Settlement growth and densification within a peri-urban polycentric region Examination of driving forces, model development and preliminary simulation results

The paper discusses research carried out within the EU funded project GEOLAND. Settlement growth will be modelled for small towns in a polycentric semi-rural region without a defined core city and with competing municipal efforts to attract industries and population on either side of the Rhine - in Austria und Switzerland. The model region is the Austrian Rhine valley. In particular, the model simulates population migration and commercial start ups controlled by regional and local factors (attractiveness/constraints). The approach concentrates on a multi agent system to simulate regional migration and allocation decisions of households and companies causing built-up area densification and land-use change patterns. Thus the models virtual "game board" is a cellular landscape (of 50x50m cells) characterised by several grid cell layers, which comprises information from high resolution earth observation data of different historic dates and cell-related additional demographic and employment data from official statistics, further landscape attractiveness and workplace accessibility. The model takes into account different settlement development velocities due to different migration/commercial start-up attractiveness and considers further the effects of introducing or not introducing green belt protection areas to hinder settlement expansion. To examine, whether competing job offers in Switzerland influence migration decisions, 2 different spatial attractiveness surface layer sets are generated – one for the those people who want to work in Switzerland and one for those who remain working in Austria. The suitability and relevance of the attractiveness criteria is examined via regression models, which explain regional migration patterns through various attractiveness factors and workplace accessibility. The very local decision where to settle is carried out by local attractiveness surface layers considering suitability for housing or commercial sites on a block level. The model is validated by control runs for past decades comparing model results with satellite image land-use classification on a regional and on a local level with different statistical parameters on a municipality - and further on cell basis. Scenario-runs till 2020 are carried out using overall population and employment forecasts for the region as growth framework together with land use patterns, zoning restrictions und future work-place accessibility conditions estimated applying an assumed future road network.