenario - much of it also coming from different fields of study. For example, in site selection for coastal aquaculture, field work data, RS imagery, secondary maps and tabular data from the fields of recreation, resource extraction, biological diversity, scenic valuation, existing land use, etc., could all be rapidly integrated into the decision-making process.

3. GISs are applications-oriented. They bring a spatial technology immediately to bear on real world problems. They can instantly show the crucial part that both location and spatial differentiation play in optimizing socio-economic well-being.

4. GISs have greatly expedited the speed of working across a whole range of functions. It has provided its user with rapid access to a huge volume of data ensuring that decisions or outcomes have an added degree of objectivity, this being accomplished in some situations in near real time.

5. Once digitized, any GIS data can be quickly and efficiently updated, meaning that more frequent revisions become possible.

6. Time series changes can be monitored rapidly, and statistical calculations of these can be quantified, allowing future projections to be estimated.

7. The cost-effectiveness is such that GIS allows a company to be competitive in various ways, e.g. in optimizing its location. It also allows public services to be more efficient and cost-effective.

8. GIS allows for the production of special purpose (perhaps 1-off) maps to be produced at low cost, and for the propagation of a large variety of maps, or other output data, which might not otherwise have been produced.

9. GIS technology enables a high quality of output to be reached and standardized by a large number of people who might have no cartographic or draughting skills.

10. For the research and academic user, GIS not only enhances the opportunity to describe, explain and predict spatial patterns and processes, but they also allow sophisticated and realistic models to be formulated and tested. Maguire (1989) suggests that GIS may stimulate the development of a new philosophy which integrates the work of human and physical geographers plus those in a number of related disciplines.

6.6.2 The Problems of GIS

1. Many of the benefits of GIS cannot be assigned a monetary value, e.g. intangible benefits such as “better decision-making”, “better planning” or “better information”. These benefits could well be greater than the actual increases which can be measured such as costs of producing maps or consultancy costs. The actual value of acquiring a GIS is very difficult to demonstrate.

2. The extremely slow process of digitizing existing maps data means that the user must incur a high cost for the digital conversion, i.e. up to 75% of total monies spent on GIS by some users is in data capture or conversion. This means that much potentially valuable data will remain “locked up” for many years.

3. The poor quality of much existing hard-copy graphical or tabular source data which, when input to the GIS, causes error propagation. Some errors are caused by the “fuzziness” of the entity, e.g. “The Alps”. How does data covering this vague region get precisely captured? Other errors occur as a result of using dated sources, poor digitizing, poor sampling, processing mistakes, conversions of data, classification mistakes, etc. Algorithms need incorporating which allow for estimated confidence limits of the output data, plus error-train analysis to assess the potential results of propagating errors.

4. Who will be responsible for updating digital data and deciding which institutional data will be regularly updated?

5. The means and the rights to gain access to information. At present they not only vary greatly from country to country, but the legal framework is often either extremely poorly determined or is unnecessarily restrictive. There are legitimate copyright problems in making all data available to the public, plus the problems of ownership of digital data, e.g. if a new set of data is produced from another owner's source material.

6. There is a critical shortage of trained people in GIS, i.e. in understanding the technology, in spatial analysis methods, in applications design, etc. There are too few training places and teachers, inadequate research and funding plus a lack of understanding that the problem even exists! This situation is likely to continue in many countries whilst governments show little interest in geographic data handling.

7. How to cope with the huge data influx which is exponentially occurring, e.g. by 1994 the EOS satellite alone will generate 10 million million bits of data per day, and a simple U.K. Ordnance Survey sheet occupies about 240 million bits. This data somehow needs storing, archiving, structuring and processing - better methods must be found. How do we evaluate if the data being acquired is really necessary?

8. There is no international format, or set of agreed standards for GIS data, which would enable data of known quality to be easily transferred among different users. Associated with this is the fact that much data is collected by groups who might not wish to exploit the spatial aspects of it, so they are not obliged to provide the data in a form which might facilitate spatial analysis.

9. Many systems still lack “user-friendly” interfaces which would make them more readily accessible to non-expert users. These people may understand what they wish to get from the GIS, but they do not wish (or have time) to absorb either vast instruction manuals or the intricacies of computer operation. In many cases this might have led to user resistance to the new technology.

10. The problem of getting access to data in the sense of actually finding what the user requires. Digital archives are hard to get information on (let alone from!), and browsing through data bases via on-line local or regional networks can be slow and ill-referenced. When data is found there can be scaling, processing or formatting problems.

6.7 Considerations in Choosing a GIS

GISs are successful when they comprehensively and consistently meet user requirements - world-wide a large number of GISs have failed to live up to expectations because in the planning stage user requirements were not identified and because users had not realized the correct potential of GIS. Users have also not made a detailed study of the data resources available or of the ability of the user or his organization to manage and maintain the system.

These considerations, plus the permutations of software and hardware available, can make the selection of GIS a complex procedure.

