Understanding spatial interaction in models of commuting behaviour

Abstract

This dissertation aims to understand the spatial and historical complexity of interaction patterns due to commuting and the change of functional regions – commuting regions. It contributes to the ongoing debate about more flexible regionalisation techniques, which allow detecting more complex spatial structures than a closed labour market. Spatial effects between the transport system and land use are investigated.</br></br>The spatial interaction is context-specific. At the municipal level, accessibilities are calculated for Germany and Switzerland. The historical accessibility combines the quality of the transport network and the attractiveness of locations. This interaction pattern describes the historical development in economically powerful regions or peripheries. This changes the perception of distance; stronger interaction lets cities move closer for example. Other historical analyses show the delineation of functional commuting regions with clear and fuzzy boundaries, which is a blurring of the interaction concept.</br></br>After the discussion of traditional methods of defining catchment areas, two techniques, the confidence ellipse and the clustering by social network algorithms, are used to visualize and measure overlaps of commuting regions. This analysis shows commuting regions in rural areas and a growth of commuting regions in the flatland of Switzerland. Two basic spatial structures of overlaps are described: "an area between with strong connection without direction" or "an area more at the edge with weak links." In addition, these structural changes are examined. An exploratory spatial model explains the influence of morphological, economic and demographic factors on size and existence of commuting regions. The commuter flows and the shape of commuting regions are controlled by municipal strategies and the individual decision of each employee. The last chapter changes the perspective from the municipal to personal one: the individual residential choice. It examines spatially different strategies of residential choice. However, the choice of a residence does not differ systematically between functional and spatial categorisations.</br></br>All chapters of this dissertation use spatial characteristics which go beyond the Euclidean distance, and altered perception of space. These spatial interaction characteristics are accessibility, access variables and functional region. The value und significance of these spatial characteristics in the modelling process is reflected. The availability of data, the geographic extent and the historical timeline characterizes this work. The model estimates are often ambiguous, which is related to the problem of the strong correlation of spatial attributes. This dissertation uses a variety of spatial analyses: maps, clustering and economic models to explain commuting regions and complex spatial interaction, in general.