Computers & Geosc~ences Vol. 11, No. 3, pp. 291-295, 1985 0098-3004/85 $3.00 + .00 Printed in the U.S.A. Pergamon Press Ltd. LIMITATIONS OF MICROCOMPUTERS IN THEMATIC MAPPING V. GARDINER and D. J. UNWIN Department of Geography, University of Leicester, Leicester, LE1 7RH, England (Received 12 June 1984) Abstract--Hardware, software, and data requirements for thematic mapping are identified. Existing microcomputer-based systems are evaluated in terms of these requirements, with particular reference made to first-generation 8-bit machines. For some applications today's systems are adequate, but for many purposes they are not. Key Words: Computer mapping, Microcomputers. INTRODUCTION In 1977 D. R. F. Taylor suggested a four-stage model of the development of automated cartography (Fig. 1), which can be regarded as being of general appli- cation. Questions arising from this which are as yet unresolved are whether a fifth stage can be recognized, representing the impact of the microchip in general and the microcomputer in particular, and, if the answer to this is positive, what form should this extension take? The aim of this paper is to examine and evaluate currently available microcomputer-based systems for thematic mapping in relation to fairly modest suggested desiderata for these systems. Atten- tion will be directed first toward requirements and desiderata, hardware, software, and data aspects of the systems being considered in turn. Then, existing microcomputer-based systems will be evaluated in the light of the desiderata identified. HARDWARE REQUIREMENTS Automated cartography places considerable de- mands upon the availability of hardware. In most applications, it is necessary to process a lot of infor- mation describing both the spatial locations of entities, such as points, lines, or areas, and the attributes located at, located in, or associated with each of these entities. Thus, a line's spatial position may be repre- sented by a string of digitizer coordinates and its nature described by feature codes indicative of, for example, parish boundary, county boundary, and so on. Therefore, for processing anything but trivial data sets, it is necessary to have available a lot of central processor addressable memory for immediate processing, as well as much secondary storage for longer term recall. The production of maps acceptable to users implies that there is available a range of good-quality display hardware, either as peripherals or as an integral part of the system core, as well as the facility to switch readily from one device to another. Rhind (1977) suggests a four-fold typology of display devices, as in Table 1. It is perhaps not unreasonable to expect realistic systems at least to have the capability of interfacing to devices representing each of these major types, as well as having some capability for interactive editing. Finally, it must be acknowledged that many the- matic mapping operations involve processes which humans can do well and quickly by manual methods. These include such tasks as point-in-polygon deter- mination, threading isolines through randomly or regularly distributed point-data values, and positioning text to satisfy aesthetic and graphic criteria. These types of tasks require long and complex algorithms to be encoded if they are to be effective and of genuine use in an automated system. Processors used in microcomputer-based systems for thematic map- ping, therefore, must be rapid. The hardware requirement, therefore, may be sum- marized as a need for fast machines with much memory, interfaced to specialized peripherals in a conjunction which until recently was only to be observed in specialist centers undertaking research into computer-aided cartography and computer graphics. soFrWARE REQUIREMENTS Programming a digital computer to draw maps is neither simple nor rapid, so that, if at all possible, most users will use packaged "ready to run" programs, standard subroutine libraries, and standard operating systems. It is necessary for software to be robust and reliable, especially if it is to be used by those without programming experience. It should be able to handle all types of permitted data, and should not fail in operation, no matter how absurd are the parameters specified, for example, by a trainee user who has misunderstood the requirement. Programs should be well documented, either by separate manuals and specifications or by integral on-screen facilities. They should be efficient in operation, especially if large data sets are to be handled, and the code in which 291