The impetus for wanting to implement a GIS has frequently come from an individual within an organization who has learned about it from a conference, workshop or the IT media, and who has advocated its adoption. This internal innovation has often been from the “bottom up” because many senior managers, unlike their junior counterparts, have failed to keep progress with IT. Tomlinson (in Dept. of Environment, 1987) notes that this “bottom up” approach has often led to unsuccessful GIS implementation because it has been piecemeal and lacking in wide company recognition. For successful implementation it is vital that the approach is systematic, looks at a very broad field and that a written report is produced which sets out a total implementation strategy, or perhaps a number of alternative plans, plus a very clear statement of the functional needs. The basic strategy report should set out to answer a number of basic questions, which we discuss here under various pertinent headings.

In compiling this chapter we will place the context of choosing a GIS on a variety of scales and in several situations. We need not look at the actual purchasing procedures and we will presume that a decision to implement computerization is agreed and that external facilities are available, e.g. office space, power supplies, etc. For further details on systems selection we recommend - de Man (1984), Burrough (1986), FICCDS (1988), Dangermond and Morehouse (1989), Guptill (1989), Montgomery (1989), Pearce (1989), Ventura (1990) and Pearce (1990).

6.7.1 Why Is a GIS Needed?

Since most companies or organizations will have progressed reasonably well without a GIS, it is very important to ask this! The answer lies both in tangible and intangible ways. From the aquaculture and viewpoint it has been made clear that there is an urgent need to secure and optimize suitable locations for fish production. From the inland fisheries viewpoint maintenance or enhancement of fish production from natural and man-made waters through ensuring good water quality and adequate quantities is of prime importance.

But as well as this it must be clear that GIS could be utilized by Fisheries departments in other tangible ways, e.g. in identifying market outlets, in identifying fish landing and processing sites, in allocating areas for exploitation for cooperative management of resources, in planning labour deployment and the supply of fishing gears, etc. Once GIS experimentation was completed and data bases were added to, then, if the experiences of other users is to be believed, other less tangible uses would be developed and deployed, e.g.:

1. In what way do fish catches vary across, and depth-wise, in an expanse of water, and why?
2. Where do we have areas of protein deficiency?
3. Is fish equally acceptable throughout our region, and if not, where and how much effort should we place on promotion?
4. What are the correlations between fish acceptability (for consumption) and income?
5. With what social or physical features do present aquaculture facilities best correlate?
6. Are aquaculture and fishery extension services properly allocated?
7. Etc.

Needs are also related to efficiency. Careful consideration would need to be given as to the range of efficiencies, and implied cost savings, that GIS would bring, e.g. perhaps in resource deployment, in transport distribution patterns, in real time answers to quite complex spatial problems, in optimizing the temporal/spatial output of production by species, etc. If GIS is not adopted, it is possible that relative efficiencies would decline, leading to loss of competitiveness and a consequent loss in commercial market share, or in the case of government, to the poor allocation of resources.

We list a number of other pertinent questions regarding user needs:

1. Can the needs be better met elsewhere? That is, is the need for a one-time assessment, or for fairly continuous analyses?
2. How urgent are our needs?
3. How would existing employees accept GIS being introduced?
4. What degree of “project awareness” should be raised in the organization?
5. How much say will the GIS users within the organization have in the acquisition decision?
6. Are our needs for a GIS expanding or declining - or are they flexible?
7. Do we need a GIS which provides simple drafting plans or highly sophisticated spatial analyses?
8. How much do we know about GIS - as a company - as individuals?
9. Where could we view a commercial or agency (rather than demonstration) GIS in action and speak to other users?

6.7.2 How Should a GIS Fit Into An Organization?

Many Fisheries departments, or other organizations contemplating securing a GIS, will have existing computing facilities, and there would certainly be a number of cost and personnel advantages to be had by sharing these facilities. But are these facilities available? - and in the right location? - and of the right “size”? - and have the desired peripherals? - etc.? Experience has shown that it is not easy to fit a GIS into existing facilities because GISs involve so many complex and systems-specific considerations, though GIS integration is certainly worthy of evaluation.

Perhaps the GIS should be a specific system which is accessible to a range of related disciplines, i.e. to personnel within a fisheries department, or to agriculture and fish production. or to food procurement generally. Or should it be even wider than this, e.g. a GIS which covers the entire resource planning or land use planning for a region or country? The answer to this will usually be determined by the structure and size of the organization contemplating GIS entry. We make the point here that much of the data to plan for agricultural development, or to assess environmental impacts are the very same data required to plan for fisheries and aquaculture. Therefore, there is a natural cohesion among the needs of various disciplines that could be satisfied by a single administration at a single location. However, it is usual that the optimum location for siting the GIS is closest to where most of the data bases are stored, or near to where data is available for capture, so this might determine the functional range of users who have access to one GIS.

