Chapter 2 Background and History Geographic information systems evolved as a means of assembling and analyzing diverse spatial data. Many systems have been developed, for land-use planning and natural-resource management at the urban, regional, state, and national levels of government agencies. Most systems rely on data from existing maps, or on data that can be mapped readily (Shelton and Estes, 1979). The development of geographic information systems has its roots in at least two overlapping areas: an interest in managing the urban environment (particularly in terms of planning and renewal), and a concern for the balancing competing uses of environmental resources. Technology has played a critical role in addressing these concerns. If we look at John Naisbitt's 1984 work Megatrends, we can see why. Megatrends discusses new directions which are transforming our lives. In Naisbitt's words, none of the megatrends discussed “is more subtle, yet more explosive than . . . the megashift from an industrial to an information society.” This information society had its beginnings in 1956 and 1957. Indeed, the advances in communications and computer technology that facilitated the widespread dissemination of the ideas and concepts contained in Rachel Carson's book Silent Spring also provided the foundations and requirements that necessitated the construction of automated geographic information systems. Today, environmental scientists and resource managers have access to more data than ever. Naisbitt (1984) estimates that scientific information is doubling every five years. The key to coping with this information explosion is the employment of systems - systems that will take the data, analyze it, store it, and then present it in forms that are useful. These are the requirements of an information system. 2.1 The Cartographic Process According to Robinson and Sale (1969), cartography is often described as a meeting place of science and art. This science/art is fundamentally directed at communicating information to a user and is central to an understanding of the strengths and weaknesses of geographic information systems technology. Much of the material contained in this book is directly related to essential elements of the cartographic process, which involves a body of theory and practice that is common to all maps. Maps are both a very important form of input to a geographic information system, as well as common means to portray the results of an analysis from a GIS. Like a GIS, maps are concerned with two fundamental aspects of reality: locations, and attributes at locations. Location represents the position of a point in two-dimensional space. Attributes at a location are some measure of a qualitative or quantitative characteristic, such as land cover, ownership, or precipitation. From these fundamental properties a variety of topologic and metric properties of relationships may be identified, including distance, direction, connectivity and proximity. As Robinson et al. (1984) observe, “a map is therefore a very powerful tool”. Indeed, maps are powerful tools for communicating spatial relationships. Following Robinson et al., maps: are typically reductions which are smaller than the areas they portray. As such, each map must have a defined relationship between what exists in the area being represented, and the mapped representation. This relationship is of primary importance. Scale sets limits on both the type and manner of information that can be portrayed on a map. involve transformations. Often in mapping, we are faced with a need to transform a surface which is not flat (such as a portion of the earth's surface). In order to represent such a surface on a flat plane, map projections are employed (see section 6.6.l). Choice of a particular projection has an impact on how a given map may be used. Plane coordinate grids are often used on maps as systems of reference. are abstractions of reality. Maps are the cartographer's representation of an area, and as such, display the data that the cartographer has selected for a specific use. Thus, the information portrayed on a map has been classified and simplified to improve the user's ability to work with the map. contain symbols which represent elements of reality. Few map symbols have universally accepted meanings, but some maps use a standardized set of symbols. portray data using a variety of marks, including lines, dots, tones, colors, textures, and patterns. In addition to these basic characteristics of maps, the user of maps and other products of a geographic information system should understand the errors which may affect them. The sources of errors fall into three categories (Burrough, 1986): obvious sources, those resulting from natural variation and original measurement, and those arising through processing. Obvious sources: The source data may be too old to be of value. The areal coverage of a given data type, within a given time frame, may not be complete. The scale of the map may restrict the type, quantity, and quality of the data which may be presented. The number of observations within the target area may not be sufficient to be able to determine the spatial patterns in the objects of interest. Practical matters such as the time, funds, and staff which are available may not permit us to produce a product of the required characteristics. Natural variation and the original measurements: Positional accuracy of the source data may not be sufficient, due to problems in the field