Local modelling of land consumption in Germany with RegioClust

Abstract

Germany is experiencing extensive land consumption. This necessitates local models to understand actual and future land consumption patterns. This research examined land consumption rates on a municipality level in Germany for the period 2000–10 and predicted rates for 2010–20. For this purpose, RegioClust, an algorithm that combines hierarchical clustering and regression analysis to identify regions with similar relationships between land consumption and its drivers, was developed. The performance of RegioClust was compared against geographically weighted regression (GWR). Distinct spatially varying relationships across regions emerged, whereas population density is suggested as the central driver. Although both RegioClust and GWR predicted an increase in land consumption rates for east Germany for 2010–20, only RegioClust forecasts a decline for west Germany. In conclusion, both models predict for 2010–20 a rate of land consumption that suggests that the policy objective of reducing land consumption to 30 ha per day in 2020 will not be achieved. Policymakers are advised to take action and revise existing planning strategies to counteract this development.