Geoinformatics is one of a number of terms being used today to describe activities in the general area of computer handling of digital geographic information. Other terms include geomatics, popular in Canada in particular because of its similar French equivalent; and geographic information science, a term of growing popularity in the U.S. The term is very broad, and includes at least three distinct areas:
First, the geographic information technologies include four areas of technology of particular significance for geographic information.
Second, geoinformatics includes concern for the supply of geographic data. Traditionally, geographic data has been supplied in the form of maps, atlases, and images by a variety of agencies, many of them at the national level. The most important type of geographic data is described as framework, because it exists to allow other activities to be tied to locations on the Earth's surface. For example, if we know that a certain observation was taken at the outlet of the Salmon River into Lake Ontario, and have access to a topographic map, then the location of the observation in latitude and longitude can be determined by recognizing these features on the map, and reading off the coordinates. The framework includes geodetic control, the network of high-accuracy monuments maintained by the National Geodetic Survey, and associated components including mathematical approximations to the shape of the Earth; topographic data providing information on the shape and locations of prominent features such as rivers and lakes; imagery from the air or space; elevation data giving information on the height of the terrain; and the gazetteer, an index listing named features and their approximate locations. Today, all of these framework data sets are considered essential parts of the National Spatial Data Infrastructure (NSDI), a term used increasingly to describe the national framework (NRC, 1993).
Third, geoinformatics includes those sciences concerned with research and development associated with geographic information technologies and geographic data. Geographic information science is concerned with the fundamental issues that arise in the digital handling of geographic data, and today overlaps increasingly with the traditional fields of cartography, remote sensing, photogrammetry, geodesy, surveying, and geography. It considers mapping methods, particularly when these are enhanced by digital technology; issues of sampling in space; problems of representing geographic data in digital form; problems of scale and resolution; and the increasingly important areas of accuracy and uncertainty.
Each of these three areas—technology, data, and science—presents numerous challenges at this time. In technology, there is constant interest in improvements to systems, through new and more efficient data structures and algorithms, and new designs for user interfaces. New sensors are being developed for more powerful remote sensing, and new developments are proposed in GPS. In the area of data, we are seeing major developments aimed at improving the organization and dissemination of geographic data, in order to promote data sharing, reduce the possibility of duplication, and obtain greater returns from investment. Recent developments in the specification of metadata, or data about data, have produced great improvement in our ability to locate data, particularly on the Internet, and to evaluate the fitness of a given data set for a given use. The search engines of the Internet now allow us to check millions of servers in search of data, though they do not yet provide effective means for detecting and cataloging geographic data, but instead are still largely based on the recognition of words of text. The widespread adoption of geographic information technologies has allowed virtually anyone to become a collector and publisher of geographic data, calling into question traditional supply arrangements which have emphasized the role of the federal government and central production. Finally, there is increasing interest in applying the principles of NSDI to the globe; unfortunately, at this time the supply and open exchange of global geographic data is confounded by many factors, including concerns for national security. Priorities in science range across a wide spectrum, as we recognize the important role that many disciplines can play in understanding and researching issues associated with geographic information. A recent summary of research issues is provided by the research agenda developed by the University Consortium for Geographic Information Science (UCGIS) in June 1996 (UCGIS is a consortium of 47 major U.S. research institutions). The agenda, which was developed at the UCGIS Annual Assembly in Columbus, is available in detail at www.ucgis.org, and is summarized in a recent paper (UCGIS, 1996). The ten topics are, in no particular order:
There are many possible 'hot topics' in this ten-point UCGIS list; in this last section I would like to discuss three current personal favorites. There is a large literature in each area which I will not attempt to review; interested readers are encouraged to examine the UCGIS agenda (UCGIS, 1996; and www.ucgis.org) or recent issues of the major journals of the field (for example, International Journal of Geographical Information Science; Geoinformatica; Transactions in GIS; Cartography and GIS).