data itself, instrument errors, and lack of rigor in the compilation process. Attribute errors also may come from a variety of sources, including both mis-identification and compilation problems. Processing: Numerical errors may include round-off or dynamic-range errors in arithmetic computations. Errors in logic may cause us to manipulate the data incorrectly, thus leading us to fool ourselves. Common problems in this area are associated with classification and generalization. The above lists focus on errors in a single map sheet, or a single GIS data layer. When working with many layers at once, the separate layers may not be completely compatible in terms of scale and accuracy, thus complicating the task of creating an accurate, final, analytic product. This, as well as other problems, such as developing efficient and cost-effective means for verifying the accuracy of map and GIS products, are active research areas. 2.2 Early History Cartography is defined in the Multilingual Dictionary of Technical Terms in Cartography (Meynen, 1973) as “the art, science and technology of making maps together with their study as scientific documents and work of art.” Map-making per se can be traced back to the ancient cultures of Mesopotamia and Egypt. The earliest known map, a regional map imprinted in a clay tablet, dates from about 2500 B.C. Yet people must have been making maps much earlier than that. Simple arrangements of sticks or pebbles were probably used to illustrate geographic relationships long before clay tablets or papyrus came into vogue. A mound of dirt, a few pebbles, and a small furrow made with a stick, could have illustrated important game trails or berry-picking locations, and could thus have been the first analog geographic information system. Parent and Church (1988) state that the origins of more sophisticated geographic information systems go back to early developments in cartography. They reference the mid-eighteenth-century production of the first accurate base maps as an important point in GIS development. As Parent and Church point out, until the development of high-quality base maps, the accurate graphic depiction of spatial attributes was not possible. These developments were followed by a rapid expansion of the use of thematic mapping. The idea of recording various layers of spatial data on a series of similar base maps was an established cartographic convention by the time of the American Revolutionary War (Harley et al., 1978). For example, a map by French military leader and cartographer Louis Alexandre Berthier (1753-1815) contained hinged overlays showing troop movements during the 1781 Siege of Yorktown (Rice and Browns, 1974). In the early part of the nineteenth century, advances in both the physical and social sciences provided geographers with important intellectual tools for the analysis of spatial data. Such fields as statistical analysis, number theory, and advanced mathematics flourished. The first geologic maps of London and Paris appeared. The work of the distinguished German geographer Alexander Freiherr von Humbolt (1769-1859) became influential. The British census of 1825 produced a tremendous amount of data to be analyzed, and the science of demography soon evolved. Church and Parent (1988) state that by 1835, technology (in particular, advanced cartographic techniques), social science, and social thought (specifically, concepts of environmental responsibility) had progressed to support new and improved levels of thematic mapping. However, it was the economic changes of the industrial revolution, according to Church and Parent, that provided the main catalyst for the early evolution of geographic information systems in this time period. The explosion in manufacturing, with the attendant demand for raw materials and labor, created the need for a new, extensive infrastructure, both social and industrial. Indicative of all these changes is a transportation study, completed in 1837, that first brought together technical, social, and scientific advances related to spatial data analysis, into a coherent whole. The Atlas to Accompany the Second Report of the Irish Railway Commissioners, appearing in 1838, consisted of a series of maps with a uniform base, depicting population, traffic flow, geology, and topography. As this example indicates, cartographers realized some 150 years ago that a single map may not contain all the data required to satisfy a given information need. Indeed, the data may not exist in map form at all, but in graphs, text, or statistical tables. As researchers and resource managers began to ask more and more complex questions about their environment, their need for improved methods of processing spatially distributed data increased. This need lead to the beginnings of automated geographic data processing in the late 1800's. Streich (1986) states that American statistician Herman Hollerith (1860-1929) was the father of automated geoprocessing. Hollerith adapted punched-card techniques, which had been used in France to program looms, to help process the information collected in the 1890 United States Census. Hollerith conceived the idea of punching raw demographic data onto cards and using machines to sort and collate this data. Streich goes on to state that". . . the move to 'electro-mechanical' data-processing technology for census tabulation characterizes a fundamental tenet of geoprocessing -- that a need exists to rapidly, accurately, and cost-effectively collect, analyze, and distribute spatially disposed information." In 1936, Charles Colby's presidential address to the Association of American Geographers (Colby, 1936) laid out research challenges in geography. Among these challenges particular emphasis was placed on the development of quantitative approaches to map-based problems. In doing so,Colby set the stage for the modern era of geographic information systems. 