In this research, we explore the integration of computational and visual approaches, to contribute to the analysis of complex geospatial data. Computational analysis based on the SOM is used in a framework for data mining, knowledge discovery and spatial analysis, for uncovering the structure, patterns, relationships and trends in the data. The framework is informed by current understanding of the effective application of visual variables for cartographic and information design, by developing theories on interface metaphors for geospatial information displays, and by previous empirical studies of map and information visualization effectiveness. It is used to facilitate the knowledge construction process by supporting user's exploratory tasks in a number of ways, including a scenario for better use of the representational spaces. The ultimate goal is to support visual data mining and exploration, and gain insights into underlying distributions, patterns and trends, and thus contribute to enhancing the understanding of geographic processes and support knowledge construction. The framework guided the initial design decisions of a prototype exploratory geovisualization environment. The visualization environment incorporates several graphical representations of SOM output. These include a distance matrix representation, 2D and 3D projections, 2D and 3D surfaces, and component plane visualization with which correlations and relationships can be easily explored. Multiple views are used to simultaneously present interactions between several variables over the space of the SOM, maps, and other graphics such as parallel coordinate plots. Some applications of the method are explored with different datasets. A usability evaluation methodology based on a taxonomy of exploratory tasks and visualization operations is developed to assess the effectiveness of the proposed exploratory geovisualization environment. A subsequent empirical usability testing is conducted and involves different options of map-based and interactive visualizations of a SOM output with the exploration of a socio-demographic dataset. The study emphasizes the visual exploration and knowledge discovery processes. The usability test results and answers to the research questions provide some guidelines for geovisualization design that integrate different representations such as maps, parallel coordinate plots and other information visualization techniques. The research shows that visual exploration can be enhanced by combining the attribute space and the geographic space visualizations. To be effective, this integration of visual tools needs to be done appropriately since these tools are found to support different visual tasks. For visual grouping and clustering, visual analysis and comparison of the patterns in the data, and for revealing relationships, the SOM was found more effective than the map. The usability test results suggest that the integration of map and other representations techniques such as parallel coordinate plot and the SOM-based visualization of the attributes space should reflect the potential of each visual tool. The attribute space visualization is effective as a visual data mining tool allowing the user to select, filter, and output results. The results of this process can be viewed in maps, since the map was generally a better representation for tasks that involve visual attention and sequencing (locate, distinguish, rank).