One reliable way to explain a new technology is by reference to the concepts associated with an older, more familiar one. When the automobile was invented, it was described as a 'horseless carriage'—similarly, today's GIS is readily described as a computer containing maps. People are familiar with the notion that a computer can be used to store and manipulate words, numbers, and images, so the idea of storing and manipulating the information found on maps is an easy extension. But while familiar technologies can provide landmarks in the unfamiliar world of new technologies, the use of such analogies ultimately works against creative understanding of the new technologies' true potential. As long as an automobile was conceptualized as a horseless carriage, it was difficult to see the full impact it would have, as later entirely new concepts were developed that had no equivalents in the older world. The use of the map metaphor to explain GIS constrains thinking in many ways. Paper maps are of necessity: flat, requiring the world to be projected onto a flat sheet—in a digital world there is no reason to flatten the world; two- dimensional, from the nature of the physical medium—there is no equivalent constraint in the digital world; static, since a map cannot be changed once printed—but digital databases can be changed easily; crisp, since maps had to be drawn with pens of fixed width—but concepts in the digital world can be fuzzy, such as the boundaries between different soil types; at a uniform level of detail—but in the digital world scale and spatial resolution can be varied within a single database; at a single level of generalization—but digital databases can be built with multiple levels of generalization linked logically into a coherent whole. After many years of development, geographic databases are only now beginning to employ new methods of representation that address some of these constraints, and move GIS design beyond the simple metaphor of the paper map.
Miniaturization and improved performance have allowed computers to evolve from massive mainframes to small desktop workstations. Recently, further technical developments have led to the laptop and palmtop, offering limited functionality in smaller and smaller formats. But GIS remains largely a technology of the office desk. Only in certain areas is there significant exploitation of the potential for miniaturization and wireless communication to allow GIS to operate truly in the field. Precision agriculture is one such area, where GIS technology is increasingly found in vehicles (tractors and harvesters) where it is used to monitor production, and control planting and the application of fertilizers and chemicals. Field GIS is also found in facilities management, and in resource management. But these applications scarcely scratch the surface. In principle, GIS is about description and modeling of forms and processes on the surface of the Earth, and its ideal environment is surely the field, where there can be direct observation of the phenomena GIS is designed to represent. A field GIS in the hands of a soil scientist could help design sampling schemes; provide a base map for improved accuracy of location; accept input from the scientist in the most convenient form; and carry out many of the operations that today have to be relegated to non- specialists in the office.
The final priority is uncertainty, a term increasingly applied to the observed difference between the real world and a digital representation of it. Uncertainty includes error, due to inaccurate measurement of variables such as location or elevation; the effects of generalization, when detail is lost; problems due to the use of an inadequate data model, which can constrain the ability to describe important elements of the real world; and problems of inadequate definition. Uncertainty can change when the data pass from one person to another, if the meaning of the digital representation is not communicated fully. It can also change as the data are manipulated or transformed. At this time, we have a very inadequate understanding of uncertainty in geographic data, and the effects it can have on the results of GIS analysis. Frequently, the actual impacts of uncertainty are demonstrably far higher than users believe, since there is a tendency to ascribe much greater significance to the results of computer processing than are actually justified. Research in this area is investigating such issues of propagation. Another area of great interest at this time is the visualization of uncertainty, since traditional cartographic methods of visualization of geographic data pay very little attention to uncertainty. There is also interest in fuzzy logic, as a means of dealing particularly with uncertainty of definition in areas such as land cover mapping.
I have attempted in this short discussion to introduce the three major areas of activity in geoinformatics at this time: technology, data, and science. I hope I have demonstrated that there are issues of major significance in all three, and abundant opportunities for collaborative research. The emphasis has been on the field itself, rather than on any particular application, such as Earth system science, on the assumption that contributions to the field are also likely to be valuable to the field's applications. But for a more specific analysis of the importance of geoinformatics to Earth system science, and a closer analysis of the issues in this application area, the reader is referred to the following: