The use of maps in the exploration of geographic data
Thanks to technologic developments, more maps are now produced and used than ever before. Furthermore, no longer do users always require cartographers to produce the maps for them. They themselves can generate the maps they need, for instance, through the World Wide Web, and have the map displays appear on the monitor of their computer system. Cartographers are now supposed to incorporate their knowledge and expertise into the mapping software that is operated by the map users themselves. To better serve the needs of these map users, and to make the maps they produce as effective as possible, cartographers are also needed to carry out map use research.
This research deals with map use in exploratory cartography. Exploratory cartography is a cognitive process in which maps are used as tools for discovery, to solve the problem of gaining insight into unknown geographic relationships. In the past, users had to rely fully on the static maps prepared for them by others (e.g. in the form of a paper atlas). Nowadays, with the widespread availability of geographic data, computers and cartographic visualization software tools, users are able to generate and / or adjust, just for themselves, the map displays they require to reveal geographic information hitherto unknown to them. Therefore, over the last decade it has been accepted in our discipline that exploratory cartography involves a high physical interaction of the individual user with, and the manipulation of, the map and the underlying geographic data, leading to private and ephemeral map displays on the computer screen. Cartography now indeed has the potential to be fully "demand-driven" instead of "supply-driven".
Map use research has been executed since the 1950s. Two complementary types of map use research can be distinguished: more holistic, functional map use research, and perceptual and cognitive research. The results have been used to improve the functioning of maps. But the research, even the occasional usability research in exploratory cartography, has often been limited to the functioning of an individual, static or interactive map display on its own. What has been lacking is research in which not only the functioning of a given map is investigated, but also the complete process of finding, selecting, retrieving, adjusting or generating map displays in, for instance, an exploratory cartography environment. In other words, research into the cognitive processes that precede the actual use of a map display has been missing. This PhD research project may be considered as one of the first attempts to obtain more information about these "pre-map-use" stages in exploratory cartography.
The objective of this research was to investigate the selection and use of maps in the process of exploring geographic data, leading to hypotheses that could be tested in further map use research. The ultimate aim is to use the research results for improving the tools for exploratory cartographic visualization, i.e. both the cartographic products and the hard- and software used to make such products available to their users.
A case study was defined with a concrete objective and involving a concrete group of users - users who could, in principle, make use of a great variety of map types in visualizing a broad spectrum of geographic data. It encompassed a case in regional exploratory studies, one of the first stages in geographic research in which geographers try to gain insight into the geography of a region unknown to them.
Hypotheses about the geographic questions posed in regional exploratory studies.
Such geographers - professional map users with different levels of expertise in regional exploratory studies - were chosen as test persons in this research project. The region selected for the case study was the province of Overijssel in the Netherlands. The test persons were asked to construct a schematic graphic model of the geography of this region, derived from the concept of chorèmes and based on the use of maps that they selected or generated themselves, with the help of a research assistant. For this, the test persons could make use of a database of Overijssel, containing digital geographic data prepared for visualization within the ArcGIS software environment, as well as numerous existing paper and digital maps. To make these items of a data-rich environment quickly and easily accessible, a metadatabase was also created. The goal was to create an environment in which the test persons could operate freely, without being "steered" by the available possibilities, i.e. a demand-driven instead of a supply-driven cartographic environment.
In order to investigate the cognitive processes of exploratory cartography, use was made of a combination of qualitative research techniques centred round the think aloud method, a method hardly used so far in the field of cartography. The test persons were asked to think aloud during the execution of their task in a specially equipped experiment room containing a unique combination of hard- and software. The thinking aloud was recorded on videotape, together with the images of the test persons interacting with the maps, the changes on the PC monitor screen on which new maps were generated from the data in the database (or existing maps adjusted), as well as the changes on another PC monitor screen on which the test persons constructed their graphic models.
Above: Impression of the research laboratory especially equipped for this project: the test subject (left) produces a graphical model on the PC to the left. The research assistant (right) is helping the test subject to retrieve existing maps form the Overijssel database and to construct new maps from geographical attribute data with ArcGIS. All actions of the test subject are recorded with the video-camera at the back and his thinking aloud is recorded with a wireless microphone.
