?
Феноменология пространственного анализа: модели географической редукции и методы пространственной статистики
The author analyzes how existing representations of spatial phenomena in geography relate to the assumptions underlying statistical and geoinformatic methods. In a broad sense, geography problematizes multi-scale manifestations of spatial homogeneity and heterogeneity, linking them to the spatial positions of observations. Models of geographic reduction are different ways of representing space, its elements, the spatial relationships among them, and the resulting spatial effects. These models determine how the positions of observations in space and the mechanisms of formation of spatial phenomena are described. The author considers three basic types of such representations (models)—fields, discrete objects, and networks—describing their key properties, providing examples from different subdisciplines of geography, and outlining differences in the application of spatial analysis and statistical methods such as kriging, moving-window operations, spatial regression, and others. Models of reduction are discussed in greater detail using population geography and mobility as examples. In contrast to network-based gravity models of mobility, which do not take the spatial context of movements into account, alternative interveningopportunities models are based on a topological description of potential movement destinations for agents, and the modeling results can be represented as a probability field of movements. Models of geographic reduction define the framework for formulating research questions and for applying statistical and geoinformatic methods. The same geographic phenomena can be represented simultaneously through different models, which opens up new ways of addressing research problems, while the methods that operate on these models can thus be transferred across different domains of geography.