It would be generally considered that the development of GIS is a highly centralized function, perhaps affecting the organization as a whole, and requiring resources which are beyond those of a “small Fisheries office”. A central place, “head office” type of location is preferable for initial GIS instigation, though as the system develops and more local data accrues, then some of the work might be suitably carried out at GIS outposts. Certainly, individual Fisheries offices should not be allowed to develop their own preferred GISs, since the chances are that a diversity of incompatible systems would develop.

Some further pertinent questions are:

1. What is the current state of information technology within the organization?
2. What spatially-referenced data is already held by the organization in computer readable form?
3. Where is this data held, and is it able to be shared and on what basis?
4. What new data sources are required?
5. Can any data held be easily integrated into a GIS, i.e. with regard to structure, topology, etc?
6. What degree of centralization or dispersal might be initially desirable?
7. Are plans underway to standardize data capture formats within the organization?

6.7.3 What Financial Considerations Should Be Made?

The instigation of any new development will entail costs and these usually need to be the subject of a cost/benefit analysis. It is important, and relatively easy to perform this analysis, but it is very much more difficult to assess the cost savings associated with the intangible advantages that the adoption of GIS brings. It is a good idea to monitor the costs of obtaining all data, so that at least for the future a bank of “cost experience” is being built up.

The actual capital costs of the initial set up will vary with the extent of the GIS application, the size and quality of items purchased and their source. The price range is enormous, varying from <$3 000 for the most basic processor, plus VDU, digitizing tablet, dot matrix printer and software, to >$200 000 for a highly sophisticated complete GIS system. If data is purchased then this can represent a very significant proportion of capital costs. The notional price, in terms of computer speed and memory plus quality of hard- and soft-copy output per unit of money spent, has been falling rapidly since the inception of GIS, and it is likely to do so for the immediate future. An important cost consideration is that developments in all aspects of GIS are occurring so rapidly, that most sub-systems are practicably dated in about five years and obsolete in ten. To keep up to date therefore, depreciation costs ought to be calculated at up to 20% per annum.

Most actual operating costs are quite low and comparable to other office activities though data capture can be very labour-intensive. However, given the sophisticated nature of the activity, the skill and managerial levels of personnel employed may be quite high, so costs here must be carefully considered. Also, in many government organizations new posts will have to be established. Thus, the problem is not only financial, but administrative as well. Burrough (1986) shows a very detailed analysis of personnel/cost inputs, and it is clear that there is a hierarchy which varies according to a general functional classification (from lowest to highest), i.e.:

1. Personnel needed for data input to GIS, e.g. digitizing, attribute entry, checking, inputting satellite imagery.
2. Personnel needed for data output, e.g. mostly different kinds of plotting.
3. Personnel needed to use software for data analysis, e.g. performing manipulation and analyze s within the GIS.
4. Personnel needed for planning and management.

6.7.4 What Sort of GIS Configuration Should We Adopt?

Since space prohibits an analysis of choice considerations pertaining to individual hardware items, we shall restrict our cover of this to a number of pertinent questions. We shall then make a consideration of alternative hardware configurations, realizing that it is difficult to isolate these from GIS cost and software considerations and that in practice it would be possible to configure hardware in an almost infinite variety of ways. The hardware choices cannot be clear-cut and decisions must relate to personal preferences, software choices, functional requirements, capital available, numbers of users, the degree of interaction with other systems, etc., plus the answers derived from some of the earlier questions.

The questions which need answering before investing in hardware include:

1. What are the relative merits of buying a “turn-key” system? (see section 6.7.7)
2. What formal evaluation methods (benchmarking) should be used to assess hardware suitability?
3. What after sales service is given and at what cost?
4. Where can independent advice on hardware or configurations be obtained?
5. How extendable, compatible, or flexible, are the various hardware components?
6. What is the hardware supplier's overall reputation?
7. How much storage space is likely to be needed, and how quickly do we need to work?
8. What quality, size, colour and range of output is required?
9. Do functional requirements dictate raster or vector data, or a combination?
10. How many users may be accessing the GIS and for what purposes?
11. How easy is the hardware to use, or the manuals to follow?

GIS configurations are combinations of hardware which, at a minimum allow one user to perform some of the GIS functional range, and at a maximum comprise hundreds of workstations linked to one another, to a complete range of hardware devices and to other computers, and which perform all GIS functions at rapid speeds backed by large memory stores.

Minimum practical configuration requirements can be met from a microcomputer (PC) based interactive graphics workstation. This system typically features a 32 bit processor, hard disk storage, perhaps a tape and disk sub-system, a digitizing tablet, a plotter, alphanumeric keyboard and a high resolution graphic display screen, all of which function together as a single “tool”. Since developments in microcomputers have been so rapid, these systems are now capable of running at 4 MIPS (million instructions per second), and of being linked with communication devices into larger networks. Systems management needs to adhere to strict guidelines since regular data back-up is seldom available. Though there are a number of other limitations on PC-based workstations such as poor ergonomics, limited processing power, restricted integrated software availability and difficulties of integrating with other equipment, they must be viewed as a very attractive start-up entry into GIS. Costs vary greatly but rise to about $30 000.