2.3 The Modern Era From these early beginnings, advances in computing, cartography, and photogrammetry laid the technological foundations for the automated geographic information systems that began to appear in the 1960s. The conceptual framework within which early geographic information systems were implemented involved individuals in many disciplines. Researchers and resource managers in diverse areas realized that there was a need for integrating data from a variety of sources, to manipulate the sets of data to analyze them, and then to be able to provide information for a resource planning and management decision process. Three important factors helped lead to the creation of digital geographic information systems in the 1960's: refinements in cartographic technique, rapid developments in digital computer systems, and the quantitative revolution in spatial analysis. These developments were very important in that they helped to provide analytic tools as well as stimulation to researchers and professionals in a variety of applications. In addition to these, we must not forget the advances in geographic thought that helped to bring about the modem GIS. Chrisman (1988) points out that the “cult of novelty and high technology”can blind us as to various disciplines' contributions to GIS development. As a geographer, he points to Sauer's early work in the upper peninsula of Michigan, as well as to the land-use research of Whillesey, Finch, and others that foreshadowed our current paradigm of stacking layers of thematic information. The roots of such map overlay may be traced back at least 100 years in the field of landscape architecture (Steinitz et al., 1976). In 1969, Ian McHarg's Design with Nature was published. This work formalized the concept of land suitability/capability analysis (SCA). SCA is a technique in which data concerning land use in a locale being studied is entered into an analog or digital GIS. SCA programs are used to combine and compare data types via a deterministic model, in order to produce a general plan map. If the model is carefully implemented, and suitable data is available, this map should be consistent with existing land-use classes and the constraints that are imposed by both natural and cultural features. Design with Nature is a seminal work, influencing the use of overlays of spatially referenced data layers in the resource planning and management decision process. McHarg's efforts in SCA have been followed by many articles in this area. Resource management concerns spurred development of spatial data-processing systems in the U.S. government during the past two decades. A system called STORET was developed by the Public Health Service for storage of spatial information about water quality (Green, 1964). Another, called MIADS, was developed by the U.S. Forest Service for the analysis of recreation alternatives and hydrology (Amidon, 1964). The Census Bureau was also heavily involved in geocoding and automated spatial data processing at this time. In the university community at this time, Harvard's Laboratory for Computer Graphics developed and made available a series of automated mapping and analysis programs. The University of Washington at Seattle also made important contributions, particularly in the areas of transportation analysis and urban planning and renewal (Gaits, 1969). Urban planning applications blossomed with the development of these kinds of tools; by 1968 thirty-five urban and regional planning agencies in the United States were using automated systems (Systems Development Corp., 1968). The first system in the modern era to be generally acknowledged as a GIS was the Canada Geographic Information System or CGIS (Peuquet,1977). Roger Tomlinson (1982), involved in the design and development of the system, states that CGIS was designed specifically for the Agricultural Rehabilitation and Development Agency Program within the Canadian government. The main purpose of CGIS was to analyze Canadian Land Inventory data, which was being collected to find marginal lands. Therefore, in the broadest sense, the first geographic information system was developed to help with an environmental problem: rehabilitation and development of Canada's agricultural lands. The Canadian Geographic Information System was implemented in 1964 (Deuker, 1979). This was one year after the first conference on Urban Planning Information Systems and Programs, a conference which led to the establishment of the Urban and Regional Information Systems Association. The New York Landuse and Natural Resources Information System was implemented in 1967, and the Minnesota Land Management Information System in 1969. In these early years, the costs and technical difficulties of implementing a GIS prevented all but national-and state-government agencies from developing these systems. In 1977, a report issued by the United States Department of the Interior's Fish and Wildlife Service compares the selected operational capabilities of 54 GISs (USFWS, 1977). This survey, which is representative of several others conducted in the late 