Above: Synchronization of the sounds of the thinking aloud, the video image of the test subject and the changes on the monitor screens of the two PC's is obtained by recording all these inputs simultaneously with a SVHS video-recorder (bottom right) on one single video tape. The images can be combined in this way with the help of a Video Quad Unit (bottom left).
These combined and synchronized video recordings were analysed and turned into verbal and action protocols. Parts of the video recordings were also discussed with the test persons in retrospect, in order to retrieve information about the cognitive processes not completely revealed by the thinking aloud. A questionnaire was used to obtain information about the characteristics of the test persons.
The verbal and action protocols resulting from the think aloud sessions were analysed on the basis of a hypothetic model of geographic problem-solving derived from the analysis of the task in hand and from cartographic and (regional) geographic theory. The building stones for this model were obtained by making a distinction between the overall objective of the geospatial data exploration, the various map use tasks executed to meet that overall objective, and the map use activities undertaken during the execution of these tasks. The map use tasks are related to a hypothetic set of geographic questions with an increasing level of complexity. Other aspects of the required modelling were the categorization of geographic themes and map types in relation to the basic geographic questions to which these map types are supposed to provide an answer. The modelling culminated in the combination of these building stones in a map use matrix and a map selection matrix. The combination of a map selection matrix and a map use matrix could be considered as a reflection of one segment of the process of geographic problem-solving. In such a segment one map is selected or generated and then used, whereas in the process of regional exploratory studies a number of maps will be used. On the other hand, the map selection and map use matrices should not only be seen as parts of the model of geographic problem-solving: they were also used as coding tools in the analysis of the recordings of the think aloud sessions.
Indeed, completed map use and map selection matrices were one of the outcomes of the user tests. The other outcomes were completed questionnaires, graphic models produced, verbal and action protocols, and shorthand overviews of the geographic questions posed, geographic themes selected, and maps selected or generated. Analysis of these outcomes could lead to a wealth of information from which many hypotheses could be derived for testing in further research - even hypotheses beyond the scope of the objectives of this particular research project. Within the framework of this present research, however, the main results, to be confirmed in further research, can be summarized as follows.
Exploratory cartography is not as interactive, private or demand-driven as assumed, at least not when it comes to regional exploratory studies. Here, the use of maps is very much of a supply-driven nature. Users - experts and novices alike - prefer to start their geographic data exploration with ready-made maps that are immediately available. They even prefer maps on paper and possibly combined into an atlas. Such maps are not selected on the basis of clear geographic questions. If users have a demand at all, and do not just browse through what is immediately available to them, the geographic theme is a more important selection criterion - no matter how that theme is actually represented. In the early stages of regional exploratory studies, users do not want to generate their own map displays, nor do they want to adjust existing ones or interact with them. The main reason behind the supply-driven nature of map use is the time factor. Map use in regional exploratory studies is a very quick process. Not only do users want to retrieve maps as quickly as possible, they also devote little time to actually consulting the map. They do not carefully consult legends; nor do they notice important aspects of, or relevant mistakes in, the data or the way in which these data are represented. Users are not cartographically aware, and do not know which map types provide the best answers to which geographic questions. Nevertheless, experienced regional geographers use the maps for confirming or rejecting their geographic hypotheses, as well as for executing high-level map use tasks. Novices use maps to come to grips with the task in hand, and mostly stick to elementary map use tasks. Users may postpone their interactive, private and demand-driven dealings with map displays until the later stages of their regional geographic studies. Therefore, and in view of the application potential of highly interactive cartographic visualization software tools, what was called exploratory cartography may better be called analytic cartography, for application in a later stage in regional studies in which the main geographic characteristics have already been discovered, but need to be analysed further.
For executing regional exploratory studies, geographers would first of all benefit from a coherent supply of ready-made map displays in atlas form, carefully prepared by cartographic visualization experts. If such an atlas comes in digital format, the maps should be quickly accessible. For an immaculate transition to the subsequent stage of analytic cartography, the atlas tool could well be extended by interactive and user-friendly (cartographic) visualization functionalities and the possibility of linking to, or adding, geographic data.