At a more sophisticated level, but using similar configurations, there is the engineering workstation or multi-user centralized system (Figure 6.21). This is based on a minicomputer, to which all the peripherals are connected. Peripheral devices are capable of little or no processing and users are dependent upon the processing power of the host computer and the data base it supports. Costs can be up to $100 000 for a complete functional GIS. They mostly operate under the UNIX operating system, and a vast amount of software covering all applications is available.

Various networks of computers, workstations and other hardware form the next level of configuration. A variation on the centralized single PC, or minicomputer engineering workstation configuration, is to form linkages of either of them based around a central processor, forming what is called a local area network (LAN) (Figure 6.22). Here the peripherals are connected to the LAN through device servers which control processing, and the limited storage problems of the PCs can be overcome by the use of a large external driver. Limitations on these configurations are the dependence on the single processor which affects response times, and the system is less amenable to incremental upgrades.

The definite trend in configuration, caused by technological advances, is towards distributed processing. Here processing power, and usually the GIS data bases, are decentralized across a network of graphics or “dumb” workstations, computer servers, file servers and other peripherals (Figure 6.23). Users have access to all the resources of the entire network and extra processors can be added when desired, as can other peripherals. Workstation nodes can be distributed throughout a large building on a LAN and, where a multi-site operation is desirable, wide area networks (WANs) are possible using microwave transmissions or cable networks for transferring data. This distributed processing approach is particularly efficient because it allows processing power and storage capacity to be available at the user's location thereby decreasing delays in response time. Data is distributed among users via DBMS software which acts as a “master librarian”. It is becoming increasingly common to bring additional data into the network from “value added network services” (VANs), i.e. companies who sell digital data via public telecommunication lines. Hardware manufacturers are increasingly offering networking capability for all of their products.

Figure 6.21 Multi-tasking Centralized Engineering Workstation (from Croswell and Clark, 1988)

Figure 6.22 Centralized Configuration Using a Local Area Network (from Croswell and Clark, 1988)

Figure 6.23 Multiple Processors and Peripheral Devices on a Distributed Network (from Croswell and Clark, 1988)

This hierarchy of GIS configurations, from small or basic to the large completely functional GIS, should in no way be seen as ranging from “bad” to “good” or from “make do” to “ideal”. The choice of configuration is governed by user's needs and, since extremely sophisticated GIS are now available at the lower hierarchical level, then user needs can be increasingly satisfied for a lower capital outlay. What a greater GIS configuration investment means is that the user's actual needs or requirements can be greatly upgraded. But before large investments are made, the user should ask himself whether the extra money would not be better spent in more comprehensive and accurate data collection systems, since they completely govern what the GIS is really capable of.

6.7.5 What Sort of Software Should We Choose?

Since there are now so many packages to choose from, the primary task is to determine which software best fits the operational requirements that have been previously established. It is unlikely that there will be a single best package. If the GIS is to be shared between departments or between operating functions, then it could well be that several more specialized packages will be required. If hardware is already owned then the nature of this, plus its operating system, will influence what software is available.

Some important points that a software purchaser might want to consider are:

1. Which software applications or system interfaces have the highest priority and why? Remember that software packages are often oriented towards providing a specific capability or supporting a specific application area, e.g. CAD, terrain analysis or image processing.

2. How the graphics are integrated with the attributes. Some systems place more emphasis on the intelligence of the data base describing one or other of these features.

3. What is the functionality of the DBMS, e.g. how well is a RDBMS able to “join” attribute data stored in one tabular list with that stored in others? Is the DBMS an integral part of the package or is it provided as an optional interface?

4. Does the software allow for continuous mapping? This is important since “seamless” data base structures are essential to many analytical problems.

5. How secure is the software supplier? - does he have a good record? - are his products known for versatility and reliability?

6. What kind of graphic user interface is desirable? - simple menus, “pop-up” menus, icons, windowing? - how user friendly is it?

7. How user-friendly is documentation? Manuals should be perused prior to purchase and if possible test runs of a package should be arranged or viewed.

8. Do we go for a raster or vector based approach? Though this consideration might be diminishing in importance as software systems more frequently cater for both, it would be important with regard to the structure of the data base sources being used.

9. Can the software easily be transferred to other GISs?

6.7.6 What Are Our Personnel Needs?

In section 6.7.3 we noted that personnel costs played a significant part in GIS costs, largely because of the skilled nature of much GIS work. In this section we will not look at skill acquisition per se; this will be done in section 6.8. Here we shall be concerned with the range of personnel needed.

Since applied GISs seldom function as a large-scale enterprise, then the numbers of personnel employed would be very low. For some of them GIS operations would only form part of their work, e.g. along with other computer work or as part of the management structure. It is likely that any skilled person working full time on GIS would need to be very versatile, i.e. because of the detailed and wide ranging knowledge required of computing, spatial analysis and GIS. In two studies, Kapetsky et al (1987) and Kapetsky (1989) it was found that for a GIS operation to be successful a cross-disciplinary team needed to be assembled. For a typical aquaculture study this might comprise a fishery biologist, a systems analyst specializing in RS image processing and GIS, a local fisheries officer, plus personnel to secure local information and information which was typically held in a central location (usually a capital city). Any of these personnel might usefully contribute to setting up a GIS, as well as advising on its further enhancement. Taking a different approach, Burrough (1986) classifies and exemplifies the range of work associated with GIS under two main categories:

1) “Low skills”. - to include typists, computer operators, map digitizers, draughtsmen, etc.

2) “High skills”: Managers - necessary for both the daily running of GIS and harmonious relationships between GIS and the rest of the organization.

Liaison - necessary to establish and maintain contacts with users. Technical - to include computer cartographers, programmers and systems development. Scientists - environmental scientists and others who use GIS for their research and applications programmes.

To obtain personnel with appropriate GIS skills or qualifications, there are a number of approaches:

1. For some of the less-skilled tasks, in-house training might be considered. This may only be a realistic alternative in a large organization where the costs can be more easily absorbed and where there are likely to be staff capable of offering instruction. Alternatively, staff skilled in computer hardware operations must be recruited.

2. Employ external instructors or send staff away on training courses.

3. For the more specialized tasks it is almost essential that outside recruiting would be necessary.

4. If some aspects of the required work was “occasional”, then it could be sub-contracted out to specialist companies.

5. If advice was being sought on any GIS matter, then outside consultants could be engaged.

The particular approach to obtaining personnel would vary not only from one organization to another, but also between countries, between functional levels of GIS and on a temporal scale, i.e. with growth in the scale at which the GIS operated.

In staffing a GIS operation any organization will need to consider very carefully its personnel mix, paying particular attention to:

1. What mix of internal and external personnel should be engaged to design or set up the GIS?

2. Do we initially fully contract out our GIS so as to see its potential?

3. What GIS jobs might we need to contract out immediately, or in the future?

4. Can we afford to recruit personnel specifically for GIS work related solely to aquaculture and inland fisheries?

5. Should the application of personnel to aquaculture and inland fisheries GIS be developed from a more general GIS operation?

6. What particular staffing mix do we need, i.e. between computer specialists and subject matter specialists?

7. What proportion of our personnel effort should be employed in data collection?

8. Through what means, and using what criteria, do we recruit our GIS personnel?

6.7.7 Procedures for Setting Up the GIS

We should now be at a stage where a strategy report has been produced so that the organization has a clear idea of the overall objectives for establishing a GIS, has some idea of particular user requirements and desired output and knows what its commitments will be in terms of costs, data and personnel, plus the capital software and hardware necessary to allow the GIS to function at a specified level of sophistication. It should also be clear where or how the GIS will fit into the organization from the operational viewpoint.

Having done this, a quite straight-forward set of procedures should be followed for project implementation, though these may vary between organizations or countries. These are:

1. Look for several suppliers of the necessary software. Nowadays there is a good chance that needs will be met by available packages, though software programs from several suppliers may be needed. A suitably shortened and revised form of the strategy report should be available for potential software and hardware suppliers which clearly sets out the relevant user requirements.

2. Before purchasing software, it is important to consider the hardware necessary. Software might have been designed to run on a specific machine, or it could run on other computers perhaps with a performance loss.

3. An alternative to looking at software and then hardware is to consider a “turn key” system, i.e. a combination of soft and hardware put together to satisfy a well identified demand. This could be either a completely “customized” system which is perhaps specialized in certain functions, or it could be an off-the-shelf system. Beware that “turn key” systems can, in practice, carry out all the functions which you have specified. And remember - you may be buying a lot of software capability which you will not require and the hardware may not be easily extendable.

4. Having narrowed down potential software and hardware suppliers to perhaps two or three serious contenders, each of these should be asked to do a “benchmark” test. This will involve running a series of practical simulations, of the types of functions identified as being important, and then comparing the results against the other suppliers and against quoted specifications. The tests might be on the time taken to perform certain operations, on ease of use or graphics positional accuracy, etc.

5. It should now be possible to choose a system(s) based on price and required performance. The normal rules that apply within an organization for supplies or equipment purchase may then be followed. Once the system has been implemented it might still be necessary to undertake certain tasks in order to maximize GIS usefulness, e.g.

   1. Customizing the output to obtain a desired “look”.
   2. Creating documentation which is specific to the user's GIS.
   3. Training personnel to the demands of your system.
   4. Looking for suitable, specific data.
   5. Carry out a post-implementation audit to see whether the original objectives for the GIS are being met. If not, why not?

