Rethinking land take futures: A cellular automata-based spatial planning approach to model urban expansion and densification under divergent growth scenarios

Land take - the conversion of greenfield land into built-up areas - poses critical challenges for sustainable urban development. Addressing this issue requires understanding the balance between outward urban expansion and inward urban densification. This study employs a Multinomial Logistic Regression-based Cellular Automata (MNL-CA) model to simulate two different future scenarios of urban development till 2050 in Wallonia, Belgium - a region experiencing rapid urbanisation. The model simulates two contrasting built-up demand scenarios. The first scenario is density-based, which follows a linear extrapolation of historical trends in built-up demand into the future. Conceptually, this scenario represents a gradual slowdown of urban expansion, indicating a shift to densification. The second applies a stress-test scenario in which urban expansion continues steadily through 2050. The results reveal that under the density-based scenario referred to as Business-As-Usual (BAU), expansion rates decrease sharply, stabilising at approximately 0 hectares per day by 2040. This indicates a shift toward compact urban forms, albeit accompanied by a continuous decline in overall demand. By contrast, the growth-based scenario or Growth-As-Usual (GAU) produces ongoing expansion at around 2.5 hectares per day in 2050, highlighting the risks of uncontrolled land consumption. Spatial metrics further demonstrate that the density-based scenario fosters compact and contiguous development, whereas the growth-based scenario results in fragmented urban forms that challenge the resilience of spatial planning policies. By evaluating both scenarios, our model provides a framework for stress-testing long-term development strategies, enabling more robust assessments for sustainable land-use policy.