1970s, provides information on the hardware environment, programming language, documentation, and characteristics of the systems. This survey lists many GISs developed by federal and state agencies, as well as universities. However, it contained information on only a few commercial GISs. Even today, few commercial firms offer fully integrated geographic information systems. Streich (1986) estimates that there are ten commercial firms offering GISs on the open market. Why is it that there are so few commercial firms offering geographic information systems, even today? There is no simple answer. A GIS is a complex hardware and software system, and requires considerable expertise in a variety of geographic, computer science and systems engineering areas. Nevertheless, some GISs are being developed in the private sector. These commercially available systems provide a wide range of well-integrated GIS capabilities. As technology and scientific understanding improve, the development of geoprocessing systems becomes more and more open to commercial firms. Instead of being a large-user in-house activity, the development of geoprocessing systems will likely be taken over by commercial firms and made available to a variety of hardware environments, discipline interests, and goals. In addition to the beginnings of commercial GIS development, the 1970s also saw significant developments in image processing and remote sensing systems, which often had some GIS functions. Such firms as the Environmental Systems Research Institute in California began operation. Image processing systems with some GIS elements were developed at the Jet Propulsion Laboratory and at the Purdue University laboratory for Applications of Remote Sensing. These latter systems incorporated GIS capabilities as the remote sensing community quickly realized that ancillary GIS data could play an important role in improving the accuracy of the interpretation of remotely sensed data. Developments in remote sensing technology and applications during this decade spurred practical and theoretical work in the areas of geometrical corrections and registration. The coupling of map and image data also drove work in raster-vector data format conversion (see Chapter 6). Today there are a number of both commercial and public-domain image processing systems that possess sophisticated GIS capabilities; however, we still believe that a great deal more work needs to be done in terms of effectively integrating remotely sensed data into traditionally vector-based geographic information systems. An interesting footnote to this brief history of geographic information systems comes from a conference held at the University of Calgary in 1982. The conference title was "Computer Assisted Cartography and Geographic Information Processing: Hope and Realism." In a session led by Roger Tomlinson, several individuals were critical of the operational status of the system. The criticisms revolved around the question of whether the system was meeting the needs of the users. Tomlinson responded that the system was originally designed for one specific class of users, who wished to analyze land inventory data and find marginal farms. This class of users had effectively disappeared, and there were now over 100 other users and a nine-month backlog of work. He concluded that this tremendous number of users and long backlog indicated that the system was successful. In response to Tomlinson's argument, it was noted that for many applications, a nine-month backlog was intolerable; when users cannot get their work done on the system, something is indeed not right. The message we derive from this story is that problems will certainly arise as geographic information systems designed and implemented for a specific class of problems are used for other purposes. Users and system managers must guard against inappropriate use of systems and must establish priorities and long-range upgrade and migration policies to meet the needs of changing user communities and changing data. In summary, the development of geographic information systems, in terms of both the underlying concepts and the technology, has drawn on the talent and experience of many researchers and investigators. It has grown out of concerns about the state of the physical and cultural environment, and it has been advanced by efforts in both the public and private sectors. Many early systems were developed to solve relatively narrow, specific kinds of problems. The past twenty years have seen an explosion in the technological base for these systems, particularly in the areas of data processing and remote sensing systems. The 1980s have seen continued growth in GIS applications, significant system refinements, and a modest expansion of the commercial availability and applicability of geographic information systems. While many operational systems may be limited in terms of the geographic area, the number of data types, and the modeling and analytic capabilities they can provide, they can perform many operations that only 25 years ago were considered unfeasible. One recent trend in the evolution of GIS technology is the inclusion of artificial intelligence into GIS design and operation. This topic was the subject of a workshop at a recent international symposium in Zurich, Switzerland (Smith, 1984). We will examine some of these far-reaching discussions in section 12.8. Let us now consider some of the generic components and functions of geographic information systems.