6. Needs such as accommodation, site, workspace size, delivery dates, finance, insurance, personnel, security and communications potential and facilities should be worked out as an ongoing operation during the project evaluation and implementation stages. Other operational considerations which should also have been considered include planning for the ongoing management of the system, maintenance of the data bases, suitable storage conditions for expensive tapes, provision of back-up tapes or disks, work scheduling arrangements and data acquisition procedures.

6.8 Guidance and Support for GIS

It is fair to say that the range of support, information and guidance in GIS was at a minimal level until the mid-1980s, although it was commensurate with the lower profile which GIS then had. With the technological-led boom in GIS, which had its seeds in earlier research and which was aided by the synthesis of scientific fields which were combining around the power of the processor, i.e. CAD, RS, computer graphics, spatial analysis, etc., then support for and from the “industry” took off rapidly. Confirmation of this is to be had in the introduction of new courses, the inception of journals and magazines, the increase in conferences and symposia, the inception of GIS learned societies, and so forth, all of which have occurred in the past three or four years. Unlike RS, where support has come from a fairly narrow range of interested groups and companies, interest in GIS is far wider. Here we can only hope to point the reader along some major lines of support - this should be sufficient given the amount of information which is now available.

6.8.1 Training in GIS

Undoubtedly there is a chronic shortage of people with the necessary skills in all aspects of GIS. This has arisen, not only because of the enormous upsurge of demand, but also because of the necessity of having such a wide knowledge base to be truly efficient in all the complexities of GIS. People having these skills have easily been able to secure promising careers, usually in the private sector, and consequently the areas of research and teaching have been very hard hit (Dept. of Environment, 1987; Green, 1987; Gittings, 1989).

In response to these needs a large number of short courses on GIS are now being offered, both by the software houses and university departments. These are typically of three to five days duration, expensive, and are aimed at either an "intensive introductory" level for management, or for introducing new additions or updates to software packages.

University departments or Higher Education Colleges are rapidly introducing GIS into their syllabuses, mainly as part of undergraduate (first degree) Geography or IT courses, or as a typically one year part-time course, or as a specialized one year Masters degree or Diploma. Some detailed curricula, setting out all the essential components or a GIS course at degree level, are now beginning to appear, e.g. Unwin and Dale (1990). Many of the GIS courses have evolved from sideways moves out of IT, RS, CAD or geography or from an integration between any of these fields.

Another kind of short course has been offered by FAO. It is a GIS and RS appreciation workshop with the objective to familiarize fisheries and aquaculture personnel from developing countries with applications and constraints of the two technologies. Workshops have been from 8 to 15 days and include field verification of RS and GIS results. The Global Environmental Monitoring System of the United Nations Environmental Programme (UNEP) also offers GIS courses for developing-country persons, but of a longer duration and aimed broadly at the environment.

One response to the lack of teaching staff in GIS has been the computerized tutor. There are two recent developments:

1. Geographical Information Systems Tutor (GIST), developed in the U.K. at Birkbeck College (Raper and Green, 1989). This tutorial program is structured in a logical functional sequence, and uses illustrated “hypercards” to explain the fundamentals of GIS. Browsing is possible in any sequence, and GIST can either be accessed through a wide area network or it can be purchased.

2. Geographic Information Starter System (GISTARS), developed at Penn State University in the USA (Myers, 1990). This is a training package aimed at promoting interest in, and understanding of, GIS technology in developing countries. It runs on PCs having one or more disk drives and a CCGA card.

6.8.2 Published Information on GIS

This can be conveniently listed under three major categories:

A) Conferences and Symposia. These are organized by a variety of learned societies. They usually occur annually, where they might be associated with an exhibition or trade show, and they invariably produce conference proceedings. International ones include:

1. International AUTO-CARTO.
2. International Automated Mapping and Facilities Management.
3. International Geographic Information Systems.
4. International Geosciences and Remote Sensing Symposium.
5. International Symposium on Computer Assisted Drawing.
6. International Symposium on Spatial Data Handling.
7. Private companies, or some governmental organizations in the USA, are now holding annual conferences/symposia, e.g. GRASS, ESRI, ERDAS, GIMMS, etc., which are “marked” internationally.

B) Journals and Trade Magazines. Few of these are totally concerned with GIS, though all of the others have a rapidly increasing GIS component, and some have issued special GIS editions. Table 6.7 lists the main ones. The trade magazines, though sometimes showing company bias, are excellent sources of information, typically offering articles plus “Trade Directories” under such headings as Training, Software Suppliers, Hardware Suppliers, Consultants, GIS Specialist Services, etc.

^* = Essentially trade publications (as are the books marked below)

C) Books. Very few books have been devoted solely to GIS - the list below includes all those currently available.

1. ASPRS (1987).
2. Aronoff (1990).
3. Burrough (1986).
4. Dept. of Environment (1987).
5. Foster and Shand (eds) (1990).*
6. GIS World Inc. (1989).*
7. GIS World Inc. (1990).*
8. Masser and Blakemore (Eds) (1990).
9. Mutunayagam and Bahrami (1987).
10. Peuquet and Marble (Eds) (1990).
11. Rhind and Mounsey (1991).
12. Shand and Moore (Eds) (1989).*
13. Star and Estes (1990).
14. Tomlin (1990).

6.8.3 Other Sources of GIS Guidance

These are necessarily rather a mixture, and make no claims to being comprehensive. They include:

1. Private companies. Notably the software houses publish extremely useful newsletters, typically two to four times per year. These are an essential way of not only promoting themselves, but also of detailing the rapid developments which are occurring. They nearly always offer back-up services and guidance on software problems.

2. Cartographic and Digital Archives. These are being compiled in some countries, but because formidable problems are faced regarding updating, legal ownership of data, storage responsibility and location, access rights and costs, etc., then progress in building up digital map archives is slow (Finch and Rhind, 1987). UNEP now offers on-line access to a catalogue of environmental digital data sets including details of the responsible agencies. Further details can usually be had from national mapping agencies or national libraries, and the FAO retains an archive of their digitized resources.

3. Consultants. Some consultants specializing in GIS are beginning to emerge, and their availability is best sought through those previously mentioned trade magazines or books, or through specialist consultancy directories.

4. Videos. A number of videos have recently been produced which look at GIS from varying perspectives. Recent ones are reviewed in Hall and MacLennan (1990).

5. Computer On-line Facilities. If these are accessible, then a wide range of bibliographic data can be acquired. This may be a very rapid way of acquiring details on published information, though the full texts may not be so easy to obtain.

6. National GIS Organizations. Some of these have been formed in the last two or three years with a view to consolidating and coordinating progress in GIS. They include:

1. National Centre for Geographic Information and Analysis. (U.S.A.)
2. The Association of Geographic Information. (U.K.)
3. Canadian Centre for Geographic Information Systems in Education.
4. National Centre of Expertise for Geographic Data-Processing. (Netherlands).
5. Centre for Spatial Information Systems. (Australia).

6.9 Future Trends in GIS

“We are convinced that a fundamental change towards the routine use of geographic data on computers has already begun. The consequences of this cannot be predicted except in general terms and by analogy. The introduction of computer accounting systems 25 years ago was hesitant at first and many mistakes were made. However, costs fell, the value of new forms of information from the systems was discovered and, in due course, these systems became commonplace in quite small organizations. A return to manual systems is now inconceivable. We believe the use of Geographic Information Systems will develop in a similar fashion.” (Dept. of Environment, 1987). Despite the fact that GIS faces huge research programmes and countless problems, we feel it right to preface the conclusion of our analysis of GIS in an optimistic way.

In a field which is moving as fast as GIS, it is imperative to review future trends, if only to see the kind of scenario which we may be facing in as little as half a decade. Some of the trends we mention will have been previously hinted at, especially in section 6.6.2 on problems in GIS, so we have not elaborated on these. The reader must be aware that research advances are also being made in many related disciplines, some of which will have a profound effect on GIS - especially those developments occurring in basic computer technology. Useful sources on future trends in GIS include: Clarke (1986), Wilkinson and Fisher (1987), Croswell and Clark (1988), Rhind (1988), Tomlinson (1989) and Ventura (1989) and Maguire (1990). From the hundreds of possible trends we have selected 20 of the most important:

1. The ability to handle both raster and vector data on the same machine, at the same time and to inter-relate them. It is inevitable that intensive research will be aimed at gaining the many advantages to be had from achieving this integration (Wilkinson and Fisher, 1987).

2. The development of “intelligent” or “expert” knowledge-based systems. These are based on formal sets of rules which may be modified and systems which are capable of learning from experience. Basically, this means that a great deal of time, or digital storage space, could be saved if during the encoding, manipulation or retrieval stages of GIS functioning, the program was able to use logical rules (knowledge) in specific situations to short-cut, improve or restructure its procedures, e.g. searches were not carried out for areas outside certain prescribed domains (Rhind, 1988; Molenaar, 1989), or in digitizing there might be automated recognition of certain features on maps by colour or symbols. Expert systems will also aid users in making decisions over which GIS functions to use under particular circumstances (Butler, 1988).

3. The display of maps, or other graphical data, will be made more effective to the final users. There has been insufficient constructive research into the whole area of map perception (visualization), even though it is clear that map presentation techniques are sometimes an obstacle to understanding what the GIS is trying to convey. Wilkinson and Fisher (1987) claim that the future acceptance of GIS may depend just as much on the quality of the visual product as on their capabilities for manipulating spatial data.

4. How to deal with the “fuzziness” of data. Fuzzy data derives from it being ill-defined or having limited accuracy. Research is in progress to find out when to use which kind of measurement to assess data in accuracies, and Butler (1988) describes other work aimed at providing a rigorous basis for expressing "fuzziness" concepts.

5. How do GIS users use the available data? It is clear that if two analysts were requested to separately seek an answer to the same specific questions using any available objective data, they would be likely to arrive at different solutions. Why is this? It must be related to the sorts of questions asked in seeking a solution (selection criteria), to perceptions of levels of importance, to imperfect knowledge, etc. We need to be precise about what operations should be performed on what data, which data is the most important under specified conditions, which mathematical functions can be legitimately performed and perhaps to revise our data gathering priorities. Tomlinson (1989) clearly outlines research priorities in this area.

6. The need for a “Geographic Information Theory”. By establishing such a theory, Molenaar (1989) claims that the GIS user would be better able to understand its possibilities, to structure the data processing sequences, to formulate criteria for measuring the performance of a system and to design a GIS from scratch.

7. The incorporation of “desk-top publishing” into GIS. This means that software packages (or programs) would be integrated into the GIS in order to greatly improve, and to control, the quality of the output, e.g. integration of maps and text, layout and appearance of maps, variations of fonts, improved colouring or shading, etc. (Schokker, 1989).

8. Developments in optical disk storage systems. They will have a large impact on GIS because of the ability to store huge volumes of information in a small, inexpensive format. Storage densities are between 10 and 50 times greater than traditional magnetic storage devices. Until recently the technology was used mostly for “write once read many” (WORM) processes since data could not be erased. This is still very useful for data archiving where huge data volumes must be stored and retrieved, but rarely changed, e.g. map outlines. However, erasable optical disk units are just becoming available (Crosswell and Clark, 1988) and these are likely to become the norm for large-scale storage by the mid-1990s.

9. Move towards a “World Referencing” or “Global Positioning” system. Whilst this seems almost essential for organizations such as FAO, which operate at the world scale, it will also greatly ease the data transference problems in other fields such as ocean monitoring, meteorology, international fisheries, pollution evaluation and control - in fact any field which breaches international borders. Global positioning satellites will help make referencing easier and hopefully a Global Positioning system will be in place during the next decade.

10. The development of high-speed telecommunications facilities. This will aid the development of efficient networking at local and national scales, allowing access to data bases over a wide distance which will be vital to many GISs. This development will require great advances, not only in network design and implementation, but also in the standardization of GIS data formats and structures.

11. The development of wholly different and more efficient data base structures. This will inevitably involve the implementation of a knowledge-based DBMS which incorporates intelligent query languages to help the GIS user rapidly retrieve required items of data. The whole area of DBMSs, and data base storage facilities, will become increasingly visible as the number of data bases and their size dramatically increases. Tomlinson (1989) maintains that we need to develop geographic data models which will allow investigations of the complex linkages which occur in any spatial area, in order to improve the ways in which data bases are designed and structured. In the meantime, Goodchild (1988) indicates that work on both hierarchical and relational data structures or models continues to bring new ideas.

12. Improvements in both automatic and semi-automatic data capture by raster scanning and line following techniques. More sensitive scanners will allow greater discrimination between data and “noise”. It is likely that paper maps of the future will incorporate special inks or bar codes to make feature recognition easier for automatic data capture equipment. There will eventually be transaction-activated data capture, where the processes of change themselves generate updated data bases.

13. There will be renewed attention to the updating of algorithms. Many existing ones were devised with limited objectives, levels of functionality and “computational elegance”.

14. Tomlinson (1989) asserts that there is not yet much potential for existing GIS software to carry out quantitative spatial analyses, i.e. considering that many of the methods are very useful, they are simple and they have been around for up to 30 years. Examples include factor analysis, modelling network changes and location optimizing routines.

15. Clearer definitions in legal ownership and copyright laws relating to the use and transference of data, and to the legal liability of the producer of data when users act upon advice and something goes wrong.

16. Making graphic user-interfaces more friendly. Much GIS software is still too difficult for novices to use, but in the near future users will be able to “drive” systems by using concepts and terms which are familiar to them.

17. The formulation of GIS standards to facilitate information exchange. Proposals for minimum and uniform standards have been laid down in some countries (NCDCDS, 1987), but world-wide a great deal needs to be done. Standards need to apply to both the quality of the data and the way it is structured, so that levels of expected accuracy can be realistically proposed.

18. Much work urgently needs to continue on error propagation - on the ways in which uncertainties in the values of control parameters and input data affect results.

19. How the GIS user can be made aware of the digitized data which exists. There is much work to be done in creating directories, archives, on-line facilities, etc., which catalogue clearly the availability, location, cost and rights to existing digitized data.

20. A continuation of the massive task of digitizing existing topographic, and some thematic, maps. This will also encompass data improvements and up-dating and will perpetuate discussions on “which levels of government will provide which type of commonly used data in which digital format?” (Tomlinson, 1989. p.292).